Anomaly detection program and fire monitoring system equipped therewith

Abstract

To provide an abnormality identification program for early identification of the occurrence of an abnormality that has a low probability of occurring under normal conditions, and a fire monitoring system equipped with the program.SOLUTION: An abnormality identification program for a fire monitoring system includes processes for: collecting a plurality of physical quantities x measured by a plurality of sensors over a predetermined time period and converting them into a normalized physical quantity y; calculating the sum Sprev of the physical quantity y multiplied by a weight w, where the weight w corresponding to the plurality of sensors are set in advance; collecting a plurality of physical quantity x measured by a plurality of sensors after a predetermined time elapses, converting them into a normalized physical quantity y, and calculating the sum Snew of the physical quantity y multiplied by the weight w; calculating the mean value and standard deviation σnew of the sum Snew; and determining whether the probability calculated by applying the mean value of the sum Snew, the standard deviation σnew, and the mean value of the sum Sprev to the cumulative distribution function is less than or equal to a threshold value, and determining that an abnormality has occurred if the probability is less than or equal to the threshold value.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an anomaly detection program that determines whether or not an anomaly has occurred based on physical quantities measured by multiple sensors, and a fire monitoring system equipped therewith. [Background technology]

[0002] For example, fire detectors, which are widely used in ordinary homes, measure physical quantities such as room temperature and smoke density, and notify of a fire if the physical quantity measured exceeds a threshold. On the other hand, large-scale facilities such as office buildings, tenant buildings, and logistics warehouses use fire monitoring systems that combine multiple analog detectors and receivers. Multiple analog detectors are installed in monitoring areas throughout the large-scale facility. Each of the multiple analog detectors outputs an analog signal to the receiver according to the measured value. The receiver determines whether a fire has occurred or not, or whether any other abnormality has occurred, based on the analog signals received from each of the multiple analog detectors. For example, the receiver can issue a warning if a physical quantity measured is below the threshold for determining whether a fire has occurred. The threshold for determining whether or not to issue a warning can be set to a different value for each analog detector or for each monitoring area.

[0003] In addition, a conventional type of fire detector known as a combined spot-type detector is available. A combined spot-type detector can measure two or more physical quantities, such as temperature and smoke concentration, and will alert to the occurrence of a fire if any one of these physical quantities exceeds a threshold.

[0004] Furthermore, Japanese Patent Publication No. 2018-88630 (Patent Document 1) and Japanese Patent Publication No. 2018-101416 (Patent Document 2) propose a fire monitoring system that generates a large number of images of a monitored area captured by a surveillance camera, and uses these images to perform deep learning on a multilayer neural network to improve the accuracy of fire detection. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2018-88630 [Patent Document 2] Japanese Patent Publication No. 2018-101416 [Overview of the project] [Problems that the invention aims to solve]

[0006] A fire monitoring system receiver can issue a warning if a physical quantity below a threshold is measured to determine whether or not a fire has occurred. However, the receiver can only compare the physical quantity measured by each of the multiple analog detectors to the threshold. The receiver cannot issue a warning by considering the interrelationships of the multiple analog detectors, such as the relationship between different physical quantities like temperature and smoke concentration, the relative positions of the multiple analog detectors, the relationships between the measurement results of the multiple analog detectors, or the effects of disturbances on each of the multiple analog detectors.

[0007] Conventional combined spot-type fire detectors alert to a fire when any one of two or more physical quantities, such as temperature and smoke concentration, exceeds a threshold. In other words, combined spot-type detectors determine whether a fire has occurred by comparing each of the two or more physical quantities to a threshold. Therefore, combined spot-type detectors cannot determine whether a fire has occurred by combining the two or more physical quantities.

[0008] The fire monitoring systems described in Japanese Patent Publication No. 2018-88630 and Japanese Patent Publication No. 2018-101416 both require deep learning of a multilayer neural network based on a large number of images. Furthermore, image-based deep learning cannot weight the data to account for the interrelationships between multiple analog detectors as described above. Moreover, the computational load for fire detection using a multilayer neural network is extremely high.

[0009] The present invention has been made in view of the above-mentioned problems, and aims to provide an anomaly detection program and a fire monitoring system equipped therewith that can detect an anomaly that has a low probability of occurring under normal conditions at an early stage by comprehensively integrating the measurement results of multiple sensors and considering the interrelationships of multiple sensors. [Means for solving the problem]

[0010] (A) To achieve the above objective, the present invention provides an anomaly detection program that determines whether an anomaly has occurred based on signals output from multiple sensors for measuring a physical quantity x that changes due to the occurrence of a fire, and includes a first process that, under normal circumstances where no fire has occurred, collects multiple physical quantities x measured by multiple sensors over a predetermined time or predetermined number of times, and converts each of the collected physical quantities x into a normalized physical quantity y, and a weight w corresponding to each of the multiple sensors is set in advance, and the sum S of the physical quantities y multiplied by the weight w. prev A second process to calculate the following: After the predetermined time has elapsed or after the predetermined number of times has been exceeded, multiple physical quantities x measured by multiple sensors are collected, each of the collected physical quantities x is converted into a normalized physical quantity y, and the sum S of the physical quantities y multiplied by the weight w is calculated. new A third process to calculate the sum S new average value [Table 1] and standard deviation σ new The fourth process involves calculating the average value. [Table 2] , standard deviation σ new and sum S prev average value [Table 3] The processor is instructed to perform a fifth process, which includes determining whether the probability calculated by applying the cumulative distribution function is below a predetermined probability threshold, and if it is below the threshold, determining that an anomaly has occurred.

