EP0572302A1 - Alarm system and method by analysis of the signal of a sensor - Google Patents

Abstract

The present invention relates to an alarm system and method, by analysis of a reception signal originating from a sensor detecting a standing wave regime set up in a vehicle passenger compartment, especially by ultrasonic waves, in such a way as to detect an intrusion into the passenger compartment via the changes in the said standing wave regime. According to the invention, the alarm system includes a microcontroller (M) which produces setting-up signals for a receiver (2) during the learning periods in order to set up, by anticipation, the sensitivity of the system. Vehicle alarms. <IMAGE>

Description

The present invention relates to an alarm method and system by analyzing a reception signal from a sensor detecting a standing wave regime established in a passenger compartment of the vehicle, in particular by ultrasonic waves, so as to detect by modifications to said standing wave regime intrusion into the passenger compartment.

In the state of the art, methods of analysis are known which include the step of measuring the level of the signal received by a sensor, then, by measuring the signal level at various frequencies, of distinguishing a shock from a intrusion.

However, a particular problem arises for methods of this kind in that the detection of an intrusion is a very rare event and which can be very brief. In addition, the intrusion detection signal can be extremely similar to signals from other events such as, for example, impacts on the vehicle, thermal or hygrometric modifications, in particular in the case of ultrasound, etc.

Therefore, there is a need for an analysis method making it possible to more clearly distinguish the intrusion from other events such as the movements of air mass of thermal origin or from shocks at a distance such as the start-up of '' a jackhammer on the public road when the vehicle is stationary, for example, and its alarm is activated.

It is an object of the present invention to provide a method which overcomes this drawback of the prior art.

In the prior art, methods have also been proposed which make it possible to adapt the processing chain of the received signal as a function of changes in particular in terms of the characteristics of the components of the chain and / or physical parameters such as the temperature with humidity and pressure inside the passenger compartment. In particular, this kind of process is intended to overcome false alarms caused for example by a sudden increase in sunshine.

However, known adaptation methods are based on an automatic gain correction (AGC), method according to which, by means of signal integration, the average level of the received signal is found and the feedback loop is increased. the gain of the processing chain so as to obtain a level as high as possible below a threshold.

In particular, when the signal level becomes too high, the automatic gain correction makes it possible to lower or decrease the overall gain, at least on the first stages, so as to lower the output signal from the processing chain below a certain level. threshold.

If one is in a standing wave node (and this can last a long time), at this movement the AGC gives the maximum gain. If the overall gain of the chain is high, and the intrusion which occurs at this time is rapid and causes an unfavorable displacement of the standing wave regime, the detection chain is saturated and it no longer makes it possible to detect the phenomenon of 'intrusion.

It is another object of the present invention to propose a means of adapting the processing chain which makes it possible to retain the possibility of detecting an intrusion even in the aforementioned adverse cases.

• à effectuer une correction des désadaptations de la chaîne de traitement, par anticipation, lors de phases répétées d'apprentissage, puis
• à détecter les bruits notamment d'origine thermique ; et
• à détecter les évènements aléatoires ne correspondant pas à des intrusions comme les chocs ; et enfin
• à détecter les intrusions proprement dites, lors de phases répétées de scrutation.

In fact, the present invention relates to a method for analyzing the detection signal of at least one sensor, such as an ultrasonic sensor, which detects the movements of a standing wave regime in the passenger compartment of a vehicle so to detect a possible intrusion. The invention is characterized in that it consists:

• to correct the mismatches in the processing chain, in advance, during repeated learning phases, then
• detecting noise, particularly of thermal origin; and
• detect random events that do not correspond to intrusions such as shocks; and finally
• to detect the intrusions proper, during repeated scanning phases.

The invention also relates to an alarm system comprising at least one transmitter producing a standing wave regime in the passenger compartment of a vehicle under alarm, and at least one receiver, characterized in that it comprises means for carrying out an adaptation of the reception characteristics of the system, the output of which is connected to processing means which in particular comprise means for extracting the detection signal from thermal noise, then means for extracting the intrusion signals from the detection signals and means alarm.

