JPH07159238A - Signal processing method for infrared ray sensor and infrared ray sensor - Google Patents

Abstract

PURPOSE: To discriminated and evaluate a disturbance signal and an invader signal simply by means of small number of parts. CONSTITUTION: An infrared sensor is provided with a sensor element 1 and an evaluation circuit connected to its output side and consisting of an amplifier 3, an A/D converter 4 and a micro-controller 6 which includes a pulse processing step 5 for converting digitized sensor signals to pulses and a fuzzy controller 7 for comparing each of supplied pulse with regulations stored in form of language variable.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for processing and evaluating the signal of a passive infrared sensor which produces an electrical signal, which will be described below, as a sensor signal for incident infrared radiation.

[0002]

In the known method of this kind, the sensor signal is examined in the simplest case to see if it exceeds a positive or negative threshold, or the number of crossings with the threshold is exceeded. Are counted. It is also known that an alarm should be triggered if the positive (or negative) threshold is exceeded before the negative (or positive) threshold is exceeded.

All sensors based on simple thresholds are generally very susceptible to failure. The reason is that a single minor accident of sufficiently large amplitude can trigger a false alarm. On the other hand, a sensor in which some pulses are counted, which may or may not be polarity dependent, is especially sensitive when an intruder is at the border of the sensitive area or moves through only one zone of the application pattern. It loses sensitivity relatively quickly.

Detection devices are also known in which the sensor signal is continuously compared with a set of stored reference patterns and triggers an alarm when there is sufficient correlation. This detector is certainly very reliable and sensitive, but it has too many parts. This means that large, expensive processors must be used to supply the required storage capacity and power.

[0005]

According to the invention, a method of the type mentioned in the preamble of the claims is specified, in which the interference signal on the one hand and the intruder signal on the other hand are detected. A distinction between the broadly overlapping categories is certainly made with high sensitivity and especially also in the perimeter surveillance area. Moreover, a simple evaluation with a small number of components is possible and can be achieved with a simple microcontroller.

The purpose is to digitize the sensor signal and process it in the form of pulses, the pulses are characterized by data, and the evaluation of the pulses is performed by fuzzy logic, so that in each case there are several. It is achieved in that the data of the series of pulses is compared with the rules stored in the form of language variables.

The invention further relates to an infrared sensor for carrying out the method, which comprises at least one sensor element for generating a sensor signal and an evaluation circuit for processing and evaluating the sensor signal. .

The infrared sensor according to the invention is characterized in that the evaluation circuit comprises a fuzzy controller, to which the pulse data is supplied.

Due to the processing of sensor signals in pulse form,
An appreciable reduction in data rate occurs, which fulfills an important requirement for easy evaluation. As defined, the pulse begins when the signal leaves the normal position in the positive or negative direction and ends when it returns to the normal position. Data characterizing the pulse, such as amplitude, duration, polarity,
The intervals etc. are stored and a series of pulse
The data, i.e. the data of a series of several pulses, is always used.

If fuzzy logic is used instead of using strictly old logic in evaluating signals, the rules of evaluation are empirically science based (this is a rather laborious and awkward way of using conventional analytical algorithms). Can only be converted to)
Has the advantage that it can be based on.

In order to check the plausibility of the alarm,
A pulse data stream consisting of a number of leading pulses is available,
The criteria for triggering an alarm are thereby formulated in the form of fuzzy logic operations on the preceding fuzzy pulse data, i.e. transformed into linguistic variables. Thus, the criteria include a science-based in verbal form for the assignment of pulsed flows to intruder-class or failure-class, so that the science-based content was obtained from observations of myriad work tests and on disturbing signals. Represents all experiences.

The fuzzy fuzzy set, as described above, also gives a fuzzy result, the defuzzyfication of which gives a clear decision for or against the alarm. The rules for language form in the microcontroller of the evaluation circuit are:
Only minimal memory requirements are needed. In addition, fuzzyfication and defuzzification (ie, conversion into distinct numbers of obscure areas, or extraction of explicit statements from obscure areas) is more numerical than the old rule processing. The demand is quite small.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will now be described in detail with reference to an embodiment shown in the accompanying drawings. FIG. 1 is a block diagram showing an embodiment of an infrared sensor according to the present invention. In the figure, the infrared sensor comprises a sensor element 1, which is connected via an associated optical system 2 of a certain focal length. It receives infrared radiation from a space that is being monitored and depends on the incident radiation and then outputs an electrical signal, referred to as the sensor signal. It should be understood that the number of sensor elements 1 is not limited to one, and two or more may be used. The sensor signal is amplified by the amplifier 3, and its output signal is digitized by the analog / digital ((A / D) converter 4 and then the microcontroller 6
To the pulse processing stage which forms part of. The microcontroller 6 also includes a fuzzy controller 7.