[0011] (B) Preferably, in the abnormality detection program described in (A) above, each of the physical quantities x collected in the first process and the third process is converted into a physical quantity y normalized by the following formula (1).

number

[0012] (C) Preferably, in the abnormality detection program of (A) or (B) above, the sum S in the second process and the third process prev and sum S new This is calculated by one of the following formulas (2) to (4).

number

number

number

[0013] (D) Preferably, in the abnormality determination program according to any one of (A) to (C) above, the probability in the fifth process is calculated based on the value of t calculated by the following formula (5). The abnormality determination program according to any one of claims 1 to 3. [Number] However, n is the average value [Table 5] of the total sum S used for the calculation of new and is the number of data.

[0014] (E) Preferably, in the abnormality determination program according to any one of (A) to (D) above, the value of the weight w used in the second process and the third process is set and changed based on the influence of the disturbance exerted on at least one of the plurality of sensors.

[0015] (F) Preferably, in the abnormality determination program according to (E) above, when at least one of the plurality of sensors is affected by the temperature of the air conditioner, based on the information regarding the operation of the air conditioner, the value of the weight w applied to at least one of the plurality of sensors is set and changed.

[0016] (G) Preferably, in the abnormality detection program described in (E) above, if at least one of the multiple sensors is affected by temperature due to sunlight, the value of the weight w applied to at least one of the multiple sensors is changed based on at least one of the time of sunlight, solar radiation, and temperature.

[0017] (H) Preferably, in any of the abnormality detection programs (A) to (G) above, in the first process, random numbers are generated to increase the significant figures of the physical quantity x, and after adding the random numbers to each of the multiple physical quantities x measured by the multiple sensors, a normalized physical quantity y is calculated.

[0018] (I) To achieve the above objective, the fire monitoring system of the present invention is a fire monitoring system that operates according to any of the abnormality determination programs of (A) to (H) above, and comprises a plurality of sensors and a receiver, the receiver comprises a receiving unit, a control unit and a notification unit and has the abnormality determination program installed, the plurality of sensors are installed in a monitoring area and measure a physical quantity x that changes due to the occurrence of a fire and output a signal, the receiving unit receives the signals output from the plurality of sensors, the processor of the control unit executes the first to fifth processes according to the abnormality determination program based on the plurality of physical quantities x measured by the plurality of sensors, and the notification unit notifies that an abnormality has occurred if it is determined in the fifth process that an abnormality has occurred.

[0019] (J) Preferably, in the fire monitoring system described in (I) above, the abnormality determination program includes a sixth process which determines whether a physical quantity x measured by at least one of the multiple sensors is equal to or greater than a preset threshold of physical quantity x, and determines that a fire has occurred if it is equal to or greater than the threshold, the processor of the control unit executes the sixth process according to the abnormality determination program based on each of the multiple physical quantities x measured by the multiple sensors, and the notification unit notifies that a fire has occurred if it is determined in the sixth process that a fire has occurred.

[0020] (K) To achieve the above objective, the fire monitoring system of the present invention is a fire monitoring system that operates according to any of the abnormality determination programs of (A) to (H) above, comprising a plurality of sensors, at least one repeater, and a receiver, wherein the repeater comprises a receiving unit and a control unit and has the abnormality determination program installed, the plurality of sensors are installed in a monitoring area and measure a physical quantity x that changes due to the occurrence of a fire and output a signal, the receiving unit receives the signals output from the plurality of sensors, the processor of the control unit executes the first to fifth processes according to the abnormality determination program based on the plurality of physical quantities x measured by the plurality of sensors, and if it is determined in the fifth process that an abnormality has occurred, it transmits a signal to another repeater or the receiver.

[0021] (L) Preferably, the fire monitoring system of (K) above includes a sixth process which determines whether a physical quantity x measured by at least one of the multiple sensors is greater than or equal to a preset threshold of physical quantity x, and determines that a fire has occurred if it is greater than or equal to the threshold, and the processor of the control unit executes the sixth process according to the abnormality determination program based on each of the multiple physical quantities x measured by the multiple sensors, and if the sixth process determines that a fire has occurred, it transmits a signal to another repeater or the receiver.

[0022] (M) To achieve the above objective, the present invention provides an anomaly detection program that determines whether an anomaly has occurred based on signals output from a plurality of sensors for measuring a physical quantity x, and includes a first process of collecting a plurality of physical quantities x measured by the plurality of sensors over a predetermined time or a predetermined number of times, and converting each of the collected physical quantities x into a normalized physical quantity y, and the sum S of the physical quantities y prev A second process to calculate the following: After the predetermined time has elapsed or after the predetermined number of times has been exceeded, multiple physical quantities x measured by multiple sensors are collected, each of the collected physical quantities x is converted into a normalized physical quantity y, and the sum S of the physical quantities y is calculated. newA third process to calculate the sum S new average value [Table 6] and standard deviation σ new The fourth process involves calculating the average value. [Table 7] , standard deviation σ new and sum S prev average value [Table 8] The processor is instructed to perform a fifth process, which includes determining whether the probability calculated by applying the cumulative distribution function is below a predetermined probability threshold, and if it is below the threshold, determining that an anomaly has occurred.

