CA2105842C - Neural network based control system - Google Patents

Neural network based control system

Info

Publication number
CA2105842C
CA2105842C CA002105842A CA2105842A CA2105842C CA 2105842 C CA2105842 C CA 2105842C CA 002105842 A CA002105842 A CA 002105842A CA 2105842 A CA2105842 A CA 2105842A CA 2105842 C CA2105842 C CA 2105842C
Authority
CA
Canada
Prior art keywords
neural network
combine
actuator
machine
function
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CA002105842A
Other languages
French (fr)
Other versions
CA2105842A1 (en
Inventor
James William Hall
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Deere and Co
Original Assignee
Deere and Co
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Deere and Co filed Critical Deere and Co
Publication of CA2105842A1 publication Critical patent/CA2105842A1/en
Application granted granted Critical
Publication of CA2105842C publication Critical patent/CA2105842C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/0265Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
    • G05B13/027Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks only
    • AHUMAN NECESSITIES
    • A01AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01DHARVESTING; MOWING
    • A01D41/00Combines, i.e. harvesters or mowers combined with threshing devices
    • A01D41/12Details of combines
    • A01D41/127Control or measuring arrangements specially adapted for combines

Abstract

A neural network is trained with a general set of data to function as a general model of a machine or process with local condition inputs set equal to zero. The network is then retrained or receives additional training on an extentd data set containing the general set of data, characterized by zero values for the local condition inputs, and data on specific local conditions, characterized by non-zero values for the local condition inputs. The result is a trained neural network which functions as a general model when the inputs for the local conditions inputs are set equal to zero, and which functions as a model of some specific local condition when the local condition inputs match the encoding of the some local data set contained within the training data. The neural network has an architecture and a number of neurons such that its functioning as the local model is partially dependent upon its functioning as the general model. This trained neural network is combined with sensors, actuators, a control and communications computer and with a user interface to function as combine control system.

Description

~ 210~8~2 NEURAL NETWORK BASED CONTROL SYSTE~
Background of the Invention This application includes a microfiche appendix including four microfiche and 300 frames.
The invention relates to a control system which utilizes a neural network, and more particularly, to such a control system for a combine harvester.
Combine grain harvesters are complex machines and carry out a complex process that currently requires human supervisory control. The many adjustments built into a combine harvester provide versatility, bùt also make set-up and operation of the machine very difficult, even for experienced operators. For nearly 30 years, researchers have attempted to control sub-systems of the harvesting process with traditional forms of automatic controls. Examples include automatic feedrate controls, threshing cylinder speed controls and cleaning fan controls. More recently, attempts have been made to model or control the harvesting process with rule-based expert systems. However, none of these proposed solutions have proven successful in the marketplace. One reason for this lack of success is that interactions exist between sub-systems. Effective optimization requires consideration of the entire machine or process rather than isolated sub-systems.
Summary of the Invention Accordingly, an object of this invention is to provide a neural network based control sy~tem for a combine harvester.
A further object of the invention is to provide such a control system which can respond to and control the entire machine rather than only portions or sub-systems of the machine.
A further object of the invention is to provide such a control system which can function in response to a general or typical set of conditions and which can function in response to different specific local conditions.
Another object of the invention i8 to provide such a control system which can function in response to different specific local conditions while retaining the ability to function in response the general or typical set of conditions.

21Q~'~4X

These and other objects are achieved by the present invention, which includes a neural network with an architecture including a plurality of neurons, including hidden layers. Each of the neurons in one of the layers has a plurality of inputs. At least one of these inputs i8 a local condition input which is set e~ual to zero while the network is trained with a general set of data. The network is then re-trained or receives additional training on an extended data set containing the general set of data, characterized by zero values for the local condition inputs, and data on specific local conditions, characterized by non-zero values for the local condition inputs. The resulting trained network will function as a general model and as a local model and the local model will function partially dependent upon the general model. This trained neural network is combined with sensors, actuators, a control and communications computer and with a user interface to function as a combine control system.
Brief Description of the Drawings FIG. 1 is a side view of a combine schematically illustrating its major systems and components.
FIG. 2 is schematic block diagram of the control system of the present invention.
FIG. 3 illustrates the graphical user interface implemented by the data acquisition/actuator control (DAAC) computer of FIG. 2.
FIGs. 4a-4c illustrate the graphical user interfaces implemented by the neural network emulation computer of FIG.
2.
FIG. 5a shows the arrangement of FIGs. 5b-5f.
FIGs. 5b-5f are combined to form a top level schematic diagram illustrating neural network emulator of FIG. 2.
FIG. 6 illustrates a neuron which is representative of the neurons which are in the first layer of the neural network of FIGs. 5b-5f.
FIG. 7 illustrates a neuron which is representative of the neurons which are in the second layer of the neural network of FIGs. 5b-5f.

~ 1 a ~ 8 ,~~ 2 FIG. 8 illustrates a neuron which is representative of the neurons which are in the third or output layer of the neural network of FIGs. Sb-5f.
FIG. 9a shows the arrangement of FIGs. 9b-91.
s FIGs. 9b-91 are combined to form a wiring diagram illustrating the senors, relays and actuators of the present invention.

FIGs. lO and ll illustrate in tabular form experimentally determined data for a local condition.

Detailed Description FIG. 1 illustrates an agricultural combine 10, such as a John Deere Model 9S00 production combine harvester, such as lS described in US Patent No. 4,967,863, issued 6 November l990 to Tei~ido et al., and assigned to the assignee of the present invention. The combine 10 has a chassis 12 and ground engaging wheels 14 and 16. Forward ground engaging wheels 14 are driven by hydraulic motor 18 located between the wheels.
Harvesting platform 22 extends forwardly from the chassis of the co~bine and is used for harvesting a crop in a field.
After harvesting, the crop is then directed through feeder house 24 and into the combine. Clean grain compartment 26 is located behind the operator's cab 20 at the top of the 2S combine. Behind compartment 26 is transverse internal combustion engine 28 which is the prime mover of the combine, powering the propulsion means, the harvesting means, and the threshing and separating means. The internal combustion engine 28 is provided with a driving means 30 for powerinq the various usage assemblies.
Between the sides of the combine, which form the chassis of the combine, is located the threshing and separating means.
The threshing and separating means separates the grain from the straw and chaff of the harvested crop. The feeder house directs the harvested grain to threshing means 31 which comprises rotating transverse threshing cylinder 32, transverse concave 34, and rotating beater 38. As the crop - 210~84~

passes between cylinder 32 and concave 34, grain and chaff fall through the concave to pan 36 and the remaining straw and unseparated grain is advanced to beater 38.
After threshing, the straw in the crop is advanced to s separating means 39. The main elements of the separating means are straw walkers 40 and 42, and cleaning shoe assembly 48. From beater 38 the crop-is advanced to the oscillating straw walkers 40 and 42 which direct the straw to the rear of the combine where it is returned to the field by straw spreader ~4. Grain and chaff falling through the straw walkers falls onto oscillating slanted pan 46 which directs the grain and chaff to pan 36. The grain and chaff are directed from pan 36 by overhead auger assemblies to cleaning shoe assembly 48 which is used to separate the chaff from the grain. The grain and chaff falling onto the chaffer 47, sieve 49 and chaffer extension 51 of the cleaning shoe assembly encounters an air stream from fan 50 which blows the lighter chaff out of the rear of the combine while the heavier grain falls through the cleaning shoe assembly and into clean grain receivinq auger 52.
Auger 52 directs the clean grain to a clean grain elevator (not shown) which in turn directs the grain to clean grain compartment 26. Tailings, that is unthreshed heads of grain, fall into tailings auger 54 which directs the 2S unthreshed heads back to the threshing cylinder!and concave.
When the clean grain compartment i8 to be unloaded, transverse unloading augers 56 and 58 direct the grain to the side of the compartment from where it comes into contact with a vertical unloading auger (not shown) which directs the clean grain through unloading tube 58. During an unloading operation, tube 58 would normally be extended outwardly from the side of the combine so that clean grain can be more readily directed into a wagon or truck. ~t should be noted that the arrangement of the threshing and separating elements is well known, and that this invention is directed to a control system for controlling the operation of these elements and the combine.