• la figure 1 : un schéma d'ensemble d'une chaîne de traitement mettant en oeuvre le procédé de l'invention ;
• la figure 2 : un schéma d'un mode de réalisation du récepteur du système de la figure 1 ;
• les figures 3a et 3b : des courbes représentatives du signal reçu et traité selon le mode de réalisation de la figure 2 lors d'un premier type de phases dans le procédé de l'invention ;
• la figure 4 : un organigramme des fonctions, qui permettent d'exécuter une détection du signal de réception en s'affranchissant largement du bruit ;
• les figures 5 et 6 : des caractéristiques spectrales de réception ;
• la figure 7 : un mode de réalisation d'un algorythme qui est utilisé dans un moyen de traitement du mode de réalisation de la figure 4.

Other characteristics and advantages of the present invention will be better understood with the aid of the description and the figures which are:

• FIG. 1: an overall diagram of a processing chain implementing the method of the invention;
• Figure 2: a diagram of an embodiment of the receiver of the system of Figure 1;
• Figures 3a and 3b: curves representative of the signal received and processed according to the embodiment of Figure 2 during a first type of phase in the method of the invention;
• FIG. 4: a flow diagram of the functions which make it possible to carry out detection of the reception signal while largely avoiding the noise;
• FIGS. 5 and 6: spectral reception characteristics;
• FIG. 7: an embodiment of an algorithm which is used in a processing means of the embodiment of FIG. 4.

In Figure 1, there is shown a preferred embodiment of an alarm system implementing the method of the invention.

The alarm system comprises at least one module 1 emitting ultrasonic waves which includes a capsule emitting ultrasonic waves when an electrode thereof is activated by oscillations, for example at 40 kilohertz produced by an oscillator circuit 5 which it is connected to it by a wired link 6.

The oscillator is connected to an output port 10 of a microcontroller M which supplies it with a control and / or power signal.

The alarm system also includes at least one receiver 2 consisting of a capsule 14 for receiving ultrasonic waves, the output 16 of which is connected to an input of an amplifier 15 with controllable gain. The amplifier 15 includes at least one gain control input connected to at least one other output port 13 of the microcontroller M.

The receiver 2 also includes a demodulator circuit 18 connected to an output 17 of the amplifier 15 with controllable gain. The output 19 of the demodulator 18 is connected to a port 20 for analog digital conversion of the microcontroller M.

The microcontroller M also comprises at least one input port 9 connected by a suitable link 8 to a sensor 7 for measuring the environment such as a temperature sensor or a relative humidity sensor, or an atmospheric pressure sensor, etc. .

The microcontroller M also comprises at least one input port 22 connected by a suitable link 24 to a control device 25 constituted for example by at least one push button operated by the user of the alarm system to activate the alarm system. this control device can be supplemented by command input means making it possible in particular to reconfigure the microcontroller M, or to load it with a new work program, or new data.

The microcontroller M also includes an output port 21 connected to a device 23 for displaying voice type messages (using a voice synthesizer), or sound (using a buzzer or other speaker ) or alphanumeric (using a console or a liquid crystal panel), or of viewing information in light and / or colored form (by means of light-emitting diodes, colored or not, the lighting and extinction indicates the state of the alarm system).

The microcontroller M comprises at least one output port 26 connected to a control input 27 of an alarm device which can be constituted in particular by an alarm siren and / or by a means producing a function invalidation signal of the protected vehicle as a signal to inhibit the injection or ignition computer, a signal to block the alternator, etc.

Finally, the microcontroller M has an electrical power supply input 11 connected to a power supply device, for example constituted by a voltage regulator connected to the vehicle battery, or on any other electricity generating device and in particular on a power supply. uninterruptible power supply to resist power outages or sabotage while the alarm system is in operation.

In Figure 2, there is shown a mode of realization of the receiver 2 of the system of FIG. 1.

The capsule for receiving the ultrasonic waves is here represented by the reference 30 and is connected between ground and a suitable input 31 of a first amplifier 32 with controllable gain whose gain control input 33 is connected by a suitable link 34 to an access terminal 46 of the module 49 integrating the microcontroller M. This access terminal 46 delivers a control signal whose value represented in particular by the DC voltage level makes it possible to determine the gain which it is desired to fix on the first amplifier 32.

The amplified output 35 of the amplifier 32 is connected to an input of a demodulator 36, the output 37 of which is connected to the input of a first filter 38, of the low-pass type, the output 39 of which is transmitted from a part at an input of a second filter 40, of high-pass type, and by a link 42 to an access 52 of the aforementioned module 49 and which is connected to a first analog-digital conversion port CAN1 of the microcontroller M.

