EP1884904A1 - Danger type determination by means of at least two signals - Google Patents

Abstract

The method involves evaluating two signals using an evaluation unit, where the signals are received from a detection unit. A receiving unit and a transmitting unit are provided for communicating with other units. The evaluation unit determines a type of danger according to a rule from the signals based on a decision tree. A difference of the two signals is determined as the rule, where the difference is compared with a predetermined value, which is a real number.

Description

The invention relates to a method for detecting a hazard and for determining the type of danger with a danger detector, at least comprising a detection unit and an evaluation unit.

Hazard detectors, such as optical scattered light smoke detectors, optical extinction smoke detectors, optoacoustic smoke detectors, etc., which are not wired, but via the air interface with another unit, such as a security control panel, a fire alarm, a hazard alarm, etc., exchange messages may show only a very small electrical power consumption, so on the one hand the electrochemical voltage sources such as batteries, accumulators etc. are spared as possible and thus a long life of the hazard alarm is guaranteed until the next revision and on the other hand as little as the costly electrochemical voltage sources until the next revision needed. The processing unit with the microprocessor of the danger detector, which evaluates the signals of the smoke detector, may therefore consume only as little as possible electrical power.
In fire detectors of the latest generation, at least two signals for distinguishing between fire and non-fire are distinguished, whereby these at least two signals are processed with special algorithms to a final result. However, these algorithms used until now in fire detectors are either very extensive or use power-intensive operations such as one or more quotient formations and therefore lead to an excessively large power consumption and thus burden the electrochemical Tension cells strong. In addition, the processing speed of the signals is often very slow in today's algorithms. This leads to an extended response time of the fire detector and thus to a delayed detection of fire. The signal processing with such algorithms with quotient formation is for example from the writings JP 2005115970 . AU 2004201100 A1 . EP 1 022 700 B1 . DE 199 02 319 A1 . DE 42 31 088 A1 . WO 01/59737 A1 and US 2004/0066512 A1 known. In the literature, the signal processing for fire detectors, for example, from A. Riemer, H. Politze, T. Krippendorf, "Realization of a wide-range optical detector using different wavelengths and multiple scattering angles", AUBE 04, Proceedings 13th International Conference on Automatic Fire Detection, 14.-16. September 2004, Duisburg, Germany, page 74 ff , and T. Fujisawa, T. Suzuki, Y. Yoshikawa, S. Ohkuma, "Optical Smoke Detector Using Dual Light Spectrum", AUBE 04, Proceedings 13th International Conference on Automatic Fire Detection, 14.-16. September 2004, Duisburg, Germany, page 527 ff , discussed. In all of these documents, the ratio of scattered light signals generated by two emitters of different wavelengths is formed. A non-fire is characterized by the fact that this ratio is less than or equal to two. For a ratio greater than two, for example, a fire is assumed as the origin, and accordingly an alarm is issued. At the documents AU 2004201100 A1 . US 2004/0066512 and A. Riemer, H. Politze, T. Krippendorf, "Realization of a wide-range optical detector using different wavelengths and multiple scattering angles", AUBE 04, Proceedings 13th International Conference on Automatic Fire Detection, 14.-16. September 2004, Duisburg, Germany, page 74 ff , In each case one emitter in the NIR range (infrared) and one in the wavelength range of the blue light is used. The signals of the forward and the backward scattering of both wavelengths are evaluated by the ratios the signals from the forward and backward scattering of the first wavelength and the second wavelength are formed. In addition, the ratio between the signals of the infrared and the blue light from the forward scattering and the same from the backward scattering is respectively formed. All ratios are then appropriately compared so that it can be determined whether there is fire or non-fire.

From the scriptures EP 1 022 700 B1 and DE 199 02 319 A1 It is known that a relationship between forward and backward scatter of a fire detector is formed. The obtained ratio is subtracted with a constant value and the result is divided by a number. The values of the scattered signals can also be multiplied in advance with environmental factors.

The document DE 42 31 088 A1 discloses scattered light signals measured from two equal or different angles, the degree of polarization of the first measurement being different by 90 ° from the polarization degree of the second measurement. From the two signals for 0 ° or 90 ° polarization, either a quotient is formed directly or first the sum and the difference and then the quotient of difference and sum are formed.