[0023] (N) Preferably, in the abnormality detection program of (M) above, a weight w corresponding to each of the multiple sensors is set in advance, and in the second process, the sum S of the physical quantities y multiplied by the weight w is prev The sum S of the physical quantities y multiplied by the weight w is calculated, and in the third process, the sum S of the physical quantities y multiplied by the weight w is calculated. new This is calculated. [Effects of the Invention]

[0024] According to the anomaly detection program and fire monitoring system equipped therewith of the present invention, by comprehensively integrating the measurement results of multiple sensors and considering the interrelationships between multiple sensors, it becomes possible to early detect when an anomaly occurs that has a low probability of occurring under normal circumstances. [Brief explanation of the drawing]

[0025] [Figure 1] Figure 1 is a schematic diagram showing a first embodiment of a fire monitoring system equipped with the abnormality detection program of the present invention. [Figure 2]Figure 2 is a block diagram showing the receiver that constitutes the fire monitoring system described above. [Figure 3] Figure 3 is an explanatory diagram illustrating the relationship between physical quantity x measured by multiple sensors and the normalized physical quantity y. [Figure 4] Figure 4 is an explanatory diagram for explaining the sum S of the physical quantities y. [Figure 5] Figure 5 is an explanatory diagram illustrating the principle of anomaly detection in an anomaly detection program. [Figure 6] Figure 6 shows the monitoring area conditions set for simulating the anomaly detection program. Figure 6(a) is a side view of the ceiling, sensors 1-4, and fire source, and Figure 6(b) is a front view of the ceiling and sensors 1-4. [Figure 7] Figure 7(a) is a graph showing the temperature at the location of sensor 1 calculated in the above simulation, and Figure 7(b) is a graph showing the temperature at the location of sensor 2 calculated in the above simulation. [Figure 8] Figure 7(a) is a graph showing the temperature at the location of sensor 3 calculated in the above simulation, and Figure 7(b) is a graph showing the temperature at the location of sensor 4 calculated in the above simulation. [Figure 9] Figure 9 is a graph showing the values ​​of t calculated in the above simulation and the probability values ​​of the t-distribution calculated from those values. [Figure 10] Figure 10 is a schematic diagram showing a second embodiment of a fire monitoring system equipped with the abnormality detection program of the present invention. [Modes for carrying out the invention]

[0026] Hereinafter, embodiments of the abnormality detection program and the fire monitoring system equipped therewith of the present invention will be described with reference to the drawings.

[0027] The anomaly detection program of the present invention determines whether or not an anomaly has occurred based on the probability of a measured physical quantity x occurring. Therefore, the target of detection by the anomaly detection program of the present invention is not limited to anomalies caused by fire. However, for the sake of explanation, anomalies caused by fire will be used as an example of the target of detection, and embodiments of the anomaly detection program and fire monitoring system will be described.

[0028] 1. First Embodiment of a Fire Monitoring System Figure 1 shows a fire monitoring system according to a first embodiment of the present invention. The fire monitoring system of this embodiment is installed in large-scale facilities such as office buildings, tenant buildings, and logistics warehouses. The fire monitoring system mainly consists of a plurality of sensors 1 to 4 and a receiver 10. Sensors 1 to 4 are installed, for example, on the ceiling of a predetermined monitoring area in a large-scale facility. Although not shown, numerous sensors with the same or different configurations as sensors 1 to 4 are installed in various monitoring areas of the large-scale facility. Signals output from sensors 1 to 4 and other sensors are input to the receiver 10 via signal lines 5, which are shown as dashed lines in Figure 1. The receiver 10 monitors the entire large-scale facility based on the signals output from sensors 1 to 4 and other sensors. The receiver 10 is capable of monitoring the occurrence of fires and other troubles. Other troubles include, for example, gas-related abnormalities, electrical-related abnormalities, abnormalities in fire prevention equipment installed in the large-scale facility, and abnormalities in the communication status with peripheral equipment that works in conjunction with the receiver 10.

[0029] 1.2 Multiple sensors The multiple sensors 1 to 4 are, for example, analog fire detectors. Sensors 1 to 4 have the same configuration as each other and are equipped with at least one detection element for measuring temperature and / or smoke concentration (physical quantity x). In a predetermined monitoring area, each of the multiple sensors 1 to 4 measures the physical quantity x that changes due to the occurrence of a fire and generates an analog signal corresponding to the physical quantity x. Furthermore, each of the multiple sensors 1 to 4 converts the analog signal corresponding to the physical quantity x into a digital signal and outputs it. This digital signal includes the address assigned to each of the multiple sensors 1 to 4. The digital signals output from each of the multiple sensors 1 to 4 are input to the receiver 10 via the signal line 5.

[0030] Note that the multiple sensors 1-4 are not limited to multiple fire detectors. For example, the multiple sensors 1-4 may be differential distributed type detectors. Differential distributed type detectors mainly consist of multiple thermocouples electrically connected at predetermined intervals along a single wire. Each of the multiple thermocouples generates an electromotive force corresponding to the amount of heat detected.

[0031] Furthermore, the multiple sensors 1 to 4 may be temperature measuring devices utilizing optical fibers. This temperature measuring device mainly consists of a single optical fiber, a laser light source, and a photodetector. The optical fiber is laid, for example, on the ceiling of a predetermined monitoring area in a large-scale facility. The laser light source emits pulsed laser light into the incident end of the optical fiber. A portion of the incident light is reflected by the glass particles of the optical fiber, generating reflected light of Rayleigh scattering, Brillouin scattering, and Raman scattering. These reflected lights return to the incident end. Of these components of reflected light, the intensity of the Raman scattering and the frequency of the Brillouin scattering change depending on the temperature of the optical fiber. Therefore, it is possible to measure the temperature based on the intensity of the Raman scattering and the frequency of the Brillouin scattering. In addition, it is possible to determine the position on the optical fiber where the reflected light was generated based on the time from when the incident light was incident until the reflected light returned.

