JPH06301870A - Fire state recognizing system - Google Patents

Abstract

PURPOSE:To provide a fire state recognizing system capable of precisely discriminating real fire from non-real fire by using fuzzy theory. CONSTITUTION:This system is constituted in such a way that a smoke sensor of scattered light type, a carbon monoxide sensor, or a thermal sensor is employed as the fire sensor, and an output signal from each sensor is processed and discriminated by a fuzzy expert system consisting of a membership function employed as fuzzy expression, an inference engine, and a knowledge source, and the inference engine finds probability confidence for all the real fire and non-real fire predicted according to a fixed rule based on the membership function in accordance with the output signal of each sensor, and the real fire can be discriminated from the non-real fire by comparing the probability confidence with the criterion of the knowledge source.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fire property grasping system for discriminating between real fire and non-real fire by using fuzzy theory.

[0002]

2. Description of the Related Art Some self-fire alarm systems use a scattered light type smoke detector as a fire detector, but the scattered light type smoke detector does not depend on the occurrence of fire such as steam or cigarette smoke. Occasionally, a false alarm or non-fire alarm was issued when an object was detected.

[0003]

However, since it is difficult for a fire receiver to distinguish between a real fire and a non-real fire even if such a non-fire warning occurs, it is necessary to manage the fire receiver every time a fire warning is issued. People had to go to the site one by one and check. Therefore, the recent self-fire alarm system is required to be able to reliably distinguish between non-fire and actual fire, and non-fire alarms are relatively common especially in hotel buildings with bathrooms and beds. Therefore, countermeasures are strongly desired.

Therefore, the present applicants previously filed Japanese Patent Application No. 3-8.
No. 49581 proposes a fire property grasping system which uses a fuzzy theory to discriminate a real fire event from a non-fire event to reduce a non-fire alarm, but the present invention further extends this prior application. The present invention is to provide a fire property grasping system, which has been reached as a result of repeated studies and intensive studies by the present inventors.

[0005]

SUMMARY OF THE INVENTION Therefore, in order to achieve the above-mentioned object, the proposed invention provides an output signal from a plurality of fire detectors for detecting a fire event by a plurality of different physical means. The present invention relates to an improvement of a fire property grasping system that discriminates between an actual fire occurrence event and a non-fire occurrence event by performing inference processing by fuzzy theory. Output signals from multiple sensors are adopted as a fuzzy expression. It is designed to be processed by a fuzzy expert system consisting of a ship function, an inference engine, and a knowledge source, so that the inference engine can calculate all real fires based on the membership function from the output signal of the sensor. , Probability confidence for non-actual fire event is calculated, and these probabilities are compared with the actual fire and non-actual fire according to the criterion of knowledge source. It is adapted to determine.

According to the second aspect of the present invention, as the fire sensor, a composite sensor including a scattered light type smoke sensor, a carbon monoxide sensor, and a heat sensor is used. Further, the invention according to claim 3 uses the case of smoke combustion of paper waste and the state of smoldering of cloth as the fire occurrence event, and uses the cases of smoking state and steam generation state as non-fire occurrence events. The membership function is defined in accordance with the present invention, and the invention according to claim 4 further adds the output signal from the pyroelectric element infrared detector as an element of discrimination to further increase the certainty factor in consideration of the presence of a person. Highly discriminating. Further, the invention according to claim 5 uses the fuzzy expert system to determine the probability of fire (fire certainty) and the duration determined for a fire event as the determining factors, and based on a defuzzification criterion, a fire forecast. It can be distinguished from the fire information.

[0007]

According to the present invention, based on the value of the membership function for each output signal from the fire detector that detects a fire-occurrence event by different physical means, all predicted values are calculated according to a certain rule. The probability of occurrence of a real fire event or a non-actual fire event is determined, and the certainty of these occurrence probabilities is inferred based on the judgment criteria of the knowledge source. Therefore, the sensor output corresponds to a non-fire event. I mean,
It is possible to accurately judge whether or not it corresponds to a real fire event.