The output 41 of the high-pass filter 40 is connected to the input of a second amplifier 43 with controllable gain, the gain control input 44 of which is connected by a suitable link 45 to an access terminal 47 of the module 49.

The amplified output of the second amplifier 43 is connected by a suitable link to an access terminal 48 of the module 49, and is transmitted to a second analog-digital conversion port of the microcontroller M.

In the embodiment of FIG. 2, the part of the assembly disposed to the left of the access terminals 46, 47, 48, 52 of the module 49 is of the analog type and is constituted by a standard circuit suitable for any type of detection according to the invention, for any vehicle. The part arranged to the right of the same terminals, and which has been designated under the term of module 49, is rather of digital nature and is specific to a type of vehicle, to a given type of application. The personalization, or specialization, of each module is carried out in software so as to facilitate the standardization of the mass production of the alarm system according to the invention, which is a notable advantage of the present invention, in the measure where it allows a significant reduction in manufacturing costs. However, this reduction in costs is directly linked to the architecture of the assembly and to the particular nature of the process used, as will be seen.

The method of the invention consists in providing a succession of learning and scanning phases, succession determined in advance, but certain phases of which can be postponed depending on the circumstances.

When the alarm system according to the invention is started, one begins with an initial learning phase during which the system transmitter 1 sends a wave train of determined shapes. The emission of a wave train is executed in such a way that the receiver 2 detects at the start and / or at the end of each pulse an ultrasonic wave in progressive mode, characteristic of the volume and of its conventional conditions.

The microcontroller, which collects the data, then performs a learning process on the assumption that there is no intrusion in progress.

To this end, the microcontroller stores the minimum value and the maximum value of the received wave. In Figure 3, there is shown the case of such reception. When the train of waves transmitted OE, having a particular envelope ENV-E obtained by a suitable control of the oscillator 5 thanks to the output port 10 the microcontroller activated according to a particular program recorded in a memory area not shown of the microcontroller M, is received by the receiver 2, the latter develops a response represented by the OR curve, representative of the received wave.

The signal received by the receiver 2 is transmitted to the analog-digital conversion port 19 of the microcontroller M, a particular program of which, recorded in a memory area not shown of the microcontroller M, makes it possible to store the minimum value Vmin and the maximum value Vmax.

$$\text{<Vmin> = (Somme (Vmin(i)))/Nt}$$

In one embodiment, this pair of values is obtained after the emission of Nt trains of similar waves OE and is calculated by an average given by: $$\text{<Vmax> = (Sum (Vmax (i))) / Nt}$$

$$\text{<Vmin> = (Sum (Vmin (i))) / Nt}$$

Other averages are possible depending on the shapes of the OE wave trains, in particular by increasing the order of the average and by supplementing the relationships with weighting coefficients.

In one embodiment, the learning phase is completed when the pair of acquired values (Vmax, Vmin) is stable, that is to say that it no longer varies during the emission of new OE wave trains. . Typically, this situation occurs after a few hundred milliseconds for 50 millisecond wave trains with ultrasound waves at 40 kilohertz. Therefore, according to another embodiment, there is a predetermined number of acquisitions corresponding to a sufficient time for the ultrasound system to be stabilized.

The microcontroller M deduces detection data, using pre-recorded tables or by calculation according to a pre-recorded law, from adaptation values of the processing chain. In one embodiment, these adaptation values are constituted by a gain control value applied to the first amplifier 32 by the microcontroller M by means of the link 34.

   où Vsat est la valeur de saturation de la chaîne de traitement et G est une valeur de garde prédéterminée en fonction des caractéristiques de la chaîne de réception.The gain value of the first amplifier makes it possible to bring the average value Vm of the reception signal SR (FIG. 3b) to a value such that: $$\text{Vm <Vsat - G}$$

where Vsat is the saturation value of the processing chain and G is a predetermined guard value according to the characteristics of the reception chain.