The cleaner assembly 48 of the combine 10 is equipped with three sieving surfaces. The upper surface, or chaffer 47, is adjusted to retain the large pieces of material other than grain, "MOG~, while allowing the grain and small pieces of chaff to pass through. The sieve 49 is located under the chaffer 47. The sieve 49 is ad~usted to remove the amall pieces of chaff, while allowing the cleaned grain to pass through. At the rear of the cleaner assembly i6 the chaffer-extension 51. Its purpose is to scavenge any remaining grain or unthreshed heads of grain before the trash is discharged from the combine 10. Each of these sieving ~urfaces i6 composed of adjustable louvers (not shown). A combination of louver opening and fan speed is used to completely clean the grain with a minimum of grain loss. In the production model of the combine, the operator must leave the cab, and make these louver adjustments from the ground at the rear of the machine.
Referring now to ~IG. 2, the control system 60 includes a variety of sensors and actuators, a data acquisition system, a data acquisition/actuator control computer (DAAC) 62 to control the actuators, and a second or neural network emulator computer (NN) 64 to run the neural network process model.
The data acquisition/control computer 62 functions as a closed loop controller for the ad~ustment actuators, gathers and processes sensor data, and feeds data to and from the neural network computer 64. The computer 62 is preferably an Apple Macintosh portable computer with 8MB of RAM and a 40 MB
hard disk, is relatively small, includes a flat LCD screen, is battery-powered, and is preferably mounted in the harvester cab. The second computer 64 functions as a neural network processor or emulator, and is preferably a Macintosh portable or laptop computer, also located in the cab of the combine 10.
Computers 62 and 64 are connected to each other via their printer ports.
Computer 62 includes an internal modem (not shown) which is connected by an RS-422 data bus 66 to an RS-485-to-RS-422 converter 68 (preferably model number M1200), which is in turn ~ i ~ ~ 8 1 2 connected to an RS-485 data bus 70. Bus 70 is in turn connected to a pair of frequency-to-RS-485 converters 72 and 74 (preferably model number M1602) and to an RS-232-to-RS-485 converter 76 tmodel 485COR from B ~ B Electronics Mfg Ottawa, IL). Converters 68, 72 and 74 are commerclally available from the Reithly Metrabyte Company, Taunton, MA.
A SCSI port of computer 62 i6 connected to SCSI bus 78.
Bus 78 is connected to a pair of 8-channel MacADIOS digital output devices 80 and to a pair of 8-channel MacADIOS analog input devices 82. Devices 80 are connected to a rack 84 of output signal conditioning circuits for 16 digital output relays. Devices 82 are connected to a rack 86 of input signal conditioning circuits for 16 analog input channels. The SCSI
digital output devices 80 are preferably model GWI-8DIO, and the SCSI analog input devices 82 are preferably model GWI-8AIN, both from GW Instruments, Somerville, MA. The digital output signal condition ~odules 84 are preferably model ODC-05, and the analog input conditioning modules 86 are preferably'~odél MB-31-03, all from Reithly Metrabyte, Taunton, MA. Power i8 supplied by a 12 VDC/110 VAC inverter.
A threshing cylinder speed adjusting actuator 90 is connected to module 84 via a speed increase relay and a speed decrease relay. A cleaning fan speed adjusting actuator 92 i6 connected to module 84 via a fan speed increase relay and a fan speed decrease relay. A concave clearance adjusting actuator 94 is connected to module 84 via a clearance increase relay and a clearance decrease relay. Electro-hydraulic or electric actuators 90, 92 and 94 are already available on the production model John Deere 9500 Combine. In the production combine these actuators are controlled through rocker-switches mounted on an operator~ 8 console. Computer control of these actuators is implemented by paralleling electrical relay modules with the contacts of the rocker sw~tches. Thu~, the actuators 90-94 may be operated manually, or they could be controlled by the computer through a digital output signal to the relays. As a safety precaution, the relays are wired in 210~

such a way that manual inputs take priority over computer inputs.
The production model harvester comes already equipped with digital tachometers 96 and 98 which provide feedback signals representing the speed of threshing cylinder 32 and the speed of the cleaner fan 50. The speed signals coming into these tachometers were frequency pulses from Hall-effect sensors (not shown) located on the rotating shafts (not shown). To transform these raw signals into a form that could -10 be read by the data acquisition/control computer 72, they are fed fir~t into the frequency-to-RS-485 converters 72 and 74 and subsequently through bus 70, converter 68 and bus 66 to the modem port of the computer 62.
The system also includes a grain moisture sensor 100 such as a "MoistureTrac" model 5010 manufactured by Shiwers Inc., Corydon, IA, with an RS-232 communications port installed by the manufacturer, and a grain temperature sensor 101. The grain moi~ture and grain temperature readings pass through converter 76, bus 70, then through converter 68 and bus 66 to the computer 62. Sensors 100 and 101 are preferably located in the clean grain compartment 26 of the combine 10.
An ambient relative humidity sensor 118 (such as a model CT-827-D-X21 sensor from Hy-Cal Corporation, El Monte, CA) provides an analog signal which is applied directly to one of the AnaloglDigital input channel~ of the rack 86.
Time-of-day signals are obtained from the internal clock (not shown) of the data acquisition computer 62.
The control system of the present invention includes a chaffer adju~ting actuator 104, a cleaner extension adjusting actuator 106 and a sieve adjusting actuator 108. Like actuators 90-94, each of actuators 104-108 is connected to module 84 by a correspondinq pair of relays. Preferably, these actuators are ball-screw electric actuators, such as Warner Electric ElecTrak model S24-17A8-02, and are ~ounted on the stationary frame (not shown) of the harvester, ~nd connected to the corresponding production adjustment linkage through flexible cables (not shown). The shaking motion of 210~

the cleaner assembly 48 is absorbed in the flexible cables, thus extending the life of the actuators 104-108.
Because of hysteresis due to wind-up in the cables and tolerance stack-up from the adjustment linkage ~oint~, it is S not possible to accurately control the louver position by monitoring the position of the actuators. To solve this problem,~three accurate rotary potentiometers 110, 112 and 114 are preferably mounted directly on the rear-most louver shafts (not shown) of the louvers (not shown), one each for the chaffer 47, the sieve 49, and the cleaner extension Sl and provide corresponding louver position feedback signals.
Because of the severe vibration and dust, a sealed and contactless sensor is preferred, such as model CP-2UX-R
potentiometers commercially available from ~idori.
An oil-filled linear potentiometer 116, such as model L010-100 from ETI, Oceanside, CA, is used to provide a position feedback signal representing the clearance between the threshing cylinder 32 and the concave 34. Such an oil-filled version is reliable in a very dusty environment.
A feedrate sensor 118 is installed near the top of the clean grain elevator (not shown) of the combine 10 to measure the grain feedrate. This sensor may be an infrared or photo-electric sensor such as described in GB Patent Specification 1,S06,329, published S April 1978 or a weight sensitive flow meter such as described in US Patent No. 4,765,190, issued 23 August 1988. The analog output of the sensor 118 is fed into one of the A/D input channels of module 86. For further information relating to the arrangement of the electrical relay modules, rocker switches, sensors and actuators, reference i~ hereby made to FIGs. 9a-91 and the following part list.