[0032] 1.3 Receiver The receiver 10 shown in Figure 1 is installed, for example, in a disaster prevention center located in a large-scale facility. The receiver 10 mainly consists of a receiving unit 11, a control unit 12, and a notification unit 13, as shown in Figure 2. In Figure 2, the receiving unit 11 is electrically connected to a plurality of sensors 1 to 4 via the signal line 5 in Figure 1. The receiving unit 11 receives digital signals output from the plurality of sensors 1 to 4. These digital signals include physical quantities x measured by each of the plurality of sensors 1 to 4. The receiving unit 11 transmits the received digital signals to the control unit 12. For the sake of explanation, the digital signals processed by the control unit 12 are referred to as physical quantities x measured by the plurality of sensors 1 to 4.

[0033] The control unit 12 includes a processor for executing arithmetic processing. The processor of the control unit 12 executes the processes of steps S1 to S10, S11 to S14, and S21 to S26 shown in Figure 2, according to the abnormality detection program of this embodiment. In addition to detecting the occurrence of a fire, similar to conventional fire monitoring systems, the abnormality detection program is capable of detecting slight abnormalities in the initial stages of a fire. The abnormality detection program of this embodiment will be described in detail later.

[0034] The notification unit 13 is composed of, for example, a large touch panel as shown in Figure 1, LEDs and a speaker (not shown), etc. Based on signals output from the control unit 11, the notification unit 13 displays various information on the touch panel. For example, if the control unit 12 detects the occurrence of a fire or an abnormality, the notification unit 13 notifies the occurrence of the fire or abnormality by displaying an image on the touch panel, changing the color of the LED light, and outputting a sound from the speaker.

[0035] 1.4 Anomaly Detection Program The processing of the control unit 12 according to the abnormality detection program of this embodiment will now be described. In Figure 2, steps S1 to S10 are the processing of the main routine of the abnormality detection program. Steps S11 to S14 and S21 to S26 are the processing of subroutines associated with the main routine.

[0036] 1.4.1 Fire detection process The control unit 12 executes the fire detection process in step S1 based on the physical quantity x measured by the multiple sensors 1 to 4. In the fire detection process S1, the control unit 12 compares the physical quantity x with a first threshold value set in step S11. For example, if the physical quantity x is temperature, a temperature of "70°C" is set as the first threshold value. In this case, the control unit 12 determines whether the physical quantity x is 70°C or higher. If the result of this determination is that it is 70°C or higher (YES), the control unit 12 sends a signal to the notification unit 13 to notify it of the occurrence of a fire. On the other hand, if it is determined that it is not 70°C or higher (NO), the control unit 12 executes the physical quantity x collection process in step S2.

[0037] In the abnormality detection program of this embodiment, the fire detection process S1 is executed at the earliest stage, so that even if a fire occurs, the notification unit 13 will quickly notify the fire. As a result, the disaster prevention center where the receiver 10 is installed can quickly carry out initial fire suppression.

[0038] 1.4.2 Physical Quantity x Collection Process In the physical quantity x collection process of step S2, the control unit 12 continuously collects the physical quantity x measured by the multiple sensors 1 to 4. The control unit 12 collects the physical quantity x for each of the multiple sensors 1 to 4 at predetermined time intervals. The time interval for collecting the physical quantity x is determined according to the number of multiple sensors 1 to 4, the length of the signal line 5, the processing power of the processor of the control unit 12, and so on.

[0039] In step S21, the collected physical quantity x is stored chronologically in a database corresponding to each of the multiple sensors 1 to 4. Each time a new physical quantity x is measured, it is added to the database. When the database's storage area is full of physical quantities x, the oldest physical quantity x is deleted and the latest physical quantity x is added. In this way, while the anomaly detection program of this embodiment is running, the physical quantities x stored in the database are constantly updated.

[0040] Here, we will explain the significance of the physical quantity x stored in the database in step S21. First, the physical quantity x stored in the database is one that was determined to be below the first threshold in the fire determination process S1 described above. In other words, all physical quantities x stored in the database were measured under conditions in which the occurrence of a fire had not been reported. Next, the physical quantity x stored in the database may have been measured under either normal or abnormal conditions. Normal conditions refer to a situation in which no event causing a fire has occurred. Abnormal conditions refer to a situation in which an event causing a fire has occurred, but the occurrence of a fire has not yet been reported. Finally, the collection of time-series physical quantities x measured under normal conditions serves as a basis for determining whether or not an abnormality occurred afterward.

[0041] 1.4.3 Random number addition process In the random number addition process of step S3, the control unit 12 generates random numbers to increase the significant figures of the physical quantity x and adds these random numbers to each of the physical quantities x collected in step S2. The random number addition process S3 is performed selectively according to the performance of the multiple sensors 1 to 4. That is, the random number addition process S3 is a process to improve the accuracy of the probability calculation process in step S9, which will be described later, and is performed when the minimum number of digits of the physical quantity x is insufficient. In the random number addition process S3, the control unit 12 generates a random number with n decimal places that does not substantially affect the physical quantity x measured by the multiple sensors 1 to 4, and adds this random number to the physical quantity x. In step S21 in Figure 2, the physical quantity x to which the random number has been added is stored chronologically in a database corresponding to each of the multiple sensors 1 to 4. On the other hand, if the physical quantity x measured by the multiple sensors 1 to 4 has n decimal places of significant figures, the control unit 12 proceeds to the physical quantity y conversion process in step S4 without executing the random number addition process S3.