Therefore, it is possible to reduce the occurrence of non-fire alarms, and as a result, it is possible to construct a highly reliable self-fire alarm system.

[0009]

1 shows a conceptual configuration of a fuzzy expert system used in a system according to the present invention, in which an inference engine receives output signals of sensing units 1A to 1C from a composite sensor 1. As shown in FIG. , From the membership function adopted as a fuzzy expression, the probability of confidence (fuzzy value) for all actual fire and non-actual fire events according to certain predicted rules is calculated, and this fuzzy value is used as the criterion of the knowledge source. In light of this, it is possible to infer and judge the actual fire or non-fire.

FIG. 2 shows a schematic configuration of the entire system. In this embodiment, the fire detector includes a scattered light smoke detector 1A, a heat detector 1B, and a carbon monoxide detector 1C. Is used, and the composite sensor 1 is installed in the center X of the ceiling of the room R of the hotel shown in FIG. 3, for example (claim 2). In such a system of the present invention, the output of the composite sensor 1 is sampled, for example, every one second, A / D-converted by the A / D converter 2, and then the fire reception having the inference and judgment function by the fuzzy theory is received. It is sent to the machine 3.

That is, here, the sensing units 1A to 1 of the composite sensor 1 are made by the fuzzy expert system.
If the output signal of C is due to a real fire event or a non-fire event, and if it is due to a real fire factor, a fire alarm is issued. Whether the forecast should be made in light of the defuzzification criteria, using the certainty factor and the duration of the state as the criteria
You may make it the structure which identifies whether it should be an actual fire report (Claim 5).

By the way, in the system of the present embodiment, the installation location of the composite sensor 1 is assumed to be the room R of the hotel, and all the events of actual fire occurrence predicted based on this assumption,
Output of each of the sensing units 1A to 1C of the composite sensor 1 (here, the composite sensor 1 at the center X of the ceiling is targeted) by setting a non-fire occurrence event and experimentally generating each state. A signal is obtained, and the signal output signal defines a membership function for each event.

That is, as the event of the actual fire, the state of flaming and burning of paper waste, the state of smoldering of bed sheets and blankets, and the event of the non-fire, the state of smoking, the bathroom r In order to define the function of the membership for these events, for example, aluminum steam on a carpeted indoor floor of scraps of about 20 sheets of A4 size paper is set. On the bed installed in the room R, cotton bed sheets (8 sheets of 10 cm square are stacked) and an acrylic blanket (1
3 pieces of 0 cm square are stacked by electric heating
A smoldering state was created by smoldering, and smoking was performed immediately below the composite sensor 1, and smoke was intentionally sprayed on the composite sensor 1.

In addition, 45 with the bathroom shielded
Hot water of about ° C was jetted from the shower head to collect steam, and the door dr of the bathroom r was opened from the start of measurement to create a state in which steam was generated in the room R. Each sensing unit 1A to 1C of the composite sensor 1 obtained by such an experiment
The membership function was defined so that the output signal from would not be inconsistent with the rules of the expert system.

The system of the present invention prepares membership functions for all real fire events and non-real fire events according to a certain rule predicted for the output signal of each sensor, and outputs the output of each sensor. Obtain the probability of confidence for the signal,
The actual fire and non-actual fire are discriminated based on these probability of certainty against the judgment criteria stored in the knowledge source. Next, a fire event adopted in the system of the present invention will be described.

The system of the present invention employs a discrimination method which does not consider the presence of the human body and a fuzzy discrimination method which is based on the presence of the human body. In the former case, the scattered light type smoke sensor and the monoxide are used as the fire sensor. Carbon sensor, heat sensor is adopted,
In the latter case, a scattered light smoke sensor, a carbon monoxide sensor, and a heat sensor are adopted as a fire sensor, and a pyroelectric element infrared sensor is added.