2105~12 PART LIST

PART PART TYPE PART FUNCTIO~

Cl 10MFD CAPACITOR VRl OUTPUT LINE NOISE SUPPRESSO~
C2 .33MFD CAPACITOR VRl AND VR2 INPUT LINE NOISE
SUPPRESSOR
Dl lN4001 DIODE Kl TRANSIENT SUPPRESSION DIODE
D2 lN4001 DIODE K2 TRANSIENT SUPPRESSION DIODE
D3 lN4001 DIODE K3 TRANSIENT SUPPRESSION DIODE
D4 lN4001 DIODE ~4 TRANSIENT SUPPRESSION DIODE
D5 lN4001 DIODE K5 TRANSIENT SUPPRESSION DIODE
D6 lN4001 DIODE R6 TRANSIENT SUPPRESSION DIODE
D7 lN4001 DIODE CHAFFER OPEN COMMAND LINE
ISOLATION DIODE
D8 lN4001 DIODE CHAFFER CLOSE COMMAND LINE
ISOLATION DIODE
D9 lN4001 DIODE EXTENSION OPEN COMMAND LINE
ISOLATION DIODE
D10 lN4001 DIODE EXTENSION CLOSE COMMAND LINE
ISOLATION DIODE
Dll lN4001 DIODE SIEVE OPEN COMMAND LINE ISOLATIO~
DIODE
D12 lN4001 DIODE SIEVE CLOSE COMMAND LINE ISOLATIO~
' DIODE
D13 lN4001 DIODE INCREASE FAN SPEED COMMAND LINE
ISOLATION DIODE

~ l ~ 3 8 ~

D14 lN4001 DIODE DECREASE FAN SPEED COMMAND LINE
ISOLATION DIODE
D15 lN4001 DIODE INCREASE CYLINDER SPEED COMMAND
LINE ISOLATION DIODE
D16 lN4001 DIODE DECREASE CYLINDER SPEED COMMAND
LINE ISOLATION DIODE
D17 lN4001 DIODE INCREASE CONCAVE CLEARANCE COMYAN~
LINE ISOLATION DIODE
D18 lN4001 DIODE DECREASE CONCAVE CLEARANCE COMMAN~
LINE ISOLATION DIODE

PART PART TYPE PART FUNC$ION

Dl9 lN4001 DIODE K8 TRANSIENT SUPPRESSION DIODE

D20 lN4001 DIODE K9 TRANSIENT SUPPRESSION DIODE

Fl 10 AMP FUSE ARMREST EXTENSION PANEL 12 VOLI
PROTECTION FUSE
Jl 37 PIN AMP SERIES 1 CONNECTS ARMREST EXTENSION TO I/( CONNECTOR HARDWARE RACK

CONNECTOR SENSOR ASSEMBLIES

CONNECTOR PRODUCTION ARMREST

Kl P & B VFM SERIES RELAY CHAFFER OPEN RELAY
K2 P & B VFM SERIES RELAY CHAFFER CLOSE RELAY
R3 P & B VFM SERIES RELAY EXTENSION OPEN RELAY

210~8 ~

K4 P & B VFM SERIES RELAY EXTENSION CLOSE RELAY
R5 P & B VFM SERIES RELAY SIEVE OPEN RELAY
K6 P & B VFM SERIES RELAY SIEVE CLOSE RELAY

X8 P & B VFM SERIES RELAY CYLINDER SPEED INCREASE RELAY
K9 P & B VFM SERIES RELAY CYLINDER SPEED DECREASE RELAY
Ml ACCULEX DIGITAL PANEL DISPLAY CHAFFER POSITION
METER

METER

METER
Rl 470 OHM RESISTOR ISOLATION RESISTOR

RVl POTENTIOMETER CHAFFER DPM SPAN

PART PART TYPE PART FUNCTION

SWl JD AH109120 ROCKER CHAFFER OPEN/CLOSE SWITCH
SWITCH

SWITCH

SWITCH

2 1 ~ 2 VRl 78M05 VOLTAGE SUPPLY 5V POWER TO DIGITAL PANEL
REGULATOR METERS

REGULATOR POTENTIOMETERS

PAN~ 2 PAR$ PART $YPE PART YU~CTION

Dl PRODUCTION PART VRl TRANSIENT SUPPRESSOR
D3 PRODUCTION PART Kl TRANSIENT SUPPRESSOR
Xl PRODUCTION PART HEADER CLUTCH INTERRUPT
Jl PRODUCTION PART CONNECTSTO VARIOUS OTHERCOMBINE
COMPONENTS

COMPONENTS

SWl PRODUCTION PART HEADER ENGAGE ON/OFF

SW7 PRODUCTION PART CYLINDER SPEED INCREASE/DECREAS~
(YUNCTION p~P~T.T.RT.P~) SW8 PRODUCTION PART OONCAVE POSITION D~U~SE/DE~SE
(FUNCTION p~T ~.T.RT.P~) ~1~5~12 PART PART TYPE PART FUNCTION

(FUNCTION P~T.T.R~.Rn) TIME DELAY PRODUCTION PARTS SEAT SWITCH/HEADER ENGAGE TIMER
COMPONENTS
VRl PRODUCTION PARTS 5 VOLT REGULATOR

C W FER ACTUATOR A88E~8~Y

PART PART TYPE PART FUNCTION

CHAFFER MODIFIED PRODUCTION
ASSEMBLY PART

SERIES EXCITATION TO CHAFFER, EXTENSION, AND SIEVE ASSEMBLIES

SERIES POSITION MOTOR

PUSH/PULL CONNECTS MOTOR TO CHAFFER ASSEMBLY
CABLE

SENSOR

~ART PART TYPE PART FUNCTION

2:L95~4~

EXTENSION MODIFIED PRODUCTION
ASSEMBLY PART

SERIES EXCITATION TO CHAFFER, EXTENSION, AND SIEVE ASSEMBLIES

SERIES POSITION MOTOR

PART PMT TYPE PART FUNCTION

PUSH/PULL CONNECTS MOTOR TO EXTENSION
CABLE ASSEMBLY

SENSOR

8IBVE ACTUATOR A88EM~LY

PART PART TYPE PART YUNCTION

SIEVE MODIFIED PRODUCTION
ASSEMBLY PART

SERIES EXCITATION TO CHAFFER, EXTENSION
AND SIEVE ASSEMBLIES

5 ~ 1 2 SERIES POSITION MOTOR

PUSH/PULL CONNECTS MOTOR TO SIEVE ASSEMBLY

SENSOR

CONCAVE POBITION ACTUATOR A88~MB~Y

PAaT PART TYPE PART ~UNCTION

POT LINEAR ACTION REPORT CONCAVE POSITION
POTENTIOMETER

as 485 BU8 INTERFACE NODULE8 PAaT PAaT TYPE PAaT ~UNCTION

M1602 (C) KEATHLEY M1602 CONVERTS PRODUCTION CYLINDER SPEED

M1602 (F) KEATH~EY M1602 CONVERTS PRODUCTION FAN SPEED

485COR B & B ELECTRONICS CONv~lS RS 232 SIGNAL FROM

21~ 3 8 ~12 GRAIN ~5018T~ 3AMPLINa 8Y~

PART TYP~ ~ART FlJ~:tOII

MODEL 5010 SHIWERS MOISTURE TRAC SA~LES GRAIN ENTERING GRAIN TANK
AND REPORTS MOISTURE CONTENT

110 VOLT AC VOI,TAGI~ ~O~C~

PAR~ PART TYP~ PART F~NCTlOII

;K

Other parts and circuits shown on FIGs. 9b-91 are production combine parts and are shown for purposes of clarity.

Referring now to FIG. 3, a graphical user interface (GUI) 200 is implemented by the data acquisition/actuator control (DAAC) computer 62. GUI 200 includes switches, digital input and output devices and indicator lights relating to concave position, cylinder speed, fan speed, chaffer opening, extension opening and sieve opening. For each of these combine functions the GUI 200 includes an auto/manual switch 202 so that the operator can select manual control of the associated actuator or automatic, closed loop control of the associated actuator. A readout 204 displays the actual value from the sensor associated with the combine function. Status lights 206 indicate whether the associated combine function is increasing or decreasing. A setpoint select switch 208 can be set to a local position wherein the setpoints are determined manually by the operator via front panel ~ettings and to a neural network position wherein the setpoints are deter~ined by the neural network regardless of the front panel settings.