[0042] 1.4.4 Physical quantity y conversion process In the physical quantity y conversion process of step S4, the control unit 12 converts each of the physical quantities x collected in step S2, or each of the physical quantities x to which random numbers have been added in step S3, into a normalized physical quantity y using the following formula (1).

number

[0043] Here, the normalization process in the conversion of physical quantity y will be explained with reference to Figure 3. As shown in Figure 3, the physical quantity x measured by the multiple sensors 1 to 4 are all continuous variables and follow a normal distribution N(μ,σ) with mean μ and standard deviation σ. However, each of the multiple sensors 1 to 4 is installed under different conditions in a predetermined monitoring area shown in Figure 1. For example, each of the multiple sensors 1 to 4 is installed at a different position on the ceiling, and the distance from the air outlet of the air conditioner 100 is also different from each other. Therefore, the mean μ and standard deviation σ of the physical quantity x measured by the multiple sensors 1 to 4 will be different values. Thus, the physical quantity x measured by the multiple sensors 1 to 4 is converted into a normalized physical quantity y using the above equation (1). As shown in Figure 3, the normalized physical quantity y follows a standard normal distribution N(0,1) with mean 0 and standard deviation 1. In other words, the normal distribution N(μ,σ) of the physical quantity x measured by multiple sensors 1 to 4 is aligned to the standard normal distribution N(0,1) of the normalized physical quantity y. The collection of normalized physical quantities y forms a population that follows a normal distribution N(0,1). In step S22 of Figure 2, the normalized physical quantities y are stored in time series in databases corresponding to each of the multiple sensors 1 to 4.

[0044] Note that the average value in formula (1) above [Table 10] The physical quantity x stored in the database in step S21 is used to calculate the average value. [Table 11] The number of data points for the physical quantity x used in its calculation is not particularly limited. For example, the mean [Table 12] At the time of calculating, all physical quantities x stored in the database in step S21 may be used. Alternatively, for example, the physical quantities x stored in the database in step S21 for the most recent tens of minutes or tens of seconds may be used as the average value. [Table 13] It may be used in the calculation of [the value].

[0045] 1.4.5 Sum S prev Calculation process Sum of Step S5 prev In the calculation process, the control unit 12 obtains physical quantities y from the database corresponding to each of the multiple sensors 1 to 4 in step S22, and calculates the sum S of the physical quantities y multiplied by the weight w. prev This is calculated using the following formula (2).

number

[0046] The weight w is set in advance in step S12 for each of the multiple sensors 1 to 4, based on the interrelationships between sensors 1 to 4 (including the relationships between identical sensors). For example, Table 1 below shows the weights w set based on the distance r between the multiple sensors 1 to 4. [Table 14]

[0047] The weight w shown in Table 1 is set to "1" when the distance r from sensor P to sensor Q is related to the height H (m) from the floor to the ceiling and r ≤ 0.18H, and to "r" in all other cases. -3 / 2 This is what it means.

[0048] The sum S of equation (2) above prev For each of sensorP, the physical quantity y of sensorQ is... (Q) Weight lol (Q,P) This is the sum of all the values ​​obtained by multiplying by . Figures 4(a) to (d) show the physical quantity y in equation (2) above when sensors P and Q are sensors 1 to 4, respectively. (Q) and weight w (Q,P) This shows the physical quantities y of sensors 1 to 4 in (a) to (d) in Figure 4. (Q) Each of them has weight lol (Q,P) The sum of all the values ​​obtained by multiplying by the given value is the sum S in equation (2) above. prev That is the case.

[0049] Note that the total sum S prev The formula used to calculate may be either formula (3) or (4) below.

number

number

[0050] Sum S prev The physical quantity y used in the calculation (p) and physical quantity y (Q)The number of data points is not particularly limited. For example, the sum S prev At the time of calculation, all physical quantities y stored in the database in step S22 (p) and physical quantity y (Q) You may also use the physical quantity y stored in the database in step S22. (p) and physical quantity y (Q) Of these, the physical quantity y in the most recent tens of minutes or tens of seconds (p) and physical quantity y (Q) The sum S prev It may be used in the calculation of [the value].

[0051] Sum S prev In calculation process S5, the sum is repeatedly calculated until a predetermined number of times is reached. prev The result is calculated. Repeated sum S prev The time interval over which this is calculated is not particularly limited. For example, the shortest time interval could be the latest physical quantity y (p) and physical quantity y (Q) Each time that is stored in the database in step S22, the sum S prev The following may be calculated. If it is longer than the shortest time interval, the sum S prev The time interval at which this is calculated can be set arbitrarily.

[0052] Sum S prev The sum S is calculated repeatedly by the calculation process S5. prev This is stored in the database in step S23 of Figure 2. The total sum S stored in the database prev Based on this, the sum S prev average value [Table 15] The average value is calculated. [Table 16] This is used in step S9, described later, to determine anomalies in the physical quantity x newly measured by multiple sensors 1 to 4.

[0053] 1.4.6 Sum S new Calculation process The control unit 12 calculates the total S prev After the calculation process S5 is completed, the sum S in step S6 is calculated. new Proceed to calculation process. Sum S new In calculation process S6, the control unit 12 collects multiple physical quantities x measured by multiple sensors 1 to 4, similar to steps S2 to S5 described above, converts each of the collected physical quantities x into a normalized physical quantity y, and calculates the sum S of the physical quantities y multiplied by the weight w. new The sum S is calculated. Equation (1) above is used to convert the measured physical quantity x to the normalized physical quantity y. new One of the above formulas (2) to (4) is used to calculate S. new In calculation process S6, the sum is repeatedly calculated until the predetermined number of times is reached. new The sum S calculated is obtained by repeatedly calculating the sum S. new This is stored in the database in step S24.

[0054] Sum S new The physical quantity y used in the calculation (p) and physical quantity y (Q) The number of data points is not particularly limited, but the sum S mentioned above is prev In calculation process S5, the sum S prev The physical quantity y used in the calculation (p) and physical quantity y (Q) It is preferable to make the number of data points the same as or approximately the same as the number of data points.