Then, the microcontroller M develops as adaptation values a gain value of the second amplifier 43, which is transmitted to it by a link 45 at its input 44, such as the maximum peak value of the output signal of the high pass filter 40 of the processing chain (in which the reception signal loses its DC component) is still less than the saturation value with a guard value g which corresponds in particular to the detection value of a possible intrusion. For this, the microncontroller M ensures that: $$\text{Vcm <Vsat - g.}$$

According to the method of the invention, it is therefore understandable that the reception chain will be suitable for maximum sensitivity detection without having to fear saturation at the time of detection of a possible intrusion.

In particular, this learning phase is repeated at predetermined intervals, for example with a predetermined periodicity, repetition which is interrupted upon detection of an intrusion. In addition, in a preferred embodiment, to avoid untimely adjustments of the sensitivity, that is to say of the two aforementioned gain values (adjustments which carry out the adaptation of the processing chain to its environment), the modifications of the previous settings adaptation values are only accepted if they are confirmed after several learning phases, for example five phases separated by 500 milliseconds.

In FIG. 4, a flow diagram of the functions is shown, which in the invention is implemented in the microcontroller M in the form of a program, so as to carry out a detection of the reception signal while largely eliminating the noise.

FIGS. 5 and 6 show spectral reception characteristics according to the invention.

In FIG. 4, the microcontroller M comprises an input routine or subroutine 60 of the data X (p), with p representing the order of the digital sample acquired on the analog conversion port 48 at the output of the second amplifier analog 43 (figure 2).

The flow of sequential data X (p) is supplied to a memory, the addressing of which makes a switch 61 towards a plurality of high-pass and / or low-pass filterings carried out by operators F1, F2, ..., FN in number N and which operate in parallel.

• un opérateur de filtrage 63 dont l'entrée 62 reçoit une donnée X(p) et dont la sortie 64 produit une valeur filtrée X(p) ;
• un opérateur de démodulation 65 qui permet de récupérer l'ondulation provenant d'une détection par rapport à l'onde directe provenant d'un émetteur et qui produit en sortie une valeur démodulée X(p)* ;
• un opérateur 67 de mesure de niveau qui produit un signal sur une entrée 69-1, connectée par un chemin 68 à la sortie de l'opérateur 67, d'un moyen 70 de sélection qui comporte des entrées 69-2, ..., 69-N connectées aux sorties respectives des filtres F2, ..., FN.

Each operator Fi, like the operator F1, successively comprises:

• a filtering operator 63 whose input 62 receives data X (p) and whose output 64 produces a filtered value X (p);
• a demodulation operator 65 which makes it possible to recover the ripple coming from a detection with respect to the direct wave coming from a transmitter and which produces at output a demodulated value X (p) *;
• an operator 67 for level measurement which produces a signal on an input 69-1, connected by a path 68 to the output of the operator 67, of a selection means 70 which comprises inputs 69-2, ... , 69-N connected to the respective outputs of filters F2, ..., FN.

The selection means 70 also includes a first output 71 of the selected average, demodulated and filtered, transmitted to a means 76 for processing and recognizing the levels and the durations, and a second output 72 which transmits the number of the filtering channel Fi to be chosen as will explain it.

FIG. 5 shows the spectral characteristics recognized in the present invention.

There is, at very low frequencies, a noise of thermal origin Br. This noise varies in a variable manner from continuous to a frequency of a few hertz, (typically, 10 Hz). It has an extension and a gradient of spectral decay Brl, Br2, Br3, Br4 which vary over time, which makes it difficult to detect a reception signal, which is often lower in Av level than that of noise. .

For this, a high pass type digital filtering is used according to an FPH spectral characteristic which is adapted according to the circumstances.

Using high pass filtering, it is possible to output the signal from the noise level, even if the signal level is lower than the noise level. However, it turns out that the spectrum of intrusion signals is most often greater than the noise limit frequency. Therefore, if the noise level is removed, it is possible to detect the signal outside of the noise which masked the signal to be detected.

• fréquence de coupure la plus basse possible ;
• niveau de détection minimum en l'absence d'une intrusion (notamment pendant une phase d'adaptation).

For this purpose, one of several filterings F1, F2, ..., FN is used, the cut-off frequencies of which are distributed from a few Hz to more than 10 Hz, possible noise limit frequency. However, you must be able to select the Fi filter (or filtering channel) which meets at least one of the following conditions:

• lowest possible cut-off frequency;
• minimum detection level in the absence of an intrusion (especially during an adaptation phase).