21~58~2 A setpoint readout 210 displays the value of the setpoints determined by the neural network. A setpoint readout 212 displays the value of the operator determined setpoint associated with the combine function. Also included is a kill s or stop button, a stopped or running status indicator and a set of error indicators.
The computer 62 functions as the GUI 200, and functions with respect to data acquisition, receiving sensor data and with respect to closed loop control of the adjustment actuators. The computer 62 is programmed to perform these functions using commercially available LabView II computer software from National Instruments. Programming in this language is very similar to constructing an analog wiring diagram. The various graphical elements of this language are selected from the LabView menus and incorporated into the GUI
200. With this software it is possible to program the computer 62, complete with GUI 200, without an extensive amount of custom programming.
For more detailed information concerning the program which causes the computer 62 to perform as desired, reference is made to the appropriate portions of the graphical computer program listing included in the microfiche appendix. This graphical computer program listing i8 in the LabView II
graphical language.
Referring now to FIGs. 4a-4c, a graphical user interface (GUI) i8 implemented by the computer 64, also using the commercially available LabView II computer software. The GUI
of FIGs. 4a-4c allows the operator to input data to the neural network emulation and to view the effects of changes of the process adjustments on the process outputs. The GUI 300 of FIG. 4a shows the screen used by the operator to enter the operator inputs on the local crop conditions, and to display sensor inputs. The threshing condition variable allows the operator to set how hard or easy threshing conditions are.
The growing condition variable allows the operator to set how wet or dry are the growinq or field conditions are. The GUI
310 of FIG. 4b allows the operator to define the relative 21058~2 weighting to be applied to the output parameters during the global optimizat~on. The GUI 320 of FIG. 4c ~llows the operator to vary the values of the process adjustments and view the effect on each of the individual output parameters, s as well as the "weighted sum~ global optimization parameter.
Referring now to FIGs. 5a-5f, a neural network 400 is emulated and implemented on the computer 64, also using the commercially available LabView II computer software. This emulated neural network is used by the operator of the combine harvester 10 to optimize the operation of the combine 10.
Neural network 400 include~ inputs for 15 input variables which are scaled to values between -1 and +1, then fed into each of five separate neural networks 402-410, one network for each output parameter. The output paraméters are grain damage, walker loss, cleaner loss, dockage and unthreshed crop. These outputs are displayed on GUI 320.
Network 402 includes a first layer with six neurons dmgl7-dmg22, a second layer with three neurons dmg23-dmg25 and a third or output layer with a single output neuron dmg26.
Network 404 includes a first layer with six neurons sepl7-sep22, a second layer with three neurons sep23-sep25 and a third or output layer with a single output neuron sep26.
Network 406 includes a first layer with six neurons lossl7-loss22, a second layer with three neuron6 loss23-loss25 and a third or output layer with a single output neuron loss26.
Network 408 includes a first layer with 8iX neurons dockl7-dock22, a second layer with three neurons dock23-dock25 and a third or output layer with a single output neuron dock26.
Network 410 includes a first layer with six neurons thrl7-thr22, a second layer with three neurons thr23-thr25 and a third or output layer with a single output neuron thr26.
FIG. 6 is representative of the first layer neurons.
Each fir~t layer neuron includes 15 inputs EXT, a ~et of weights W associated with the various inputs and arithmetic means for determining an output signal as a function of the inputs and the weights.

2 ~ 8 ~ 2 FIG. 7 is representative of the second layer neurons.
Each second layer neuron includes 6 inputs EXT, a set of weights W associated with the various inputs and arithmetic means for determining an output signal as a function of the s inputs and the weights.
FIG. 8 is representative of the third or output layer neurons. Each third or output layer neuron includes 3 inputs EXT, a set of weights W associated with the various inputs and arithmetic means for determining an output signal as a function of the inputs and the weights.
Each individual neuron within the network is similar to the classic representation of a neuron, with the inputs being multiplied by a weight, summed together, and the sum passed through a squashing function, in this case a hyperbolic tangent. For complete details relating to the neural network 400, and all the neurons and weights therein reference is made to the appropriate portions of the graphical computer program listing included in the microfiche appendix.
The neural network 400 is the result of a multiple part training process described below and it functions as an accurate model of the operation of the combine 10. This model can be exercised to determine the optimum setting of the combine controls. To save time during the training process, the network may be trained off-line on a small workstation computer (not shown), then the connection weights of the trained network were transferred to the emulation program running on the laptop computer 64.
A neural network with the architecture of neural network 400 is first trained to function as a general model of the operation of the combine 10. The training data consists of many examples specifying the process inputs and the corresponding outputs. These examples are selected a6 representative of normal combine operation, and are characterized by setting the inputs for local condition A and local condition B equal to zero. The network i8 then trained with this general training data u6ing a conventional back-propagation training algorithm.

~105~,4~

Next the training data is supplemented with an additional set of examples representing some specific local condition.
To distinguish the local data from the general data, examples from a given local condition are given a unique coding using s the local condition inputs. For example, all data from one specific local condition might have the coding local condition A = 1 and local condition B = 0. The neural network i8 then retrained on this combined data set using the same back-propagation training algorithm. The network can be trained to represent multiple local conditions by adding additional local data sets to the set of training examples, each with a unique coding.
The result i8 that the trained neural network functions as a general combine model when the inputs for the local conditions inputs are both set equal to zero, and functions as a model of some specific local condition when the local condition inputs match the encoding of some local data set contained within the training data.
By limiting the number of connections within the neural network, it is possible to control the degree of independence between the local models and the general model. With fewer connection~, the models tend to have a degree of dependence.
In cases where there i8 a large amount of data available for the general model, and only limited data for a local model, this property provides a method for knowledge inheritance from the general model to the local model. The limited degrees of freedom within the neural network are used to ~odel the relationships in which the data indicates explicit differences between the general and the local model. For areas of the local model in which no explicit relationship is given by the data, the limited degrees of freedom force the local model to inherit the relationships contained in the general model.
Stated another way, the neural network disclosed herein is trained with a general set of data to function as a general model of a machine or proce~s and i6 trained with a cosbined set of data including the general set of data and a local set of data to function as a local model or submodel of the 2~5~2 machine or process. The neural network has an architecture in which the degrees of freedom available for forming models are constrained by the limited number of nodes in the hidden layers. The result is that this neural network i8 able to accurately function as a general model of the ~achine or process, but it has insufficient degrees of freedom for the submodels to function completely independent of the general model. Instead, the functioning of the submodels is partially dependent upon the functioning as the general model, and the submodels will deviate from the general model with respect to local training data which differs from the general training data, but will have properties similar to the general model where the local training data does not contradict the general training data.
The general model may be trained from data taken from a variety of sources, including test data on combines operating under typical conditions and numerical representations of expert heuristics. The submodel training is done by operating the combine under specific local conditions and making several local measurements of the combine in operation, noting both the input and output parameters. Typically four to eight measurements for each distinct local condition are sufficient.
Replicate sets of these local examples are added to the set of general training cases BO that the number of new examples represents at least 10-20% of the total training cases.
As described earlier, the local condition inputs are used as indicators to inform the network whether a given data set represents a general condition or a specific local condition.
This local condition input is set to zero for general model training and operation, and is set to some non-zero value for a given local condition. Several different local conditions might be keyed to various values of this local condition input. When the network, that had been previously trained with the general data set, is given further training on the enhanced or combined data set, it forms numerically accurate submodel6 of the local harvesting processes represented by the local data. These local process models incorporate the - ~ 21~812 specific relationships and numerical values that were contained in the local data, and inherit characteristics from the general model which were not explicitly contained in the local data. This ability allows the network to mimic human-li~e decision making and generalization. This capability also produces a process model which can be easily and locally adapted to satisfy specific local needs.
With such a numerically accurate submodel of the harvesting process, it is possible to quickly and easily predict the effect of various combine adju~tments on combine performance. This is far superior to the current practice of making a change in adjustment, stopping the machine and dismounting, observing the effect, making an additional change, and so on. When implemented on a portable computer, this process model could provide the basis for manually optimizing the harvesting process.
For example, an operator can manually and systematically adjust the six operator controls via GUI 200 of FIG. 3 to optimize the outputs (minimize the weighted sum on GUI 320 of FIG. 4c) of the combine as determined by the relative importance or weight of the combine outputs set via GUI 310 of FIG.4b. This can be done utilizing a repetitive, ~one-variable-at-a-time~ technique, or by utilizing a well known optimization technigue such as gradient descent optimization.
a numerically accurate, neural network based process model can also form the basis of an automatic combine ad~ustment controller. Several techniques are well known in the art for integrating a model into a process optimization routine. One example would be to perform a gradient search optimization using the model to predict the effect of each change in adjustments. Such optimization techniques could be easily automated by the addition of computer software optimization algorithm.
The accuracy of the overall optimization/control strategy could be enhanced by the addition of feedback sensor~ for some of the process outputs. Thi8 technique would le~~en the level of accuracy required by the local model, and would perhaps 21û~8~2 permit the process optimization to be performed from the general model itself without local calibration.
In the case of a combine harvester, there are 8iX prOCeS8 outputs to be optimized, but sensors are available for only one of these outputs. If accurate sensors were available for all outputs, optimization of the harvester ad~ustments could be done with traditional control techniques. Since feedback data on the outputs are not available, knowledge processing technique~ are being used to form an accurate process model.
An accurate model can then provide accurate estimates of the process outputs, which are used for open-loop optimization of the process.
In addition to its applicability for combine harvester process control, the general techniques outlined above have broad application to a variety of process control and optimization problems.
The above also provides:
a technique for controlling and optimizing an entire process simultaneously, rather than attempting to control isolated sub-systems;
a technique to fuse knowledge from a variety of sources into a single model of a general process;
a simple technique for transforming a general model into numerically accurate submodels of a process under specific local conditions. This local model can then be used for process optimization.
The generalization process used in this technique forms models which are similar to those obtained by skilled human operators operating under similar conditions. This provides the possibility of replacing skilled human operators with computer-based controllers for complex process control.
The system is capable of continuous self-improvement. As information becomes available it can be easily incorporated into the model. Over time, the specific knowledqe available from human experts, either about the total process or about process subsystems, can be accumulated and integrated into the system.