[0055] Also, the sum S new The number of times and time intervals in which the calculation is repeated are not particularly limited, but the sum S mentioned above prev In calculation process S5, the sum S prev It is preferable that the number of times the calculation is repeated and the time intervals are the same as or approximate to this.

[0056] 1.4.7 Average [Table 17] Calculation process Average value of Step S7 [Table 18] In the calculation process, the control unit 12 calculates the sum S new average value [Table 19] Calculate the average value. [Table 20] The sum S calculated in calculation process S7 new average value [Table 21] This is stored in the database in step S25.

[0057] 1.4.8 Standard deviation σ new Calculation process Standard deviation σ of step S8 new In the calculation process, the control unit 12 calculates the sum S new Standard deviation σ new The standard deviation σ is calculated. new The standard deviation σ calculated in calculation process S8 new This is stored in the database in step S26.

[0058] 1.4.9 Probability Calculation Process In the probability calculation process of step S9, the control unit 12 calculates the average value [Table 22] , standard deviation σ new and sum S prev average value [Table 23] Apply it to the cumulative distribution function. As a variable used in the cumulative distribution function, for example, the following formula (5) is used. The control unit 12 calculates the probability value of the t-distribution based on the value of t calculated by the following formula (5).

Number

Table 24

[0059] 1.4.10 Abnormality discrimination processing In the abnormality discrimination processing of step S10, the control unit 12 compares the probability value calculated in the probability calculation processing of step S9 with the second threshold value preset in step S13. As the second threshold value, for example, a low probability value within the range of 1.0×10 -4 ~1.0×10 -8 is set.

[0060] The control unit 12 determines whether the calculated probability value is less than or equal to the second threshold value. If it is determined that the calculated probability value is less than or equal to the second threshold value (YES), the control unit 12 sends a signal to the notification unit 13 to notify the occurrence of an abnormality. Then, the control unit 12 acquires the updated physical quantity y from the database in step S22 and repeats the processing of steps S5 to 10. On the other hand, if it is determined that the calculated probability value is not less than or equal to the second threshold value (NO), the control unit 12 acquires the updated physical quantity y from the database in step S22 without sending a signal to the notification unit 13 and repeats the processing of steps S5 to S10.

[0061] 1.4.11 Principle of abnormality discrimination The principle of anomaly detection in the anomaly detection program described above will be explained with reference to Figure 5. The mean of a sample population sampled from a population with a normal distribution N(μ,σ) with mean μ and standard deviation σ is [Table 25] , the standard deviation is σ samp In this case, the value of t calculated by equation (6) below follows the t-distribution shown in Figure 5.

number

[0062] Therefore, in the abnormality detection program of this embodiment, in steps S1 to S5 of Figure 2, the sum of the population physical quantities y S is determined based on the physical quantity x under normal conditions. prev The following is calculated. Then, in step S6 of Figure 2, the newly measured physical quantity x is converted to a normalized physical quantity y, and the sum S of these physical quantities y is calculated. new Calculate the sum S new is the sum S prev Although this is not a sample population sampled from, it is summed up S prev Considering this as a sample population, t is calculated in step S9 of Figure 2, and the probability value of the t-distribution is derived from the value of t. In step S10, if the probability value is below the second threshold, it is determined that an event with a low probability of occurring under normal circumstances, i.e., an anomaly, has occurred.

[0063] 1.4.12 Changing the setting of weight w The value of the weight w that is set in step S12 in Figure 2 may be changed based on the influence of disturbances on at least one of the multiple sensors 1 to 4.

[0064] For example, when at least one of the plurality of sensors 1 to 4 is affected by the temperature of the air conditioner 100 shown in FIG. 1, based on the information regarding the operation of the air conditioner 100, the value of the weight w applied to at least one of the plurality of sensors 1 to 4 is set and changed. The information regarding the operation of the air conditioner 100 includes, for example, the ON / OFF information of the air conditioner 100, the set temperature information of the air conditioner 100, and the like. When these information are provided to the control unit 12, the control unit 12 executes the setting change process of step S14. For example, when the sensor 3 shown in FIG. 1 is affected by the temperature of the air conditioner 100, the control unit 12 changes the value of the weight w set corresponding to the sensor 3 to "0" as shown in Table 2 below. Thereby, the measurement result of the sensor 3 is excluded from the calculation of the sum S prev and the sum S new of the above formulas (2) to (4).

Table 27

[0065] Also, for example, when at least one of the plurality of sensors 1 to 4 is affected by the temperature due to sunlight, the value of the weight w applied to at least one of the plurality of sensors may be set and changed based on at least one of the sunlight time, solar radiation amount, and ceiling temperature.

[0066] 1.5 Operational Effects According to the abnormality determination program of the present embodiment and the fire monitoring system equipped with the same, by integrating the physical quantities x measured by the plurality of sensors 1 to 4 and applying the weight w considering the mutual relationship of the plurality of sensors 1 to 4, it is possible to early determine that an abnormality that has a low probability of occurring during normal times has occurred.

[0067] Also, the sum S used for determining whether an abnormality has occurred prevSince it is calculated based on physical quantities x measured by multiple sensors 1 to 4, there is no need to deep learn a neural network. Furthermore, the calculation process performed in steps S3 to S10 in Figure 2 can determine whether or not an anomaly has occurred, and the computational load for this determination is extremely small.

[0068] 2. Simulation of the anomaly detection program The anomaly detection program of the present invention was simulated using "Microsoft Excel®," a spreadsheet software product from Microsoft Corporation®.