20 to 23 show the output characteristics of the scattered light smoke detector, carbon monoxide detector, heat detector and pyroelectric element infrared detector for each event adopted in the system of the present invention.
Shown in. Here, FIG. 20 shows the output characteristic for each event of flaming fire, FIG. 21 for smoking the cotton, FIG. 22 for smoking by cigarette, and FIG. 23 for the event of steam.

Basic idea of fuzzy expert system 1) When the existence of human body is not taken into consideration Although it is attempted to grasp and judge the nature of the phenomenon at an early stage, it is basically to work on the safe side, so in a suspicious situation. We judge that it is a real fire and make an alarm. Fire rules have priority over non-fire rules, and if any one of smoke, heat, and carbon monoxide has an abnormal value, or if the abnormal value continues, it is judged as an actual fire and an alarm is issued. Let

Non-fire is judged by smoke, heat and carbon monoxide.
It is performed after all the elements return to normal values, but according to the rule, it is performed after returning to the forecast level or lower. In addition, since non-fire phenomena such as smoking and steam may develop into fire,
The display of the degree of certainty that there is a high possibility of non-fire is treated as an intermediate process. How to proceed with inference Each time the output signal of each sensor is sampled, the probability of probability of a fire occurrence event or non-fire occurrence event changes,
The actual fire and non-actual fire are judged according to the situation.

When a fire assertion rule (a rule consisting of a fire certainty factor and its duration) to be described later is established, it is judged as an actual fire and an alarm is issued, but after all three elements have returned to the forecast level or lower, non-fire occurs. When the assertion rule is established, it is determined as non-fire and the initial state is restored. 2) When based on the presence of the human body A pyroelectric element infrared sensor (PIR sensor) is used to detect the presence of the human body, but a scattered light smoke sensor, a carbon monoxide sensor, and a heat sensor are used. Adopt a fire detector as a fire detector.

In the pyroelectric element infrared sensor, the fact that the magnitude of the output value differs depending on the value detected by the pyroelectric element is added to the determination factor, so that the certainty factor is further improved. The pyroelectric element infrared sensor has an output value of f0 when nothing is detected, an output value of f1 when detecting a human body, and an output value of f2 regardless of the presence of a person when detecting a flame. Become. This means that three kinds of output values can be obtained by providing an appropriate threshold value.

The output value of the pyroelectric element infrared sensor is
Since it is f2 in the case of a flaming fire and f1 in the case of a cigarette, this difference is added to the determination method. Rule of each event adopted in the present invention 1) Method of discriminating each event when the existence of human body is not taken into consideration Each event of flaming fire, smoldering, smoking, and steam has the following rule characteristics. Therefore, the fuzzy expert system captures this feature and obtains a fuzzy value for each event. [Rules for flaming fire] Rules (1) and (2) exist although they are distinguished only by heat and carbon monoxide or heat.

Rule (1): If the temperature rises significantly,
Confidence also increases. In this event only temperature rise is used. The membership function is shown in FIG. Rule (2): When the temperature and the concentration of carbon monoxide are abnormally increased, the certainty factor is also increased. In this event, temperature increase and CO concentration change are used. The membership function is shown in FIG.

Since it consists of two elements, a fuzzy value for each element is obtained, and the product of these is used as the probability probability of the event. [Kunfire fire rule] Not only when both smoke and carbon monoxide concentration react, but also when only smoke or only CO (when showing a gradual upward trend or when an abnormal value continues) Judge and make an alarm. There are the following four rules.

Rule (1): The temperature does not rise so much, but the confidence increases when the smoke concentration and the CO concentration rise abnormally. This event uses smoke concentration, CO concentration, and temperature rise. The membership function is shown in FIG. Since it consists of three elements, fuzzy values for each element are obtained, and the product of these is taken as the probability of certainty of the event.

Rule (2): Confidence increases when CO concentration increases significantly. In this event, only CO concentration changes are used. The membership function is shown in FIG. Rule (3): When the smoke density shows a gradual increase tendency, the certainty factor also increases. This event uses smoke density changes and the average smoke density over the last minute. The membership function is shown in FIG.