A portion of the disclosure of this patent document contains material which is subject to a claim of copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all other rights whatsoever.

.

Claims (8)

1. A machine control system for controlling a machine which operates in a variety of locations and conditions and which produces an end result, the control system comprising:
a plurality of actuators, each for controlling a particular function of the machine in response to an actuator control signal;
a plurality of actuator sensors, each generating an actuator condition signal representing a condition of a corresponding one of the actuators;
a plurality of input condition sensors, each generating an input condition sensor signal representing an input condition which influences operation of the machine;
an actuator control unit for generating the actuator control signals as a function of the actuator condition signals and as a function of setpoint signals;
a neural network trained prior to and apart from normal production use of the machine with a set of general training data to function as a general model of the machine and trained to function as a submodel with respect to a set of local condition data together with the set of general training data, the neural network processing the input condition sensor signals and data collected prior to normal production use of the machine representing desired machine performance quality to produce a set of machine adjustments intended to produce the desired machine performance quality, the neural network generating the setpoint signals based upon predicted responses of the machine to varying conditions;
a data communication system comprising means for communicating the actuator signals to the actuator control unit, means for communicating the sensor signals to the neural network, and means for communicating the setpoint signals to the actuator control unit, the neural network and the actuator control unit cooperating to control operation of the machine without measuring the machine performance quality in connection with normal production use of the machine; and operator controlled means for selectively causing the neural network to function as the general model or as the submodel.
2. A control system for a combine for operating in a variety of locations and conditions and which harvests a crop, the combine having a variable speed threshing cylinder, a concave having a variable clearance, a variable speed cleaning fan, a sieve, a chaffer, and a cleaner extension, the control system comprising:

a set of actuators, each for controlling a particular function of the combine in response to an actuator control signal;
a set of actuator sensors, each generating an actuator condition signal representing a condition of a corresponding one of the actuators;
a set of input condition sensors, each generating an input condition sensor signal representing an input condition which influences operation of the combine;
an actuator control unit for generating the actuator control signals as a function of the actuator condition signals and as a function of setpoint signals;
a neural network trained prior to and apart from normal production use of the combine with a set of general training data to function as a general model of the combine and trained to function as a submodel with respect to a set of local condition data together with the set of general training data, the neural network processing the input condition sensor signals and data collected prior to normal production use of the combine representing desired crop harvesting performance quality to produce a set of combine adjustments intended to produce the desired crop harvesting performance quality, the neural network generating the setpoint signals based upon predicted responses of the combine to varying conditions;
a data communication system comprising means for communicating the feedback signals to the actuator control unit, means for communicating the sensor signals to the neural network, and means for communicating the setpoint signals to the actuator control unit, the neural network and the actuator control unit cooperating to control operation of the combine without measuring the crop harvesting performance quality produced by the combine; and operator controlled means for selectively causing the neural network to function as the general model or as the submodel.
3. A control system for a combine for operating in a variety of locations and conditions and which harvests a crop, the combine having a plurality of controllable components, the components comprising a variable speed threshing cylinder, a concave having a variable clearance, a variable speed cleaning fan, a sieve, a chaffer and a cleaner extension, the control system comprising:
a set of actuators, each for controlling a corresponding one of the components in response to a corresponding actuator control signal;

a set of actuator sensors, each generating an actuator condition signal representing a condition of a corresponding one of the actuators;
a set of input condition sensors, each generating an input condition sensor signal representing an input condition which influences operation of the combine;
an actuator control unit for generating the actuator control signals as a function of the actuator condition signals and as a function of setpoint signals;
a neural network trained prior to and apart from normal production use of the combine with a set of general training data to function as a general model of the combine and trained to function as a submodel with respect to a set of local condition data together with the set of general training data, the neural network processing the input condition sensor signals and data collected prior to normal production use of the machine representing desired crop harvesting performance quality to produce a set of combine adjustments intended to produce the desired crop harvesting performance quality, the neural network generating the setpoint signals as a function of predicted responses of the combine to varying conditions;
a data communication system comprising means for communicating the actuator signals to the actuator control unit, means for communicating the sensor signals to the neural network, and means for communicating the setpoint signals to the actuator control unit, the neural network and the actuator control unit cooperating to control operation of the combine without measuring the crop harvesting performance quality produced by the combine; and operator controlled means for selectively causing the neural network to function as the general model or as the submodel.
4. The combine control system of claim 3, wherein the set of condition sensors comprises:
a relative humidity sensor;
a grain moisture sensor; and a grain temperature sensor.
5. The combine control system of claim 3, wherein the set of combine parameter sensors comprises:
a grain feedrate sensor;
a threshing cylinder speed sensor;
a concave clearance sensor;

a cleaning fan speed sensor;
a sieve position sensor;
a chaffer position sensor; and a cleaner extension position sensor.
6. A machine control system for controlling a machine which operates in a variety of locations and conditions and which produces an end result, the control system comprising:
a plurality of actuators, each for controlling a particular function of the machine in response to an actuator control signal;
a plurality of actuator sensors, each generating an actuator condition signal representing a condition of a corresponding one of the actuators;
a plurality of input condition sensors, each generating an input condition sensor signal representing an input condition which influences operation of the machine;
an actuator control unit for generating the actuator control signals as a function of the actuator condition signals and as a function of setpoint signals;
a neural network trained prior to and apart from normal production use of the machine with a set of general training data to function as a general model of the machine and trained to function as a submodel with respect to a set of local condition data together with the set of general training data, the training data being obtained prior to normal production use of the machine and from sources other than the specific machine being controlled and not directly or indirectly from the specific machine being controlled, the neural network generating the setpoint signals based upon predicted responses of the machine to varying conditions;
a data communication system comprising means for communicating the actuator signals to the actuator control unit, means for communicating the sensor signals to the neural network, and means for communicating the setpoint signals to the actuator control unit, the neural network and the actuator control unit cooperating to control operation of the machine; and operator controlled means for selectively causing the neural network to function as the general model or as the submodel.
7. A machine control system for controlling a machine which operates under a variety of conditions and which produces an end result, the control system comprising:

a plurality of actuators, each for controlling a particular function of the machine in response to an actuator control signal;
a plurality of actuator sensors, each generating an actuator condition signal representing a condition of a corresponding one of the actuators;
a plurality of input condition sensors, each generating an input condition sensor signal representing an input condition which influences operation of the machine;
an actuator control unit for generating the actuator control signals as a function of the actuator condition signals and as a function of setpoint signals;
a neural network trained prior to and apart from normal production use of the machine with a set of general training data to function as a general model of the machine and trained to function as a submodel with respect to a set of local condition data together with the set of general training data, the neural network processing the input condition sensor signals and data collected prior to normal production use of the machine representing desired machine performance quality to produce a set of machine adjustments intended to produce the desired machine performance quality, the neural network generating the setpoint signals based upon predicted responses of the machine to varying conditions;
a data communication system comprising means for communicating the actuator signals to the actuator control unit, means for communicating the sensor signals to the neural network, and means for communicating the setpoint signals to the actuator control unit, the neural network and the actuator control unit cooperating to control operation of the machine; and operator controlled means for selectively causing the neural network to function as the general model or as the submodel.
8. A control system for a combine for operating in a variety of locations and conditions and which harvests a crop, the combine having a variable speed threshing cylinder, a concave having a variable clearance, a variable speed cleaning fan, a sieve, a chaffer, and a cleaner extension, the control system comprising:
a set of actuators, each for controlling a particular function of the combine in response to an actuator control signal;

a set of actuator sensors, each generating an actuator condition signal representing a condition of a corresponding one of the actuators;
a set of input condition sensors, each generating an input condition sensor signal representing an input condition which influences operation of the combine;
an actuator control unit for generating the actuator control signals as a function of the actuator condition signals and as a function of setpoint signals;
a neural network trained prior to and apart from normal production use of the combine with a set of general training data to function as a general model of the combine and trained to function as a submodel with respect to a set of local condition data together with the set of general training data, the neural network processing the input condition sensor signals and data collected prior to normal production use of the combine representing desired crop harvesting performance quality to produce a set of combine adjustments intended to produce the desired crop harvesting performance quality, the neural network generating the setpoint signals based upon predicted responses of the combine to varying conditions;
a data communication system comprising means for communicating the feedback signals to the actuator control unit, means for communicating the sensor signals to the neural network, and means for communicating the setpoint signals to the actuator control unit, the neural network and the actuator control unit cooperating to control operation of the combine; and operator controlled means for selectively causing the neural network to function as the general model or as the submodel.
CA002105842A 1992-09-10 1993-09-09 Neural network based control system Expired - Fee Related CA2105842C (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US94529392A 1992-09-10 1992-09-10
US945,293 1992-09-10

Publications (2)

Publication Number Publication Date
CA2105842A1 CA2105842A1 (en) 1994-03-11
CA2105842C true CA2105842C (en) 1998-11-03

Family

ID=25482924

Family Applications (1)

Application Number Title Priority Date Filing Date
CA002105842A Expired - Fee Related CA2105842C (en) 1992-09-10 1993-09-09 Neural network based control system

Country Status (6)

Country Link
US (1) US5586033A (en)
EP (1) EP0586999B1 (en)
JP (1) JP3478855B2 (en)
AU (1) AU658066B2 (en)
CA (1) CA2105842C (en)
DE (1) DE69323943T2 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111751133A (en) * 2020-06-08 2020-10-09 南京航空航天大学 Intelligent fault diagnosis method of deep convolutional neural network model based on non-local mean embedding