The document WO 01/59737 A1 describes an optical scattered-light-suction device with at least two light sources, which may also be designed as a fire alarm. The at least two light sources in this device either emit two different wavelengths or emit light with different polarizations. However, the scattered light signals can also be recorded at different angles. By forming the quotients from the respective signal pairs, it is possible to deduce the type of particles present in the device Among other things, whether the particles in the device come from a fire or from a non-fire.

The publication EP 1630758 A1 describes algorithms for evaluating the scattered light signals from two different scattering angles, which use two multipliers and the difference and the sum of the signals of forward and backward scattering to infer the type of smoke present. Extensive comparison operations, subtractions and additions are used to determine the type of smoke.

The object of the present invention is to be seen in that a simple and efficient evaluation method for determining the type of smoke and fire is proposed.

The object is achieved in each case by the subject matters of the independent claims. Further developments of the invention are specified in the subclaims.

A core of the invention is to be seen in that for the detection of a hazard and for determining the type of danger with a hazard detector, at least comprising a detection unit and an evaluation unit, the evaluation unit of the hazard detector evaluates at least two signals received by the detection unit, according to at least one rule the at least two signals on the basis of a decision tree, the danger type is determined. The type of danger considered is a type of smoke, the presence of fire, the type of gas, the type of movement or species of a living being, the type of fire, etc. By using a decision tree according to the invention, complex and extensive arithmetic operation can be avoided. First, the difference of at least two signals is formed as an at least first rule.

This difference is then used to determine the type of danger using the decision tree by comparing this difference with a previously defined value as at least second rule. As at least one third rule, this difference is used to subtract once more the signal used in the first subtraction rule. Now the second rule is carried out again and then the third rule follows again, etc. Of course, the signals for determining the type of danger are already normalized before the evaluation, ie. H. the long-term drift, the zero point signal etc. have already been taken into account. The decision tree checks whether a result provides a true statement by comparison with a previously defined value. If this is not the case, a further iteration step is initiated, i. H. from the previous difference, the signal first used for the subtraction is again subtracted and again checked with a previously defined rule. If this comparison yields a true statement, no further iteration step is initiated. Otherwise, a new iteration step is initiated. This happens until the comparison yields a true statement or the maximum number of previously defined iteration steps has been reached. The previously defined value used for the comparison is a real number. This previously defined value is a parameter that can be adapted to the environmental conditions of the hazard alarm.

In order to increase or reduce the accuracy of the determination of danger, in the first rule a signal is multiplied by an accuracy factor, a real number. The number of iteration steps in the decision tree is influenced by this accuracy factor, a real number. The value determined by the processing unit, calculated by multiplying the accuracy factor by one of the signals and subtracting the other signal, becomes then compared with the second rule with the previously defined value, a real number. Now, as already mentioned above, the third rule is applied, etc.

According to the invention, the danger detector is a hazard detector which communicates via the air interface with a danger warning center or with other units of a mobile communication network. The danger detector is powered by electrochemical voltage sources. As a hazard detector any danger detector can be used, in which at least two signals are evaluated. Such hazard detectors may be fire detectors, optical scattered light smoke detectors, optical extinction smoke detectors, optoacoustic smoke detectors, gas detectors, intrusion detectors, flame detectors, etc.

A great advantage of the inventive method is that, for example, a clear distinction between fire and non-fire is possible. The results of the decision tree are invariant or nearly invariant to the dustiness of the hazard alarm, to the long-term attenuation of the sensitivity, to the efficiency of the light sources used in the hazard detector or to the photoreceivers or to microphones and generally to proportionally varying signal magnitudes during the run of the event (for example, a fire, a gas alarm, etc.). For example, with only one iteration step, the smoke type or a present fire is determined only from a multiplication, a subtraction and a comparison.

A further advantage is that several decision trees can be combined with each other and thus even more differentiated statements for the event or the danger are possible.

With one or more decision trees can thus be decided energy-saving, very safe and very fast, for example, if there is a fire or not. This greatly reduces the likelihood of a false alarm and the hazard alarms can thus trigger an alarm more quickly and have a significantly lower power consumption than previous hazard alarms.