Rule (4): The temperature does not rise so much, and the confidence increases when the smoke density continues to be an abnormal value. This event uses smoke concentration changes, temperature rises, and durations of 10% or more of smoke concentration. Each membership function is shown in FIG. [Smoking rules] The aim is to capture cases in which the sensor responds when the smoker naturally smokes. In this situation, the smoke density often increases rapidly, so it is determined that it is a cigarette together with CO. However, if the cigarette is intentionally blown excessively, it cannot always be determined.

Further, when the abnormal value of smoke continues, it is regarded as an abnormal state and the certainty factor of the cigarette is lowered. The rules in this case are as follows. Rule (1): When the temperature does not rise so much, the CO concentration rises to some extent, and the smoke concentration tends to rise sharply, the confidence level rises. However, if the smoke density continues to be an abnormal value, it is considered as an abnormal state, and the certainty factor of the cigarette is reduced.
In this event, the maximum value of the product of the fuzzy value of the smoke concentration change and the fuzzy value for the average concentration change of the smoke concentration for the past 1 minute, the fuzzy value for the maximum CO concentration, the fuzzy value for the temperature rise, and 10% of the smoke concentration Find the product of the fuzzy values for the above durations. Each membership function is shown in FIG. [Rule of steam] Similar to the case of smoking, it is judged as water vapor by grasping the feature that the smoke density rapidly rises. In the case of water vapor, it is required that CO does not react.

Further, if the abnormal value of smoke continues, it is regarded as an abnormal state, and the certainty factor of water vapor decreases. This rule may hold for other non-fire events such as dust. Rule (1): When the temperature does not rise so much, the CO concentration hardly rises, and the smoke concentration tends to rise rapidly, the confidence level increases. However, if the smoke density continues to be an abnormal value, it is regarded as an abnormal state and the certainty factor decreases. In this event, fuzzy value of smoke density change and smoke density past 1
The maximum value of the product of fuzzy values with respect to the change in the average concentration for a minute,
The product of the fuzzy value for the maximum CO concentration, the fuzzy value for the temperature rise, and the fuzzy value for the duration of smoke concentration of 10% or more is obtained. Each membership function is shown in FIG. 2) Distinguishing each event based on the presence of the human body [Rules for flaming fire] Distinguish only by heat and CO or heat, but the output of the pyroelectric element infrared sensor (PIR sensor) detects flame. Is added to the condition. There are two rules:

Rule (1): The output of the infrared detector of the pyroelectric element becomes f2, and when the temperature rises significantly, the certainty factor also rises. In this event only temperature rise is used. The membership function is shown in FIG. Rule (2): The output of the infrared detector of the pyroelectric element becomes f2, and when the temperature and the CO concentration increase significantly, the confidence level also increases.

In this event, temperature increase and CO concentration change are used. The membership function is shown in FIG. The product of the fuzzy values for the two elements is obtained, and the output f2 of the PIR sensor is added to make the determination. [Kunyaki fire rule] Not only when both smoke and carbon monoxide concentration react, but also when only smoke or only CO (when it shows a gradual rising tendency or when an abnormal value continues) Judge and make an alarm. There are four rules:

Rule (1): The temperature does not rise so much, but the confidence increases when the smoke concentration and the CO concentration rise abnormally. This event uses smoke concentration, CO concentration, and temperature rise. The membership function is shown in FIG. Since it consists of three elements, fuzzy values for each element are obtained, and the product of these is taken as the probability of certainty of the event.

Rule (2): Confidence increases when CO concentration increases significantly. In this event only CO concentration is used. The membership function is shown in FIG. Rule (3): When the smoke density shows a gradual increase tendency, the certainty factor also increases. This event uses smoke density changes and the average smoke density over the last minute. The membership function is shown in FIG.