[0069] 2.1 Setting conditions for the monitoring area First, the conditions for the monitoring area to which the anomaly detection program of the present invention is applied were set as shown in Figures 6(a) and (b). In Figures 6(a) and (b), the monitoring area was set to have a ceiling and a floor, with no wall between the ceiling and the floor. Four sensors 1 to 4 were installed on the ceiling, and one fire source was installed on the floor.

[0070] The ceiling is 10m x 10m = 100m 2 The area was designed as a square, and the sensors were positioned at a height of H=8m from the floor. The four sensors 1-4 were positioned 5m apart from each other, and 2.5m away from each of the two sides of the ceiling. In other words, the ceiling was divided into four sections of 5m x 5m = 25m. 2 One sensor will be located at the center.

[0071] As shown in Figure 6(b), the fire source was positioned 1m vertically and 1m horizontally from the center point of the ceiling towards sensor 1. A six-tiered crib constructed of wood at equal intervals was assumed as the fire source. This assumption of a six-tiered crib is reflected in the value of the fire growth coefficient α used in the calculation of the heat generation rate Q of the fire source, which will be described next.

[0072] 2.2 Calculation of the heat generation rate Q of a fire source The quadratic function model Q=αt is used as a fire growth model for a fire source. 2 The following was used: Q is the rate of heat generation of the fire source, representing the amount of heat generated per second (kW). α is the fire growth coefficient, and t is time (seconds). The fire growth coefficient α is assumed to be 0.007 kW / s, assuming a 6-stage crib. 2 The rate of heat generation Q of the ignition source is calculated repeatedly over a predetermined period of time, for example, by adding 1 second at a time to t. When repeatedly calculating the rate of heat generation Q, the point in time t=0 is taken as the ignition point of the ignition source.

[0073] 2.3 Calculation of Ceiling Flow Temperature T By substituting the heat generation rate Q (kW) of the ignition source, the ceiling height H (m), and the horizontal distance r (m) between the ignition source and the sensor into Alpert's equation (7) below, the ceiling flow temperature rise ΔT (°C) at each position of sensors 1 to 4 is calculated. For example, the ceiling flow temperature rise ΔT over a predetermined time is calculated based on the heat generation rate Q calculated every second.

number

[0074] ΔT represents the "temperature rise" of the ceiling flow, and T (=ΔT+T0), obtained by adding the ambient temperature T0 to ΔT, represents the "temperature" of the ceiling flow at each of the sensor locations 1 to 4. Furthermore, in this simulation, random numbers that follow a standard normal distribution N(0,1) with a mean of 0 and a standard deviation of 1 are added to the ceiling flow temperature T calculated by ΔT+T0.

[0075] 2.4 Specific Examples of Calculating Ceiling Flow Temperature T The following describes specific examples of calculating the ceiling flow temperature T at each of the sensor locations 1 to 4. The following examples are single calculations with time t = 250 seconds and ambient temperature T0 = 20°C.

[0076] First, calculate the heat generation rate Q of the ignition source. For a time t = 250 seconds, the heat generation rate Q is 0.007 × 250.2 It becomes 437.5 kW. This means that 437.5 kJ of heat is generated by combustion per second.

[0077] Next, the horizontal distance r (m) between the heat source and the sensor is calculated. When the horizontal distance between the heat source and sensor 1 shown in Fig. 6(b) is r1, the horizontal distance between the heat source and sensor 2 is r2, the horizontal distance between the heat source and sensor 3 is r3, and the horizontal distance between the heat source and sensor 4 is r4, the respective horizontal distances r1, r2, r3, and r4 are as shown in the following formulas (8) to (11).

Equation

Equation

Equation

Equation

[0078] Next, the ceiling flow temperature rise ΔT (°C) at each position of sensors 1 to 4 is calculated using the above formula (7).

[0079] Since the horizontal distance r1 of sensor 1 is 2.12 m and the ceiling height H is 8 m, r1 / H = 0.265. Therefore, the ceiling flow temperature rise ΔT1 at the position of sensor 1 is calculated using the formula corresponding to (0.18 < r / H) among the two formulas in the above formula (7). The ceiling flow temperature rise ΔT1 is as shown in the following formula (12), and ΔT1 = 23.5 °C.

Equation

[0080] Since the horizontal distance r2 of sensor 2 is 3.81 m and the ceiling height H is 8 m, r2 / H = 0.476. Therefore, the ceiling flow temperature rise ΔT2 at the position of sensor 2 is calculated using the equation corresponding to (0.18 < r / H) among the two equations in the above equation (7). The ceiling flow temperature rise ΔT2 is as shown in the following equation (13), and ΔT2 = 15.9 °C.

Number

[0081] Since the horizontal distance r3 of sensor 3 is 4.95 m and the ceiling height H is 8 m, r3 / H = 0.619. Therefore, the ceiling flow temperature rise ΔT3 at the position of sensor 3 is calculated using the equation corresponding to (0.18 < r / H) among the two equations in the above equation (7). The ceiling flow temperature rise ΔT3 is as shown in the following equation (14), and ΔT3 = 13.3 °C.

Number

[0082] Since the horizontal distance r4 of sensor 4 is 3.81 m and the ceiling height H is 8 m, r4 / H = 0.476. Therefore, the ceiling flow temperature rise ΔT4 at the position of sensor 4 is calculated using the equation corresponding to (0.18 < r / H) among the two equations in the above equation (7). The ceiling flow temperature rise ΔT4 is as shown in the following equation (15), and ΔT4 = 15.9 °C.

