EP0646901B1 - Method for processing passive infrared detector signals and infrared detector for carrying out the method - Google Patents

Description

The present invention relates to a method for processing the signals of a passive Infrared detector, which depends on an incident infrared radiation in the following generates electrical signals referred to as sensor signals and then evaluates them, wherein the sensor signals are digitized and processed in the form of pulses, their evaluation by means of fuzzy logic (7), the data being a series of several Pulses are compared with rules stored in the form of linguistic variables,

In known methods of this type, in the simplest case, either the sensor signal on the Exceeding or falling below positive or negative threshold values is examined, or it is counted the number of threshold passes. It is also known to trigger an alarm if a successive crossing of a positive and a negative threshold or vice versa.

In principle, all detectors based on simple threshold values are very susceptible to failure, since a single interference pulse with a sufficiently large amplitude can trigger a false alarm. On the other hand, those detectors in which several pulses are counted are lost this depends on the polarity or not, relatively quickly in sensitivity, especially when an intruder is at the edge of the detection area or only through a zone of the Covering pattern moves.

Detection systems are also known in which the sensor signal is continuous with a Set of stored reference patterns compared and an alarm if there is sufficient correlation is triggered. Although these systems are very reliable and sensitive, they do require a high numerical effort. And that means that in the detector to provide the necessary storage capacity and performance a large and thus expensive processor can be provided got to.

In a device for detecting a human being described in JP-A-50 018 827 The signals from an infrared detector are processed with fuzzy logic, the body Presence of a human body is then assumed if that from the infrared detector measured temperature is in the range of a person's body temperature. Not to mention of the fact that it is difficult to absolute temperature at a distance of several meters in a clothed person essentially the temperature of his Clothes and not those of his body, but in the case of a coat the clothes rather Will have room temperature as body temperature. For this reason, this is known device unsuitable for use as intrusion detector.

The invention is now to provide a method of the type mentioned at the beginning a good discrimination of the strongly overlapping classes of interference signals on the one hand and intrusion signals on the other hand, with a high detection power and in particular also in the peripheral surveillance area. In addition, a simple evaluation possible with little numerical effort and realizable with a simple microcontroller be.

According to the invention, this object is achieved in that the pulses are characterized by data are used, with the amplitude and / and / or or duration can be used and that to convert the digitized sensor signals into Pulse the course of the signals is examined, and at a certain distance of the signal from a pulse start is set in its rest position and a pulse end is set on returning to the rest position.

The invention further relates to an infrared detector for carrying out the method mentioned, with at least one sensor element for generating the sensor signals and with an evaluation circuit for their processing and evaluation, which contains a fuzzy controller to which the Pulse data are supplied, characterized in that the fuzzy controller has a rule base and contains an inference machine and forms part of a microcontroller which has a pulse processing stage for processing the digitized sensor signals in pulses and for storing the contains the data describing pulses.

By processing the sensor signals in pulse form, the data rate is greatly reduced instead, which creates an essential prerequisite for simple evaluation. By definition A pulse begins when the signal is in the positive or negative direction away from the rest position and it ends when it returns to the rest position. They will be the ones Pulse-characterizing data, such as amplitude, duration, polarity, distance and the like, are stored, and there is always a pulse data series for the evaluation, that is the data of one Series of successive pulses.

The use of fuzzy logic instead of strict classic logic for signal evaluation has the advantage that the rules for evaluation are based on an empirical Knowledge base that can be used in classic analytical algorithms only would be much more cumbersome and complex to implement.

The pulse data series of a number always stands for the examination for alarm plausibility past pulses are available, the criteria for triggering the alarm in Form of fuzzy links of the previously fuzzified, that is, in linguistic Variable transformed, pulse data are formulated. So the criteria include in linguistic form a knowledge base about the assignment of a pulse series to the class Burglary or disruption, the content of the knowledge base being the entirety of the observation of countless walking tests obtained and those made with interference signals Represents experiences.

The fuzzy sets that are by definition out of focus deliver an equally blurred result, whose defuzzification provides a sharp decision for or against an alarm. The rules in linguistic form in the microcontroller of the evaluation circuit need minimal memory requirements. In addition, the fuzzification and defuzzification, So the conversion of sharp numbers into fuzzy areas, or the Obtaining clear or sharp statements from fuzzy areas, numerically much less demanding than the processing of classic rules.

Fig. 1

ein Blockschaltbild eines Ausführungsbeispiels eines erfindungsgemässen IR-Detektors,

Fig. 2

ein Detail des Schaltbildes von Fig. 1; und

Fig. 3

ein Diagramm zur Funktionserläuterung.

The invention is explained in more detail below using an exemplary embodiment and the drawings; show it:

Fig. 1

2 shows a block diagram of an exemplary embodiment of an IR detector according to the invention,

Fig. 2

a detail of the circuit diagram of Fig. 1; and

Fig. 3

a diagram to explain the function.