Families Citing this family (123)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5212765A (en) * 1990-08-03 1993-05-18 E. I. Du Pont De Nemours & Co., Inc. On-line training neural network system for process control
EP0775612B1 (en) * 1993-06-28 2003-09-24 CNH Belgium N.V. Process for the control of self-propelled agricultural harvesting machines
DE19514223B4 (en) * 1995-04-15 2005-06-23 Claas Kgaa Mbh Method for optimizing the use of agricultural machinery
US5923004A (en) * 1997-12-30 1999-07-13 Otis Elevator Company Method for continuous learning by a neural network used in an elevator dispatching system
DE19800238C1 (en) * 1998-01-07 1999-08-26 Claas Selbstfahr Erntemasch System for setting a self-propelled harvester
DE19807145C2 (en) 1998-02-20 1999-12-09 Claas Selbstfahr Erntemasch Combine with automatic cleaning adjustment device
US6389884B1 (en) * 1998-10-08 2002-05-21 Claas Selbstfahrende Erntemaschinen Gmbh Device and method for measuring the moisture of crop material in agricultural machines
US7110526B1 (en) 1998-10-14 2006-09-19 Rockwell Electronic Commerce Technologies, Llc Neural network for controlling calls in a telephone switch
US6453241B1 (en) * 1998-12-23 2002-09-17 Rosetta Inpharmatics, Inc. Method and system for analyzing biological response signal data
US6213870B1 (en) 1999-03-24 2001-04-10 Caterpillar Inc. Stall prevention system
DE19931844A1 (en) * 1999-07-09 2001-01-11 Claas Selbstfahr Erntemasch Device for adjusting the sieve opening width on combine harvesters
CA2298195A1 (en) 1999-07-16 2001-01-16 Deere & Company Cleaning shoe adjustment mechanism for a harvester
US6876991B1 (en) 1999-11-08 2005-04-05 Collaborative Decision Platforms, Llc. System, method and computer program product for a collaborative decision platform
DE10030262A1 (en) 2000-06-20 2002-01-31 Deere & Co cleaning device
DE10036971A1 (en) * 2000-07-28 2002-02-28 Siemens Ag Method for remote diagnosis of a technological process
US6591145B1 (en) * 2000-09-21 2003-07-08 Bechtel Bwxt Idaho, Llc Systems and methods for autonomously controlling agricultural machinery
DE10064356A1 (en) * 2000-12-21 2002-07-11 Claas Selbstfahr Erntemasch Device and method for conveying crops in agricultural machines
DE10111531A1 (en) * 2001-03-10 2002-09-26 Deere & Co Screen for a combine harvester
US6632136B2 (en) 2001-06-05 2003-10-14 Deere & Company Remote adjustment mechanism for a combine harvester cleaning element
US6553300B2 (en) * 2001-07-16 2003-04-22 Deere & Company Harvester with intelligent hybrid control system
US20030028353A1 (en) * 2001-08-06 2003-02-06 Brian Gventer Production pattern-recognition artificial neural net (ANN) with event-response expert system (ES)--yieldshieldTM
DE10147733A1 (en) * 2001-09-27 2003-04-10 Claas Selbstfahr Erntemasch Method and device for determining a harvester setting
US6655351B2 (en) 2001-10-24 2003-12-02 Deere & Company Vehicle engine control
US7535454B2 (en) * 2001-11-01 2009-05-19 Immersion Corporation Method and apparatus for providing haptic feedback
US20030088321A1 (en) * 2001-11-05 2003-05-08 Creger Todd D Method for compensating for variations in modeled parameters of machines
DE10162357A1 (en) * 2001-12-18 2003-07-17 Claas Selbstfahr Erntemasch Method and device for optimizing the operation of an agricultural vehicle
US6726559B2 (en) 2002-05-14 2004-04-27 Deere & Company Harvester with control system considering operator feedback
GB2428284A (en) * 2002-11-01 2007-01-24 Immersion Corp Method and apparatus for providing a switch with a processor controlled force
US8463441B2 (en) 2002-12-09 2013-06-11 Hudson Technologies, Inc. Method and apparatus for optimizing refrigeration systems
DE10306726A1 (en) * 2003-02-17 2004-09-30 Claas Selbstfahrende Erntemaschinen Gmbh Method for optimizing adjustable parameters
DE10324045B3 (en) * 2003-05-27 2004-10-14 Siemens Ag System characteristics modelling method for dynamic system using similarity analysis for modification of known system characteristics supplied to neural network structure for causality analysis
US7194320B2 (en) * 2003-06-05 2007-03-20 Neuco, Inc. Method for implementing indirect controller
DE10336771A1 (en) 2003-08-08 2005-03-10 Claas Selbstfahr Erntemasch Method and device for operating machines
US7519604B2 (en) * 2003-08-29 2009-04-14 Nokia Corporation Troubleshooting engine and method for using same
KR100543707B1 (en) * 2003-12-04 2006-01-20 삼성전자주식회사 Face recognition method and apparatus using PCA learning per subgroup
US7272544B2 (en) * 2004-01-15 2007-09-18 Honeywell International Inc. Integrated modeling through symbolic manipulation
DE102004011982A1 (en) * 2004-03-10 2005-10-06 Claas Selbstfahrende Erntemaschinen Gmbh Dew point sensor for harvesters
US8504707B2 (en) * 2004-12-07 2013-08-06 Cisco Technology, Inc. Method and system for sending and receiving USB messages over a data network
US7644211B2 (en) * 2004-12-07 2010-01-05 Cisco Technology, Inc. Method and system for controlling transmission of USB messages over a data network between a USB device and a plurality of host computers
DE102005026159A1 (en) * 2005-06-06 2007-01-25 Claas Selbstfahrende Erntemaschinen Gmbh Method for controlling a harvesting machine
US7632179B2 (en) 2005-08-01 2009-12-15 Cnh America Llc System and method for detecting combine rotor slugging
DE102005047335A1 (en) * 2005-09-30 2007-04-12 Claas Selbstfahrende Erntemaschinen Gmbh Self-propelled harvester and operating method for it
US20070260379A1 (en) * 2006-05-05 2007-11-08 Congdon Thomas M Method of operating a machine for conditioning a work material
US8930089B2 (en) * 2006-05-25 2015-01-06 Barry H. Reeves System and method for indicating to a remote location operation of an auger or the like
US20070288147A1 (en) * 2006-05-25 2007-12-13 Reeves Barry H System and method for indicating to a remote location operation of an auger or the like
US7386380B2 (en) * 2006-07-31 2008-06-10 Deere & Company Cotton harvester row unit speed synchronization control
US7452267B2 (en) 2007-03-20 2008-11-18 Cnh America Llc System and method for automatically deslugging an agricultural combine
US7689309B2 (en) * 2007-07-03 2010-03-30 Delphi Technologies, Inc. Method for generating an electro-hydraulic model for control design
DE102007046678A1 (en) * 2007-09-27 2009-04-09 Claas Selbstfahrende Erntemaschinen Gmbh Agricultural work vehicle
DE102008027906A1 (en) * 2008-06-12 2009-12-17 Claas Selbstfahrende Erntemaschinen Gmbh Agricultural harvester
US8627306B2 (en) * 2008-08-06 2014-01-07 Caterpillar Inc. Method and system for updating an information management system configuration
FR2958911B1 (en) * 2010-04-19 2012-04-27 Snecma METHOD AND SYSTEM FOR MONITORING THE OIL LEVEL CONTAINED IN A RESERVOIR OF AN AIRCRAFT ENGINE
DE102010017687A1 (en) * 2010-07-01 2012-01-05 Claas Selbstfahrende Erntemaschinen Gmbh Method for adjusting at least one working member of a self-propelled harvester
DE102010017676A1 (en) 2010-07-01 2012-01-05 Claas Selbstfahrende Erntemaschinen Gmbh Driver assistance system for agricultural machine
US10321624B2 (en) 2011-03-11 2019-06-18 Intelligent Agriculture Solutions LLC Air seeder manifold system
US9631964B2 (en) 2011-03-11 2017-04-25 Intelligent Agricultural Solutions, Llc Acoustic material flow sensor
US10318138B2 (en) * 2011-03-11 2019-06-11 Intelligent Agricultural Solutions Llc Harvesting machine capable of automatic adjustment
US9629308B2 (en) 2011-03-11 2017-04-25 Intelligent Agricultural Solutions, Llc Harvesting machine capable of automatic adjustment
US9582178B2 (en) 2011-11-07 2017-02-28 Immersion Corporation Systems and methods for multi-pressure interaction on touch-sensitive surfaces
CN102998973B (en) * 2012-11-28 2016-11-09 上海交通大学 The multi-model Adaptive Control device of a kind of nonlinear system and control method
US9198344B2 (en) 2013-01-09 2015-12-01 Cnh Industrial Canada, Ltd. Setup wizard for agricultural equipment
CN103310272B (en) * 2013-07-02 2016-06-08 南京邮电大学 Based on the DIVA neural network model manner of articulation that sound channel action knowledge base is improved
DE102014216593A1 (en) 2014-08-21 2016-02-25 Deere & Company Operator assistance system for an agricultural working machine
US10085379B2 (en) 2014-09-12 2018-10-02 Appareo Systems, Llc Grain quality sensor
US9756785B2 (en) 2014-09-12 2017-09-12 Appareo Systems, Llc Grain quality sensor
US9706709B2 (en) 2015-09-10 2017-07-18 Deere & Company Harvester fan speed control based on yield
DE102015121210A1 (en) 2015-12-07 2017-06-08 Claas Selbstfahrende Erntemaschinen Gmbh Agricultural working machine
DE102016207509A1 (en) 2016-05-02 2017-11-02 Deere & Company Method and arrangement for optimizing working parameters of a harvester
US9959498B1 (en) 2016-10-27 2018-05-01 Google Llc Neural network instruction set architecture
US10360163B2 (en) 2016-10-27 2019-07-23 Google Llc Exploiting input data sparsity in neural network compute units
US10175980B2 (en) 2016-10-27 2019-01-08 Google Llc Neural network compute tile
CN110740635A (en) * 2017-03-21 2020-01-31 蓝河技术有限公司 Combine harvester including machine feedback control
JP6673293B2 (en) * 2017-05-24 2020-03-25 トヨタ自動車株式会社 Vehicle system
IT201700058505A1 (en) * 2017-05-30 2018-11-30 Volta Robots S R L Method of control of a soil processing vehicle based on image processing and related system
US11191215B1 (en) * 2017-12-28 2021-12-07 Brian G. Robertson Dynamically operated concave threshing bar
RU2693680C9 (en) * 2018-03-27 2019-07-16 Александр Игоревич Галькевич An intelligent, integrated digital platform for managing large systems, and a way to manage large systems
US11510404B2 (en) 2018-05-24 2022-11-29 Blue River Technology Inc. Boom sprayer including machine feedback control
CN108934416A (en) * 2018-07-06 2018-12-07 巢湖学院 One kind being based on BP neural network combined harvester multi-parameter operation-control system and method
US11589509B2 (en) 2018-10-26 2023-02-28 Deere & Company Predictive machine characteristic map generation and control system
US11957072B2 (en) 2020-02-06 2024-04-16 Deere & Company Pre-emergence weed detection and mitigation system
US11653588B2 (en) 2018-10-26 2023-05-23 Deere & Company Yield map generation and control system
US11079725B2 (en) 2019-04-10 2021-08-03 Deere & Company Machine control using real-time model
US11467605B2 (en) 2019-04-10 2022-10-11 Deere & Company Zonal machine control
US11672203B2 (en) 2018-10-26 2023-06-13 Deere & Company Predictive map generation and control
US11641800B2 (en) 2020-02-06 2023-05-09 Deere & Company Agricultural harvesting machine with pre-emergence weed detection and mitigation system
US11178818B2 (en) 2018-10-26 2021-11-23 Deere & Company Harvesting machine control system with fill level processing based on yield data
US11240961B2 (en) 2018-10-26 2022-02-08 Deere & Company Controlling a harvesting machine based on a geo-spatial representation indicating where the harvesting machine is likely to reach capacity
CN109657782B (en) * 2018-12-14 2020-10-27 安徽寒武纪信息科技有限公司 Operation method, device and related product
US11778945B2 (en) 2019-04-10 2023-10-10 Deere & Company Machine control using real-time model
US11234366B2 (en) 2019-04-10 2022-02-01 Deere & Company Image selection for machine control
US20200375104A1 (en) * 2019-05-31 2020-12-03 Deere & Company Virtual sensor for grain loss in combine harvester
US11690320B2 (en) 2019-07-11 2023-07-04 Deere & Company Work machine control based on machine capabilities relative to work assignment criteria
US20210015045A1 (en) * 2019-07-19 2021-01-21 Deere And Company Federated harvester control
US11864494B2 (en) * 2019-10-29 2024-01-09 Landing AI AI-optimized harvester configured to maximize yield and minimize impurities
EP3839443B1 (en) * 2019-12-16 2022-06-08 Sick Ag Encoder device and method for determining a kinematic value
US11308738B2 (en) 2020-01-06 2022-04-19 Deere & Company Mobile work machine performance detection and control system
US11477940B2 (en) 2020-03-26 2022-10-25 Deere & Company Mobile work machine control based on zone parameter modification
DE102020117941A1 (en) 2020-07-07 2022-01-13 Grimme Landmaschinenfabrik Gmbh & Co. Kg Method for adapting a process model for operating one of several working machines, in particular root crop harvesting machines
DE102020117940A1 (en) 2020-07-07 2022-01-13 Grimme Landmaschinenfabrik Gmbh & Co. Kg Method for operating one of several working machines, in particular root crop harvesting machines
CN112013203B (en) * 2020-07-18 2021-09-24 淮阴工学院 Pipe network detection system based on DRNN neural network
US11946747B2 (en) 2020-10-09 2024-04-02 Deere & Company Crop constituent map generation and control system
US11844311B2 (en) 2020-10-09 2023-12-19 Deere & Company Machine control using a predictive map
US11845449B2 (en) 2020-10-09 2023-12-19 Deere & Company Map generation and control system
US11592822B2 (en) 2020-10-09 2023-02-28 Deere & Company Machine control using a predictive map
US11874669B2 (en) 2020-10-09 2024-01-16 Deere & Company Map generation and control system
US11711995B2 (en) 2020-10-09 2023-08-01 Deere & Company Machine control using a predictive map
US11650587B2 (en) 2020-10-09 2023-05-16 Deere & Company Predictive power map generation and control system
US11675354B2 (en) 2020-10-09 2023-06-13 Deere & Company Machine control using a predictive map
US11864483B2 (en) 2020-10-09 2024-01-09 Deere & Company Predictive map generation and control system
US11895948B2 (en) 2020-10-09 2024-02-13 Deere & Company Predictive map generation and control based on soil properties
US11635765B2 (en) 2020-10-09 2023-04-25 Deere & Company Crop state map generation and control system
US11849671B2 (en) 2020-10-09 2023-12-26 Deere & Company Crop state map generation and control system
US11889788B2 (en) 2020-10-09 2024-02-06 Deere & Company Predictive biomass map generation and control
US11825768B2 (en) 2020-10-09 2023-11-28 Deere & Company Machine control using a predictive map
US11727680B2 (en) 2020-10-09 2023-08-15 Deere & Company Predictive map generation based on seeding characteristics and control
US11474523B2 (en) 2020-10-09 2022-10-18 Deere & Company Machine control using a predictive speed map
US11849672B2 (en) 2020-10-09 2023-12-26 Deere & Company Machine control using a predictive map
US11927459B2 (en) 2020-10-09 2024-03-12 Deere & Company Machine control using a predictive map
US11871697B2 (en) 2020-10-09 2024-01-16 Deere & Company Crop moisture map generation and control system
US11889787B2 (en) 2020-10-09 2024-02-06 Deere & Company Predictive speed map generation and control system
DE102021205386A1 (en) * 2021-05-27 2022-12-01 Robert Bosch Gesellschaft mit beschränkter Haftung Method for operating a hydraulic cylinder of a working machine
CN113837399B (en) * 2021-10-26 2023-05-30 医渡云(北京)技术有限公司 Training method, device, system, storage medium and equipment for federal learning model
CN116859830B (en) * 2023-03-27 2024-01-26 福建天甫电子材料有限公司 Production management control system for electronic grade ammonium fluoride production