Rule (4): The temperature does not rise so much, and the confidence increases when the smoke density continues to be an abnormal value. This event uses smoke concentration changes, temperature rises, and durations of 10% or more of smoke concentration. Each membership function is shown in FIG. [Smoking rules] The aim is to capture cases in which the sensor responds when the smoker naturally smokes. In this situation, the smoke density often rises rapidly, so it is determined to be tobacco together with CO.

When the cigarette is intentionally blown excessively, it cannot always be discriminated. In addition, it is supposed that an actor must always be present in smoking, and the output of the pyroelectric element infrared detector should be human body detection. Further, when the abnormal value of smoke continues, it is regarded as an abnormal state, and the certainty factor decreases. There are the following rules.

Rule (1): The output of the pyroelectric element infrared sensor is f1. When the temperature does not rise so much, the CO concentration rises to some extent, and the smoke concentration tends to rise sharply, the confidence increases. However, if the smoke density continues to be an abnormal value, it is considered as an abnormal state, and the certainty factor of the cigarette is reduced. In this event, the maximum value of the product of the fuzzy value of the smoke concentration change and the fuzzy value for the average concentration change of the smoke concentration for the past 1 minute, the fuzzy value for the maximum CO concentration, the fuzzy value for the temperature rise, and 10% of the smoke concentration The product of the fuzzy value with respect to the above duration is calculated, and the output of the PIR sensor is added to the judgment. That is, the output of the PIR sensor is f1
If so, the certainty of smoking increases because there is a person. Each membership function is shown in FIG. [Rule of steam] Similar to the case of smoking, it is judged as water vapor by grasping the feature that the smoke density rapidly rises. In the case of water vapor, it is required that CO does not react.

Further, if the abnormal value of smoke continues, it is regarded as an abnormal state, and the certainty factor of water vapor decreases. This rule may hold for other non-fire events such as dust. Rule (1): When the temperature does not rise so much, the CO concentration hardly rises, and the smoke concentration tends to rise rapidly, the confidence level increases. However, if the smoke density continues to be an abnormal value, it is regarded as an abnormal state and the certainty factor decreases.

In this event, the maximum value of the product of the fuzzy value of the smoke density change and the fuzzy value for the average density change of the smoke density in the past one minute, the fuzzy value for the maximum CO density value,
The product of the fuzzy value for the temperature rise and the fuzzy value for the duration of 10% or more of the smoke density is obtained. Each membership function is shown in FIG. Next, a method of discriminating between fire and non-fire in the system of the present invention will be described.

In the present invention, the scattered light smoke detector, carbon monoxide detector, and heat detector are divided into the cases described above based on the presence of a person and cases where the presence of a person is not taken into consideration. From the output signal from the sensor, the probability of probability for each of the above-mentioned fire and non-fire events is obtained, and the forecast, fire notification, or non-fire discrimination is performed according to the following criteria. In this case, regarding the fire certainty factor, [fire certainty factor] = each certainty factor of the flaming fire rules (1), (2) and the smoldering fire rules (1), (2), (3), (4) The maximum value of. Using the fire certainty factor thus obtained and the duration of the fire as a judgment factor, a forecast or a fire report is issued in accordance with the following defuzzification judgment criteria for fire assertion. Forecast Fire confidence: 45% or more and duration: 6 seconds (or) Fire confidence: 40% or more and duration: 9 seconds (or) Fire confidence: 35% or more and duration: 12 seconds Fire report Fire confidence Degree: 90% or more and duration: 6 seconds (or) Fire confidence: 80% or more and duration: 9 seconds (or) Fire confidence: 70% or more and duration: 12 seconds [Non-fire assertion] Smoke , After returning to below the forecast level for both heat and CO, it shall be determined as non-fire according to the following criteria for non-fuzzy non-fuzzy judgment criteria. If smoking maximum confidence value> 70%, smoking If vapor maximum confidence value> 70%, water vapor If a pyroelectric infrared sensor is added, the temperature difference is within 10 ° C for the human body The original output is usually about 2V, but in the case of a flame, the temperature difference is 50 ° C. or more, and the original output is 10V or more. Therefore, the human body and the flame can be distinguished by setting the threshold value to 10V. The presence or absence of a human body can be identified by setting the threshold value to 0.4V. That is, if the original output <0.4V, the output of the pyroelectric element infrared sensor is f0, and normally 0.4V <Original output <10V, the output of the pyroelectric element infrared sensor is f1, and the human body detection If 10 V <original output, the output of the pyroelectric element infrared sensor = f2, and flame detection is performed.