However, the FPH filtering is performed digitally, for example on P0 samples X (p) according to a formula of the type: $$\text{<X (p)> = {a (p - P0 + 1) .X (p- P0 + 1) + a (p - P0 + 2) .X (p- P0 + 2) + ... + a ( p) .X (p)} / P0}$$

We can use relations where the weighting coefficients a (i) are equal, and relations where the samples X (p - k) are raised to an integer power b (filtering of order b).

However, before the first value <X (p> is available, a certain period of time elapses during which the P0 successive samples X (p) are acquired and during which the average value <X (p)> is calculated.

Therefore, if the chosen filtering channel Fi corresponded to a situation where the noise level was higher than the level of the filtered signal, it might be too late to change the filter Fi if an intrusion took place at this time.

According to one aspect of the invention, which is not necessary but preferred, the fact of having a battery of filtering channels makes it possible, during the adaptation (or learning) step, to early select the filtering channel Fi0 matched which produces on a 69-i0 output a minimum voltage value. This value i0 is transmitted by the means 70 to a selection member by a link 74 connected to its output 72 and which makes it possible to direct the following data acquired during the next step of scanning (or monitoring) to the channel of filtering Fi0 suitable.

In FIG. 6, another spectral characteristic has been shown in which the filters all have the same cut-off frequency fc, but slopes of order 1, order 2, etc., respectively. The filtering having the lowest slope (lowest order) possible and / or at minimum detection in the absence of an intrusion.

It is possible to combine batteries of high pass filters using in combination the two methods of varying the slope (filtering order) or the cutoff frequency).

In a preferred embodiment, the filters are successively calculated using a single subroutine, to which, during the previous adaptation phase, a command (equivalent to the aforementioned switching command 74) is transmitted which regulates the frequency of cutoff and / or filter order.

Returning to FIG. 4, we will explain the operation of the means 76 for processing and recognizing the levels and the durations.

In the invention, the inventors have measured that in general, the events which did not correspond to intrusion situations and which triggered the false alarms of the systems of the prior art, have close spectral characteristics, but durations and different levels.

To this end, the processing means 76 performs a measurement of the levels received <X (p> * selected and compares them to predetermined thresholds. When it is observed that the duration of a received signal, of level higher than the predetermined threshold, exceeds a certain value, the output 77 of the processing means 76 is placed at an active level which activates an alarm device or any alarm action indicating that an intrusion has occurred.

In FIG. 7, there is shown an embodiment of an algorithm which is used in the processing means 76.

The data which comes from the filtering means Fi selected as described above, are here designated by X (n). They are sampled at a predetermined frequency, equal to 100 Hz in a preferred embodiment, during a step S1.

• d'un filtrage analogique du premier ou du second ordre, suivi
• d'un filtrage numérique du deuxième ordre.

They undergo a filtering which can be carried out by the combination:

• first or second order analog filtering, followed by
• second-order digital filtering.

Therefore, a high-pass filtering is carried out at a cutoff frequency greater than a noise threshold frequency, equal in a preferred embodiment to 10 Hz.

The output X1 (n) is transmitted to a demodulation means, which performs, during a step S3, the calculation of the absolute value of the filtered signal.

The output | X1 (n) | = X2 (n) is subjected during a step S4 to a smoothing. In one embodiment, this smoothing is performed by a low-pass filter having a cutoff frequency equal to 5 Hz and a slope of order 1.

The smoothing output X3 (n) is then subjected to a test during step S5 so as to detect that the value X3 (n) is greater than a predetermined threshold S which is a predetermined function f of a parameter linked to the system sensitivity level: S = f (N). This threshold value is used, at least in part, to define an intrusion threshold which makes it possible to trigger an alarm action.

If the result of the test S5 is positive (YES), we go to a step S11 where a first counter C is incremented by one and a second counter D is reset to zero.

During a test S12, it is detected whether the first counter C is less than a predetermined value Cmax. This value Cmax is taken equal to 80 in one embodiment, which corresponds to a duration of 800 milliseconds for a detection signal level greater than the detection threshold. This makes it possible to correctly detect, without the risk of false alarms, an intrusion.

If the response to test S12 is positive (YES), we An alarm situation is therefore deduced therefrom and an alarm action is executed during step S13.