Figur 1

einen erfindungsgemässen Entscheidungsbaum zur Bestimmung der Rauchart,

Figur 2

einen erfindungsgemässen Entscheidungsbaum mit variierbarer Genauigkeit zur Bestimmung der Rauchart,

Figur 3

einen weiteren erfindungsgemässen Entscheidungsbaum mit variierbarer Genauigkeit zur Bestimmung der Rauchart,

Figur 4

einen erfindungsgemässen Entscheidungsbaum mit variierbarer Genauigkeit zum Bestimmen ob Feuer vorliegt,

Figur 5

einen minimalen Entscheidungsbaum mit variierbarer Genauigkeit zur Bestimmung der Rauchart und

Figur 6

zwei miteinander verknüpfte Entscheidungsbäume.

The invention will be explained in more detail with reference to an embodiment shown in a figure. Show

FIG. 1

a decision tree according to the invention for determining the type of smoking,

FIG. 2

an inventive decision tree with variable accuracy for determining the type of smoking,

FIG. 3

another decision tree according to the invention with variable precision for determining the type of smoke,

FIG. 4

an inventive decision tree with variable accuracy for determining whether there is fire,

FIG. 5

a minimal decision tree with variable accuracy for determining the type of smoke and

FIG. 6

two linked decision trees.

FIG. 1 shows a decision tree according to the invention for determining the type of smoke using the example of a scattered-light detector. The difference c of the two scattered light signals S1 and S2 is determined, which is examined by further differences in a decision tree and, as a result, provides the type of smoke available. The signals S1 and S2 have been normalized and freed, for example, from the long-term drift, the zero-point signal, etc. This is basically carried out prior to application of the method according to the invention. is If c is less than or equal to a previously defined value p, then it is the smoke type 1, otherwise a second iteration step is initiated. For this purpose, the further difference between c and the signal S2 is formed. In Figs. 1-6, these subtractions are listed in the form of computer instructions. Where c is a variable and S ₁ , S _{2 are} constants. The computer command c: = c - S ₂ now states that first the difference between c and S ₂ should be formed and that this new difference should then be assigned to the variable c as a new value. If the new result c is less than or equal to p, then it is the smoke type 2. If the result is greater than p, then the process is continued until the result obtained c gives a true statement. P as previously defined value is a real number and can also be set equal to 0.

The scattered light signals change continuously during a fire, with the changes in the two scattered light signals, which were measured, for example, under different scattering angles, generally being proportional to one another in both signals. Also, the changes of scattered light signals generated at different wavelengths change proportionally to each other during a fire. It has been shown that the results of the decision tree are invariant with respect to such proportional changes when p = 0 is chosen and approximately invariant when p is close to 0.

The method according to the invention can preferably be used for signals of scattered light optical smoke detectors, optical extinction smoke detectors, flame detectors or optoacoustic smoke detectors. However, the method according to the invention could also generally be applied to devices, for example hazard alarms, intrusion detectors, gas detectors, etc. in which at least two signals must be evaluated.

FIG. 2 shows a decision tree according to the invention with variable accuracy for determining the type of smoking. If the distinction of the type of smoking is to be less accurate or more accurate, the signal S1 can still be multiplied by an accuracy factor, for example k1 = 0.01; .100, in the first difference. With k1 <1 the accuracy becomes smaller, with k1> 1 it gets bigger. The comparison of the results c takes place as in FIG. 1.

FIG. 3 shows a further decision tree according to the invention with variable precision for determining the type of smoke, in which case the scattered light signals S ₃ and S ₄ are evaluated, the accuracy factor k 2 is used and the value q used for the comparison is q. If more than two signals are to be compared with one another, then the method according to the invention is used analogously with more than two signals. As already explained in FIGS. 1 and 2, in a first step the difference d is compared with a previously defined value q and in further steps of the decision tree the result d is compared with this value.

FIG. 4 shows a decision tree according to the invention with variable accuracy for determining whether there is fire. With the factors k1, k2, the number of iterations in the decision tree can be varied and thus also the fine classification of the types of smoke or the accuracy of the decision tree can be determined. The factors k1, k2 are real numbers. If the types of smoke occurring at a certain location of the smoke detector, which can be from fire and from non-fire, are known, the decision tree can be simplified by a suitable choice of k1, k2, that only a distinction is made between fire and non-fire. By varying the parameters p and q used for the comparison, the passage through the decision tree can also be influenced and the alarming properties of the fire detector can thus be optimally adapted to the ambient conditions. This adaptation of p, q can be done in the fire detector after its installation in a building.