[0040]

According to the first aspect of the present invention, the outputs from a plurality of fire detectors for detecting a fire occurrence event by different physical means are used to detect an actual fire occurrence event and a non-actual fire occurrence event by fuzzy logic. Since it distinguishes, it is possible to accurately discriminate between a real fire and a non-real fire, so that it is possible to reduce non-fire alarms and build a highly reliable self-fire alarm system.

According to the second aspect of the present invention, since a composite sensor including a scattered light type smoke sensing section, a carbon monoxide sensing section, and a heat sensing section is used, a plurality of sensing areas can be installed in one sensor. A vessel can be installed. Further, the invention according to claim 3 uses the flaming combustion of paper waste and the state of smoldering cloth as the fire occurrence event, and uses the smoking state and the steam generation state as the non-fire occurrence events. It is possible to determine with 100% the probability of flaming combustion, which is a fire occurrence event that occurs in a hotel room, and steam, which is a non-fire occurrence event. In addition, smoking and smoldering can be determined with high probability, and there is an effect that it is possible to construct a self-fired alarm system suitable for indoor monitoring such as in a hotel.

Furthermore, in the invention according to claim 4, since the presence of the human body can be identified by adding the output signal from the pyroelectric element infrared sensor to the determination element, it becomes possible to more accurately determine smoking. The accuracy of judgment is further improved. In addition, the invention according to claim 5 can identify the forecast by using the fire certainty factor and the continuation time obtained by the fuzzy expert system as the judgment factors, so that more accurate correspondence can be achieved.

[Brief description of drawings]

FIG. 1 is a conceptual configuration diagram of a fuzzy expert system used in an embodiment of the present invention.

FIG. 2 is an overall configuration diagram of the system of the present invention.

FIG. 3 is an explanatory diagram of an installation place of the fire detector of the present invention.

FIG. 4 is an explanatory diagram of a membership function (rule (1)) in a flaming fire event.

FIG. 5 is an explanatory diagram of a membership function (rule (2)) in a flaming fire event.

FIG. 6 is an explanatory diagram of a membership function (rule (1)) in a smoldering event.

FIG. 7 is an explanatory diagram of a membership function (rule (2)) in a smoldering event.

FIG. 8 is an explanatory diagram of a membership function (rule (3)) in a smoldering event.

FIG. 9 is an explanatory diagram of a membership function (rule (4)) in a smoldering event.

FIG. 10 is an explanatory diagram of a membership function in a smoking event.

FIG. 11 is an explanatory diagram of a membership function function in a steam occurrence event.

FIG. 12 is an explanatory diagram of a membership function (rule (1)) in a flaming fire event when a PIR sensor is used.

FIG. 13 is an explanatory diagram of a membership function (rule (2)) in a flaming fire event when a PIR sensor is used.

FIG. 14 is an explanatory diagram of a membership function (rule (1)) in a smoldering event when a PIR sensor is used.

FIG. 15 is an explanatory diagram of a membership function (rule (2)) in a smoldering event when a PIR sensor is used.

FIG. 16 is an explanatory diagram of a membership function (rule (3)) in a smoldering event when a PIR sensor is used.

FIG. 17 is an explanatory diagram of a membership function (rule (4)) in a smoldering event when a PIR sensor is used.

FIG. 18 is an explanatory diagram of a membership function in a smoking event when a PIR sensor is used.