Family Cites Families (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4495577A (en) * 1982-08-06 1985-01-22 Deere & Company Model reference adaptive implement control
US4527241A (en) * 1982-08-30 1985-07-02 Sperry Corporation Automatic combine harvester adjustment system
JPS6165307A (en) * 1984-09-05 1986-04-03 Konishiroku Photo Ind Co Ltd Control method of photographic emulsion manufacture process by nonlinear approximation
GB2186777B (en) * 1986-02-20 1989-11-22 Massey Ferguson Services Nv Ground speed control
US4967863A (en) * 1987-05-15 1990-11-06 Deere & Company Compact combine drive system
US4874963A (en) * 1988-02-11 1989-10-17 Bell Communications Research, Inc. Neuromorphic learning networks
GB8814936D0 (en) * 1988-06-23 1988-07-27 Ford New Holland Nv Combine ground speed control system
US4967544A (en) * 1989-01-05 1990-11-06 Deere & Company Automatic speed control system for a harvesting assembly
JP2656637B2 (en) * 1989-11-22 1997-09-24 株式会社日立製作所 Process control system and power plant process control system
CA2031765C (en) * 1989-12-08 1996-02-20 Masahide Nomura Method and system for performing control conforming with characteristics of controlled system
JPH04211802A (en) * 1990-07-25 1992-08-03 Toshiba Corp Neural network device
US5167009A (en) * 1990-08-03 1992-11-24 E. I. Du Pont De Nemours & Co. (Inc.) On-line process control neural network using data pointers
US5282261A (en) * 1990-08-03 1994-01-25 E. I. Du Pont De Nemours And Co., Inc. Neural network process measurement and control
US5142612A (en) * 1990-08-03 1992-08-25 E. I. Du Pont De Nemours & Co. (Inc.) Computer neural network supervisory process control system and method
JPH04252302A (en) * 1991-01-28 1992-09-08 Kawasaki Steel Corp Control model identifying method and control method for continuous rolling mill
US5251144A (en) * 1991-04-18 1993-10-05 Texas Instruments Incorporated System and method utilizing a real time expert system for tool life prediction and tool wear diagnosis
US5268834A (en) * 1991-06-24 1993-12-07 Massachusetts Institute Of Technology Stable adaptive neural network controller
US5253329A (en) * 1991-12-26 1993-10-12 The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration Neural network for processing both spatial and temporal data with time based back-propagation
GB2266602B (en) * 1992-04-16 1995-09-27 Inventio Ag Artificially intelligent traffic modelling and prediction system
US5353207A (en) * 1992-06-10 1994-10-04 Pavilion Technologies, Inc. Residual activation neural network

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111751133A (en) * 2020-06-08 2020-10-09 南京航空航天大学 Intelligent fault diagnosis method of deep convolutional neural network model based on non-local mean embedding

Also Published As

Publication number Publication date
DE69323943D1 (en) 1999-04-22
EP0586999A2 (en) 1994-03-16
AU658066B2 (en) 1995-03-30
EP0586999B1 (en) 1999-03-17
US5586033A (en) 1996-12-17
EP0586999A3 (en) 1995-03-29
JP3478855B2 (en) 2003-12-15
CA2105842A1 (en) 1994-03-11
AU3993093A (en) 1994-03-17
DE69323943T2 (en) 1999-08-12
JPH06176179A (en) 1994-06-24

Similar Documents

Publication Publication Date Title
CA2105842C (en) Neural network based control system
CA2383201C (en) Harvester with intelligent hybrid control system
EP3453239B1 (en) System for optimizing agricultural machine settings
CA1330113C (en) Combine ground speed control system
US7630809B2 (en) Method for controlling a harvesting machine
US8935060B2 (en) Driver assistance system for agricultural working machine
US9002594B2 (en) Agricultural working machine having at least one control unit
CN111837626A (en) Information inference for agronomic data generation in sugarcane applications
US6119442A (en) Combine setting autoadjust with machine vision
US5577373A (en) Agricultural vehicle including a system for automatically moving an implement and improved ground height sensing to a predetermined operating position
DK1446997T4 (en) Process for optimizing adjustable parameters.
US20120004813A1 (en) Method for adjusting at least one working unit of a self-propelled harvesting machine
CA2596471C (en) Engine load control for hydrostatically driven equipment
EP0122343A1 (en) Combine harvester
CA2414990C (en) On-board computer system having two operating modes
CN108934416A (en) One kind being based on BP neural network combined harvester multi-parameter operation-control system and method
US10897848B2 (en) Combine harvester with fan speed adjust
EP2840565A1 (en) Agricultural vehicle simulator
US11246264B2 (en) Harvesting control system with cutter bar load sensing
CN102929146A (en) Device and method for realizing model reference adaptive control on operation speed of combine harvester
JPS60178506A (en) Kind selector for automatic controller of agricultural implement and machinery
CN211090702U (en) Loss control system is cleaned in maize results based on CAN bus
EP3744167B1 (en) Virtual sensor for grain loss in combine harvester
Peschke et al. Model Predictive Control for Combine Harvesters using Geospatial Data
CA1129977A (en) Control of harvesters

Legal Events

Date Code Title Description
EEER Examination request
MKLA Lapsed