If the response to test S12 is negative (NO), we go to an "end" routine during which we test if we must go into the learning / adaptation phase (not shown in Figure 7) or if we must remain in the scanning or detection phase, in particular by testing the transition to the active state of a deactivation signal supplied by a vehicle access management center (locking center, etc.).

If no action is in progress, we return to the acquisition (step S1).

If the response to the test S5 is negative (NO), it is tested during a step S6 if the current value of the first counter C is not zero.

If the response to test S6 is negative (NO), we jump to the above-mentioned "end" routine.

If the response to the test S6 is positive (YES), a second counter D is incremented, which therefore measures the duration during which the level of the detection signal is below the intrusion threshold S.

Then, a test S8 is executed to find out whether the current value of the second counter D is greater than a threshold value N which moreover serves to determine the detection threshold S of step S5. In the executed embodiment of the invention, N = 30 was chosen which corresponded to a duration of 300 milliseconds, less than the duration Cmax of 800 milliseconds on the first counter. It can therefore be seen that the two counters C and D make it possible to overcome situations of false alarms in which, in particular, except during periods of silence less than the second maximum duration of 300 milliseconds, repeated shocks for example, could be confused with intrusions.

If the response to test S8 is negative (NO), we go to step S10 of "end" above.

If the response to test S8 is positive (YES), the two counters C and D are reset to zero, and the routine "end" mentioned above is passed.

The present invention can be implemented in other forms than those described above. In particular, it is possible to use entirely digital processing means, or even entirely analog means without going beyond the scope of the appended claims.

Claims (8)

Method for analyzing the detection signal of at least one sensor, such as an ultrasonic sensor, which detects the movements of a standing wave regime in the passenger compartment of a vehicle so as to detect a possible intrusion, characterized in what it consists of: - to correct the mismatches of the processing chain in advance;
during repeated learning phases, - to remove the level of noise, particularly of thermal origin; and - to detect random events that do not correspond to intrusions such as shocks, and finally - detect intrusions proper; during repeated scanning phases. Method according to claim 1, characterized in that, during at least one learning phase, a wave train of predetermined shapes is emitted, then the detection data is collected by producing adaptation values of the chain processing so that the data collected is stable. Method according to claim 2, characterized in that changes to previous settings of the adaptation values are only accepted if they are confirmed after several learning phases. Method according to claim 1, characterized in that the step of removing the noise level comprises the step of choosing from several filterings (F1, F2, ... FN) operating simultaneously, that whose spectral characteristic responds at least to one of the conditions: - lowest possible cut-off frequency, and / or - lowest possible slope, and / or on condition of a minimum detection level in the absence of an intrusion. Method according to claim 4, characterized in that the step of choosing a filtering (F1, F2 ... FN) takes place in advance during a learning phase. Method according to claim 1, characterized in that, when the detection value is greater than a predetermined intrusion threshold value, a counting of a first duration (C) is started and when the detection value then becomes less than said intrusion threshold value (S), a counting of a second duration (D) is started so as to retain only among the detection signals those: - which have a first duration (C) at least equal to a predetermined threshold value (Cmax); and - which have no interruption at this higher value for a period (D) at most equal to a predetermined threshold value (N), so as to overcome situations of false alarms. Alarm system comprising at least one transmitter (1) producing a standing wave regime in the passenger compartment of a vehicle under alarm, and at least one receiver (2), characterized in that it comprises means (M ) to adapt the reception characteristics of the system, the output of which is connected to processing means which in particular comprise means for extracting the noise detection signal, then means for extracting the intrusion signals from the detection signals and alarm means. System according to claim 7, characterized in that it comprises a first amplifier (32) with gain controllable (33) by a microcontroller (M), which produces an output signal whose amplitude is adapted, during a phase learning, at a level below one threshold value in advance, signal transmitted to a low-pass filter (38), the output of which is transmitted on the one hand to the input of a low-pass filter (40) and of an analog-digital conversion port of the microcontroller (M); the output of the high pass filter (40) is connected to the input of a second amplifier (43) with controllable gain (44) by said microcontroller (n);
the output of the second amplifier (43) being connected to another analog digital conversion port of the microcontroller (M);
the microcontroller (M) comprising means for carrying out the adaptation by anticipation of the system by adjusting the first and second gains.

Images