FIG. 19 is an explanatory diagram of a membership function function in a steam generation event when a PIR sensor is used.

FIG. 20 is an output characteristic diagram of a scattering smoke detector, a carbon monoxide detector, a heat detector, and a pyroelectric element detector in a flaming fire event.

FIG. 21 is an output characteristic diagram of a scattering type smoke sensor, a carbon monoxide sensor, a heat sensor, and a pyroelectric element sensor in a smoldering event.

FIG. 22 is an output characteristic diagram of a scattered smoke detector, a carbon monoxide detector, a heat detector, and a pyroelectric element detector in a smoking event.

FIG. 23 is an output characteristic diagram of a scattering smoke detector, a carbon monoxide detector, a heat detector, and a pyroelectric element detector in a steam generation event.

[Explanation of symbols]

 1 Combined fire detector 1A Scattered light smoke detector 1B Carbon monoxide detector 1C Heat detector 2 A / D converter 3 Fire receiver

Front Page Continuation (72) Inventor Naofumi Hosokawa 3-14-1, Nakahara, Mitaka City, Tokyo Inside the Fire Defense Research Institute, Ministry of Home Affairs (72) Inventor Hiroyuki Tamura 3-14-1, Nakahara, Mitaka City, Tokyo Fire Department, Ministry of Home Affairs In the laboratory (72) Takashi Kurio 1048, Kadoma, Kadoma, Osaka Prefecture Matsushita Electric Works, Ltd. (72) Inventor Shinji Nakanishi, 1048, Kadoma, Kadoma, Osaka Prefecture (72) Inventor, Shigenori Kusanagi Osaka Prefecture Kadoma City 1048 Kadoma Matsushita Electric Works Co., Ltd. (72) Inventor Toru Fujioka 1048 Kadoma Osaka Kadoma City Matsushita Electric Works Co., Ltd. (72) Inventor Yoshifumi Watabe 1048 Kadoma Kadoma City, Osaka Matsushita Electric Works Incorporated (72) Inventor Shinji Kiribata 1048, Kadoma, Kadoma, Osaka Prefecture Matsushita Electric Works, Ltd.

Claims (5)

[Claims] 1. A method for inferring an output signal from a plurality of fire detectors for detecting a fire occurrence event by a plurality of different physical means by fuzzy theory to distinguish an actual fire occurrence event from a non-fire occurrence event. In the fire property monitoring system, the scattered light smoke detector, carbon monoxide detector, and heat detector were adopted as fire detectors, and the output signals from these detectors were adopted as fuzzy expressions. A fuzzy, composed of a ship function, an inference engine, and a knowledge source.
It is configured to judge processing by an expert system,
For that purpose, the inference engine obtains the probability certainty for all real fire and non-real fire events predicted according to a certain rule based on the membership function corresponding to the output signal of each of the above sensors, and these probability confidences A fire property comprehension system that distinguishes between actual fires and non-actual fires based on the judgment criteria of the knowledge source. 2. The fire detector is a composite detector including a scattered light smoke detector, a carbon monoxide detector, and a heat detector. Fire property monitoring system. 3. As the above-mentioned fire occurrence event, the membership function of each is defined by using the case of the smoke combustion of paper waste and the smoldering state of the cloth, and the non-fire occurrence event is the smoking state and steam. The fire property grasping system according to claim 1, wherein each membership function is defined by using an example of an occurrence state. 4. An output signal from a pyroelectric element infrared sensor is further added, and by a fuzzy expert system,
The fire property ascertaining system according to claim 1, wherein the actual fire and the non-actual fire are discriminated by obtaining probability probabilities of all the actual fire occurrence events and the non-fire occurrence events. 5. The fire property ascertaining system according to claim 1 or 4, wherein the fuzzy expert system uses the probability certainty factor and the continuation time obtained for the actual fire event as the determinative criteria for defuzzification. Based on this, a fire property grasping system that distinguishes between fire forecasts and fire reports.