The data rate of the digitized sensor signals is sharply reduced in the pulse processing stage 5 by storing such sensor signals as "pulses". As defined, such a pulse begins when such a sensor signal moves positively or negatively far enough from its normal position and ends when it returns to its normal position. Each pulse is represented by data that characterizes it, such as amplitude, duration, polarity, interval, etc., and these data are stored.

In a known manner, the optical system 2 corresponds to a number of optical focusing means and focuses infrared radiation from a number of fan-type radiation sensitive areas to the sensor element 1 (eg GB-A-2). , 047,886 or EP-A-
0,361,224) Includes a mirror system. These radiation sensitive areas are distinct zones, so that an object passing through such a zone upon incidence will produce a positive sensor signal and on exit it will produce a negative sensor signal, both characterized together. Generate a signal. Such characterization signals can be represented within a time interval as, for example, small positive pulses, large positive pulses, large negative pulses, and small negative pulses. During the evaluation, the pulses derived from the sensor signal are immediately examined there to see if they have a certain type and arrangement characteristic of human intrusion, and a group of several successive pulses. Is always under investigation.

What has been found in the example is that three or at most four such pulses can usually be obtained from a digitized sensor signal within the observation time window,
Investigating more than four pulses is impractical. In the method under investigation, the latest 4 pulses are always stored and investigated, and this investigation is based on the fuzzy controller 7
It happens inside.

FIG. 2 is a more detailed block diagram of the fuzzy controller 7 shown in FIG.
The controller 7, in a known manner, comprises a rule base 8, an inference machine 9, a process interface 10 and an action interface 11, which when it detects the presence of an unwanted intruder in the surveillance space obtains an alarm signal AS at its output. You can For fuzzy logic, there are various publications on this subject, such as Crew
See the book "Fuzzy Set Theory and Its Applications" by HJZimmermann, published in 1991 by Academic Publishers (Kluwer Academic Publishers).

In a known manner, the rule base 8 comprises a set of linguistic rules for the evaluation of pulses. An algorithm is constructed based on these linguistic rules, whereby values are defined as so-called fuzzy sets, or unclear quantities. A language variable is a word or expression in a national or natural language. These language variables are natural language expressions (small, medium,
Larger) should be acceptable and these representations are the names of the fuzzy sets mentioned above.

Like old logic, the rules of fuzzy logic consist of a conditional or predicate part and an inference part. In FIG. 2, the condition part is symbolized by the process interface 10 and the inference part is symbolized by the action interface 11. The reasoning machine 9 links the direction of influence with the magnitude of the instantaneous states in the fuzzy set based on empirical technical knowledge.

FIG. 3 is an operational illustration showing the main features of the fuzzy controller 7 with the help of a graphical representation with typical fuzzy rules. Rule 1 in FIG. 3 is “if A = large and B = normal, then X = small”
Rule 2 is, for example, "if B = normal and C = small, then X = normal". Here, A, B and C are input variables. The sentence part beginning with "if" is the condition part, and the part beginning with "then" is the inference part.

The central term of fuzzy logic is a fuzzy set, that is, an indefinite quantity, whereby the membership of elements in a fuzzy set is determined by a so-called membership function. 1 is a affiliation and 0 in a definite amount
Means a non-affiliation relationship, but 0 for fuzzy sets
Or, not only 1 but any value between 0 and 1 can be accepted as the value of the membership function.

The conversion from explicit numbers to indeterminate quantities is termed fuzzification, whereby each input variable or in fact, for example, a signal has at least one function represented as a matrix. The x-axis of this function has the numerical equivalent in each sensor signal,
And the y-axis corresponds to the true content, ie the degree of approximation to the corresponding statement, and can take all values from 0 to 1. The degree of this approximation is calculated by the membership function.

Because of the statement in the conditional part, the variable of the belonging relation value is searched by an appropriate operator, and if this variable is the minimum value of the belonging relation function, then the operator is as shown in FIG.
Is the smallest operator as in and this is the average of the two fuzzy sets of input variables A and B. The result of the inference of the two rules, Rule 1 and Rule 2, is therefore the average over the fuzzy set of A and B, or B and C.

The clear output variables are these inferences (FIG. 2).
It is calculated from the action interface 11). If inference is available from several rules, as in FIG. 3, then the membership values of each rule are combined. This is accomplished, for example, by comparing the inferred parts of the rules to obtain the maximum of their membership values and creating a new membership function. This process is called the maximum operator, which represents the unity of the inference part.