1, an IR detector according to the invention contains a sensor element 1, which has an associated optics 2 of a certain focal length with IR radiation from the room to be monitored and depending on the Level of the incident radiation, hereinafter referred to as the sensor signal emits electrical signal. The use of a single sensor element 1 is not to understand restrictively; it can of course also be two or more Sensor elements may be provided. The sensor signal is from an amplifier 3 amplified, and its output signal is fed to an analog / digital converter 4 and after digitization takes place in a pulse processing stage 5, the part a microcontroller 6 forms. The microcontroller 6 also contains one Fuzzy controller 7.

In the pulse processing stage 5, the digitized sensor signals are first in the Data rate greatly reduced by storing them as "pulses". Such a pulse By definition, begins when the signal is positive or negative Direction far enough from the rest position, and it ends at the Return to the rest position. Each pulse is characterized by data such as Amplitude, duration, polarity, distance and the like are described and this data are saved.

The optics 2 contains a mirror system in a known manner, which is a variety of corresponds to optical bundling means and the IR radiation from a variety of fan-like radiation reception areas focused on sensor 1 (see for example GB-A-2 047 886 or EP-A-0 361 224). These radiation receiving areas are discrete zones, one passing through such a zone Object a positive sensor signal when entering and a negative sensor signal when leaving causes, both together give a characteristic signal. Such a thing The signal can be, for example, small positive, large positive, large negative through the pulses and described as small negative within a certain period of time. In the The pulses derived from the sensor signal are then evaluated examines whether they are of a type and characteristic of the intrusion of a person Have configuration, always a group of several successive Pulse is examined.

Practice has shown that within the time window considered, the digitized Sensor signal generally obtained a maximum of three, a maximum of four, such pulses so that it does not make sense to examine more than four pulses. The procedure in this investigation is such that the last four pulses are always used are stored and examined, the examination being carried out in the fuzzy controller 7 he follows

2, the fuzzy controller 7 contains a rule base 8 in a known manner, an inference engine 9, a process interface 10, and an action interface 11 whose output is monitored when an unwanted intruder is detected in the An alarm signal AS is available in the room. Regarding the fuzzy logic, the meantime extensive literature on this subject referred to, for example, the Book "Fuzzy Set Theory and its Applications" by H.-J. Carpenter, Kluwer Academic Publishers, 1991.

The rule base 8 contains a set of linguistic rules for the in a known manner Evaluation of the pulses. Based on these rules, an algorithm is constructed where the values are defined as so-called fuzzy sets, which are fuzzy sets are. Linguistic variables are words and expressions of everyday language or a natural language. These variables should be the natural language values Expressions (small, medium, large) can take, these expressions There are names for the fuzzy sets mentioned.

The rules of fuzzy logic, like classic logic, consist of one Condition or premise part and from a conclusion part. The condition part is in Fig. 2 by the process interface 10 and the conclusion part symbolized by the action interface 11. The inference machine 9 links the Direction of influence and the strength of the current states in the fuzzy sets due to of empirical technological knowledge.

3 shows a graphical representation with a typical fuzzy rule Basics of a fuzzy controller. The rule designated Rule 1 in the figure reads: "If A = LARGE and B = NORMAL, then X = SMALL". Rule 2 is: for example: "If B = NORMAL and C = SMALL, then X = NORMAL". A, B and C are input variables, X are output variables. The sentence beginning with "if" is the condition part, the part that starts with "then" is the conclusion part.

The fuzzy sets or fuzzy sets are the central concept of fuzzy logic the membership of elements in a fuzzy set by the so-called membership or membership function is defined. While with sharp quantities one One, which means belonging and a zero not belonging, are in the fuzzy sets as values for the membership function not just zero or one, but any Values in between allowed.

The conversion of sharp numbers to fuzzy amounts is called fuzzification designated. Each has an input variable, which is in practice, for example a sensor signal, at least one function depicted as a matrix. x-scaling this function has a numerical equivalent in the respective sensor signal, and the y-scale corresponds to the truthfulness or degree of approximation the corresponding statement and can take any value from 0 to 1. That degree the approximation is calculated by the membership function.

A suitable operator is used for the statements in the condition section searched for a size for the membership values; this size is the minimum value the membership function, then the operator is the minimum operator as in FIG. 3 and this in turn is the average of the two fuzzy sets for the input variables A and B. The result of the conclusion of the two rules 1 and 2 is the average of the fuzzy sets for A and B or B and C.

A sharp starting point is now calculated from these conclusions (Action interface 11, Fig. 2). If so as in Fig. 3 conclusions from several Rules exist, then the membership values for each Rules synthesized. And that happens, for example, by a comparison between the conclusion parts of the rules to the maximum value of the membership values to get the conclusion parts and a new membership function to create. This process is called the maximum operator; he puts the Union of the conclusion parts.

The unsharp one supplied by the inference machine 9 (FIG. 2) then becomes Result a sharp output size is calculated, for example, by The center of gravity of the synthetic membership function is calculated.