FIG. 5 shows a minimal decision tree with variable accuracy for determining the type of smoking. Very often, the types of smoke of fire and non-fire differ so that, for example, in non-fire small signals S1 are detected and in real fires, for example, generally more large signals S1 are present. The signals S2 are almost equal in both situations. For such cases, a minimal decision tree can be established by the appropriate choice of k1, k2. In this minimal case, the decision tree consists only of a multiplication, a subtraction and a comparison. The figure b) shows the individual operations.

Figure 6 shows two linked decision trees. Two or even more decision trees can be linked together and thus result in a differentiated evaluation of an event or a fire. Since the results of the individual decision trees are invariant (for p = q = 0) or approximately invariant (for p, q close to 0) in terms of proportional changes, the combination of the results of the linked decision trees is also invariant with respect to such changes.

FIG. 7 shows a fire detector BM according to the invention with a receiving unit EE and a transmitting unit SE for communicating with further units GMZ, such as, for example, a danger alarm center. another fire detector, etc. The fire detector has at least one detection unit DE and an evaluation unit AWE for carrying out the method according to FIGS. 1 to 6.

Claims (18)

Method for detecting a hazard and for determining the type of danger with a hazard detector (BM), at least comprising a detection unit (DE) and an evaluation unit (AWE), characterized in that the evaluation unit (AWE) at least two of the detection unit (DE) received signals (S1, S2, S3, S4) evaluates by determining the type of danger based on a decision tree according to at least one rule from the at least two signals (S1, S2, S3, S4). A method according to claim 1, characterized in that as at least one rule, a difference (c) from the at least two signals (S1, S2, S3, S4) is determined. Method according to claims 1 and 2, characterized in that the further difference between the difference (c) and a received signal (S1, S2, S3, S4) is determined as at least one rule. Method according to claims 1 to 3, characterized in that the difference (c) is compared with a predetermined value (p, q) as at least one rule Method according to one of the preceding claims, characterized in that the predetermined value (p, q) is a real number and is used as a parameter for adapting the alarming properties of the hazard alarm to the environment Method according to one of the preceding claims, characterized in that the predetermined value (p) is equal to zero Method according to claims 2 to 4, characterized in that the at least one rule is applied sequentially until the comparison of the at least two signals and / or the difference gives a true statement. Method according to one of the preceding claims, characterized in that a type of smoke, the presence of fire, the type of gas, the type of movement of a living being, the species of a living being and / or the type of fire are considered as the type of danger. Method according to one of the preceding claims, characterized in that as at least one rule for the accuracy of the determination of the type of danger, a received signal (S1, S3) is multiplied by an accuracy factor (k1, k2). A method according to claim 9, characterized in that a real number is used as the accuracy factor (k1, k2). Method according to one of the preceding claims, characterized in that, as at least one rule, the number of iterations to be used for the determination of the type of danger in the decision tree is determined by a factor (k1, k2). Method according to one of the preceding claims, characterized in that as at least one rule for determining the type of danger by a multiplication of the accuracy factor (k1, k2) with a signal (S1, S3) and the subtraction of the at least two signals (S1, S2, S3, S4) is compared with a previously defined numerical value. Method according to one of the preceding claims, characterized in that the signals (S1, S2, S3, S4) are scattered light signals of a scattered light detector. Method according to one of the preceding claims, characterized in that the hazard detector (BM) communicates via the air interface with a further unit (GMZ). Method according to one of the preceding claims, characterized in that the hazard detector (BM) is powered by electrochemical voltage sources with electricity. Method according to one of the preceding claims, characterized in that a fire detector, an optical scattered-light smoke detector, an optical extinction smoke detector, an opto-acoustic smoke detector, a gas detector, an intrusion detector and / or flame detector are used as danger detector (BM). Hazard detector (BM) for detecting a hazard and for determining the type of danger, at least comprising a detection unit (DE) and an evaluation unit (AWE), with a receiving (EE) and a transmitting unit (SE) for communicating with at least one further unit (GMZ), with the evaluation unit (AWE) for evaluating at least two signals (S1, S2, S3, S4) received by the detection unit (DE) and for determining the type of danger according to at least one rule from the at least two signals (S1, S2, S3, S4) based on a decision tree. Fire detector according to claim 17, characterized in that a fire detector, an optical scattered light smoke detector, an optical extinction smoke detector, an optoacoustic smoke detector, a gas detector, an intrusion detector and / or flame detector are provided as a hazard detector.

Images