Due to the obscure results obtained by the inference machine 9 (FIG. 2), an explicit output variable is calculated, which is done, for example, by calculating the center of mass of the composite membership function.

The design of the fuzzy controller 7 (FIGS. 1 and 2) is generally carried out in the following steps: Definition of all input and output variables: In this example,
The input variables are the data characterizing the pulse obtained from the sensor signal, and the time window, and the output variables are the values that simply describe whether an error or an unrecognized entry is included. 2. Definition of uncertain quantities (fuzzy sets) for language variables. 3. Rule Creation: For example, a suitable rule is that if the condition of a pulse stream consisting of three successive pulses with small positive, large negative, and small positive amplitudes is met for long periods in the time interval. For example, the entry not permitted at that time will be applied. 4. A set of inference machines: operators such as a specific AN
For the D function, the so-called fuzzy AND of F = y * min (A, B) + 0.5 * (1-y) * (A + B) is chosen, where A and B are input variables and y is It is a gamma factor. Since the gamma factor y = 1, the fuzzy AND operator becomes the minimum operator (FIG. 3). 5. Definition of the calculation of the explicit output variable: This operation is called defuzzification, where the explicit variable is derived from an unobtrusive quantity by the output membership function, and is implemented by producing the center of mass as in FIG. Is preferred.

[0027]

The above-mentioned signal processing in the infrared sensor is
It facilitates good and accurate discrimination between intruder signals and jamming signals with high sensing performance. In particular, extremely noisy signals and signals from surrounding surveillance areas can also be clearly evaluated. The storage of the sensor signal in the form of pulses will significantly reduce the memory capacity, in particular the memory capacity required for rules in the linguistic form. In addition to this, fuzzy and defuzzification have relatively low numerical demands and require only simple outputs (which can be easily realized with a simple microcontroller).

Because of the obscure formulation that represents fuzzy logic, it is unlikely that a signal will be rejected because it barely escapes a state. The process described above is rather very different and corresponds to an obscure intruder signal. Due to the fuzzy formulation, the algorithm in mind is simple and straightforward. Once written, it is also valid to change states, in which case only certain constants have to be corrected (so-called parameterization). The constants are optimized based on tests and simulations.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an infrared sensor according to the present invention.

FIG. 2 is a more detailed block diagram of the fuzzy controller shown in FIG.

FIG. 3 is a graph for explaining the operation of the present invention.

[Explanation of symbols]

 1 Sensor element 6 Micro controller 7 Fuzzy controller 8 Rule-based 9 Inference machine

Claims (9)

[Claims] 1. A method of processing and evaluating a signal of a passive infrared sensor which produces an electrical signal, which is described below as a sensor signal for incident infrared radiation, wherein said sensor signal is digitized and processed in the form of pulses, The pulse is characterized by data, and the evaluation of the pulse is carried out by fuzzy logic (7), so that in each case a series of several pulses of data are compared with rules stored in the form of language variables. A method for processing a signal of an infrared sensor, comprising: 2. The waveform of the sensor signal is examined for conversion of the digitized sensor signal into pulses,
2. Infrared sensing according to claim 1, characterized in that the start of the pulse is set at a certain displacement of the sensor signal from its normal position and the end of the pulse is set at the return to the normal position. Signal processing method. 3. The signal processing method for an infrared sensor according to claim 2, wherein at least one of an amplitude and a duration of the pulse is used as characterization data for explaining the pulse. 4. The signal processing method for an infrared sensor according to claim 2, wherein at least one of polarity and mutual interval is used as characterization data for explaining the pulse. 5. The data of the last n successive pulses in each case are compared with a stored rule, the value of which lies between 2 and 4 and is preferably 3. 4. A signal processing method for an infrared sensor according to any one of items 1 to 3. 6. Infrared sensing for implementing the signal processing method of claim 1, comprising at least one sensor element for generating said sensor signal and an evaluation circuit for processing and evaluating said sensor signal. Infrared detector, characterized in that the evaluation circuit comprises a fuzzy controller (7), to which the pulse data of the fuzzy controller is supplied. 7. Infrared according to claim 6, characterized in that the fuzzy controller (7) comprises a rule base (8) and a reasoning machine (9) and forms part of a microcontroller (6). sensor. 8. The rules stored in the rule base (8) of the fuzzy controller (7) are of a type whose conditional part includes the number and data of the pulses and the time intervals of the occurrence of the pulses. The infrared sensor according to claim 7, wherein 9. The infrared sensor according to claim 7, wherein the inference machine of the fuzzy controller has a fuzzy AND function as an operator.