• Definition aller Eingangs- und Ausgangsvariablen: Die Eingangsvariablen sind im vorliegenden Fall die die aus dem Sensorsignal gewonnenen Pulse charakterisierenden Daten und ein Zeitfenster, die Ausgangsvariable ist ein Wert, der angibt, ob es sich um eine blosse Störung oder um ein unbefugtes Eindringen handelt.
• Definition der unscharfen Mengen (Fuzzy-Sets) für die linguistischen Variablen.
• Aufstellen der Regeln: Eine geeignete Regel ist beispielsweise die, dass unbefugtes Eindringen dann vorliegt, wenn die Bedingung einer Pulsreihe aus drei aufeinanderfolgenden Pulsen mit den Amplituden klein positiv, gross negativ und klein positiv im Zeitraum lang erfüllt ist.
• Festlegung der Inferenzmaschine: Es wird als Operator beispielsweise eine spezielle UND-Funktion, das sogenannte FUZZY-UND der Form F = y * min(A,B) + 0.5 * (1-y) * (A+B) gewählt, wobei A und B die Eingangsvariablen sind und y ein Gamma-Faktor. Für den Gamma-Faktor y=1 geht der FUZZY-UND-Operator in den Minimum-Operator (Fig. 3) über.
• Definition der Berechnung der scharfen Ausgangsgrössen: Diese auch als Defuzzyfizierung bezeichnete Operation, bei der aus einer unscharfen Menge über eine Ausgangszugehörigkeitsfunktion eine scharfe Grösse gewonnen wird, erfolgt vorzugsweise so wie in Fig. 3 durch Schwerpunktbildung.

The design of the fuzzy controller 7 (FIGS. 1, 2) is carried out roughly in the following steps:

• Definition of all input and output variables: In the present case, the input variables are the data characterizing the pulses obtained from the sensor signal and a time window, the output variable is a value that indicates whether it is a mere fault or an unauthorized entry.
• Definition of the fuzzy sets for the linguistic variables.
• Establishing the rules: A suitable rule is, for example, that unauthorized intrusion exists if the condition of a pulse series consisting of three successive pulses with the amplitudes small positive, large negative and small positive is met for a period of time.
• Definition of the inference machine: As an operator, for example, a special AND function, the so-called FUZZY-AND of the form F = y * min (A, B) + 0.5 * (1-y) * (A + B) where A and B are the input variables and y is a gamma factor. For the gamma factor y = 1, the FUZZY AND operator changes to the minimum operator (FIG. 3).
• Definition of the calculation of the sharp output variables: This operation, also known as defuzzification, in which a sharp variable is obtained from an unsharp quantity using an output membership function, is preferably carried out as in FIG. 3 by forming a center of gravity.

The signal processing described in IR detectors enables good and clean Separation between intrusion and interference signals with high detection performance. In particular, very noisy sensor signals and signals from the peripheral monitoring area can be clearly evaluated. By saving of the sensor signals in the form of pulses results in a strong reduction in the Memory requirements, especially those for the rules in linguistic form. In addition, the fuzzification and defuzzification are numerically relatively undemanding are and require little effort, even with a simple Microcontroller is realizable.

The fuzzy logic typical of fuzzy logic makes it unlikely that a signal discarded due to a narrowly missed condition becomes. The processing described corresponds rather to the very different and fuzzy Intrusion signals. The algorithm is based on the fuzzy formulation Simple and transparent core. As soon as it is written, it also applies to itself changing framework conditions, in which case only a few constants are changed must be (so-called parameterization). The constants are due to Trials and simulations optimized.

Claims (6)

1. Process for processing the signals of a passive infra-red detector, which generates electrical signals, described below as sensor signals, in relation to incident infra-red radiation, then evaluates these electrical signals, wherein the sensor signals are digitized and processed in the form of pulses, whose evaluation is carried out by means of fuzzy logic (7), wherein in each case the data of a series of several pulses are compared to rules stored in the form of linguistic variables, characterised in that the pulses are characterised by data, wherein as characterising data for describing the pulses, their amplitude and/or duration are used, and that for the conversion of the digitized sensor signals into pulses the waveform of the signals is examined, and at a certain excursion of the signal from its normal position a pulse start is set and on return to the normal position a pulse end is set.
2. Process according to Claim 1, characterised in that, as characterising data for describing the pulses, their polarity and/or their mutual spacing are used.
3. Process according to Claim 1 or 2, characterised in that, in each case the data of the last n successive pulses are compared with the stored rules, where the value of n lies between 2 and 4, and is preferably 3.
4. Infra-red detector for implementing the process according to Claim 1, having at least one sensor element for generating the sensor signals and having an evaluation circuit for processing and evaluating them, which contains a fuzzy controller (7) to which the pulse data are fed, characterised in that the fuzzy controller (7) contains a rule base (8) and an inference machine (9) and forms part of a microcontroller (6), which contains a pulse processing stage (5) for processing the digitized sensor signals in pulses and for storing the data describing the amplitude and/or duration of the pulses.
5. Infra-red detector according to Claim 4, characterised in that the rules stored in the rule base (8) of the fuzzy controller (7) are of the type that their condition part contains number and data of pulses and the time interval of the occurrence of the pulses.
6. Infra-red detector according to Claim 4 or 5, characterised in that the inference machine (9) of the fuzzy controller has a FUZZY-AND function as the operator.

Images