Number

[0083] Next, the ambient temperature T0 is added to each of the ceiling flow temperature rises ΔT1 to ΔT4 to calculate the ceiling flow temperatures T1 to T4 at the respective positions of sensors 1 to 4. As described above, the ambient temperature T0 was set to 20 °C. The ceiling flow temperatures T1 to T4 at the respective positions of sensors 1 to 4 are as shown in the following equations (16) to (19).

Number

number

number

number

[0084] Finally, by adding random numbers that follow a standard normal distribution N(0,1) with a mean of 0 and a standard deviation of 1 to each of the ceiling flow temperatures T1 to T4, the temperature at each of the sensor locations 1 to 4 at time t = 250 seconds can be obtained.

[0085] 2.5 Temperatures of sensors 1-4 obtained by calculation Based on the settings of the monitoring area shown in Figures 6(a) and (b), and the heat generation rate of the ignition source Q = αt 2 Furthermore, using equations (7) to (19) above, the temperature at each position of sensors 1 to 4 from -60 seconds to 360 seconds was calculated every second. In this case, the ambient temperature T0 was assumed to be 20°C.

[0086] Figure 7(a) shows the temperature of sensor 1, Figure 7(b) shows the temperature of sensor 2, Figure 8(a) shows the temperature of sensor 3, and Figure 8(b) shows the temperature of sensor 4. In these figures, the point at 0 seconds is considered the ignition point of the ignition source, and the temperature from -60 seconds to 0 seconds was calculated assuming the heat generation rate Q of the ignition source was 0 (kW). In other words, before the ignition point of the ignition source, the ceiling flow temperature rise ΔT1 to ΔT4 are all 0, and the temperatures of sensors 1 to 4 are values ​​obtained by adding a random number to the ambient temperature T0 of 20°C. After the ignition point, the heat generation rate Q of the ignition source is Q = αt 2 The temperature was calculated from 1 second to 360 seconds by adding 1 second to the time t.

[0087] 2.6 Conversion from physical quantity x to physical quantity y The temperature readings per second from sensors 1-4 shown in Figures 7(a) and 8(a) and 8(b) correspond to a physical quantity x with a random number added to it. Using equation (1) above, each of the temperature readings per second from sensors 1-4 (physical quantity x) is converted to a physical quantity y. The average value in equation (1) above. [Table 28] And to calculate the standard deviation σ, multiple physical quantities x calculated before the physical quantity x to be converted to physical quantity y are used. For example, when converting the physical quantity x at 270 seconds to physical quantity y, 20 physical quantities x from 250 seconds to 269 seconds are used, and the average value in equation (1) above is used. [Table 29] And calculate the standard deviation σ.

[0088] 2.7 Sum S prev Calculation In this simulation, using multiple physical quantities y calculated before the ignition point at 0 seconds shown in Figures 7(a) and (b) and Figures 8(a) and (b), the sum S of the physical quantities y of sensors 1 to 4 is calculated using equation (2) above. prev Calculate the sum S. For example, the total sum S prev The calculation uses 20 physical quantities y ranging from -40 seconds to -21 seconds. Also, the weight w in equation (2) above is used. (Q,P) Use the values ​​from Table 1 above.

[0089] 2.8 Sum S new Calculation S is the sum of the physical quantities y from sensors 1 to 4. prev After calculating the above equation (2), every second a new physical quantity y is calculated, the sum S of the physical quantities y of sensors 1 to 4 is calculated using equation (2). new Calculate the sum S at the 270-second mark. For example, the sum S at the 270-second mark. new The calculation uses 20 physical quantities y ranging from 251 seconds to 270 seconds. Also, the weight w in equation (2) above is used. (Q,P) Use the values ​​from Table 1 above.

[0090] 2.9 Average [Table 30] Calculation S is the sum of the physical quantities y from sensors 1 to 4. new Each second in which the sum S is calculated, new average value [Table 31] Calculate the sum S at the 270-second mark. For example, the sum S at the 270-second mark. new At the time of calculation, the sum of 20 values ​​from 250 seconds to 269 seconds S new Using the average [Table 32] Calculate.

[0091] 2.10 Standard deviation σ new Calculation S is the sum of the physical quantities y from sensors 1 to 4. new Each second in which the sum S is calculated, new Standard deviation σ new Calculate the sum S at the 270-second mark. For example, the sum S at the 270-second mark. new At the time of calculation, the sum of 20 values ​​from 250 seconds to 269 seconds S new Using the standard deviation σ new Calculate.

[0092] 2.11 Calculation of t-values ​​and probability values ​​of the t-distribution average value [Table 33] , standard deviation σ new and sum S prev average value [Table 34] The value of t is calculated by applying this to equation (5) above, and the probability value of the t-distribution is calculated based on this value of t. Figure 9 is a graph showing the value of t calculated every second from -22 seconds to 360 seconds, and the probability value of the t-distribution calculated from this value of t. Note that the sum S in equation (5) above prev average value [Table 35] For example, the sum S of 20 values ​​calculated every second between -40 seconds and -21 seconds. prev It is calculated using [this method].

[0093] As shown in Figure 9, approximately 90 seconds after ignition of the fire source, the value of t is approximately 6, and the probability value of the t-distribution is 1.0 × 10⁻⁶. -4 ~1.0×10 -5 It decreased during this period. For example, in the anomaly detection program of the present invention, if the probability threshold is set to 1.0 × 10 -4 ~1.0×10 -5 If set within this range, it becomes possible to detect temperature abnormalities caused by a fire approximately 90 seconds after the ignition of the fire source.

[0094] On the other hand, as shown in Figures 7(a) and (b) and Figures 8(a) and (b), the temperature measured by each of the four sensors 1 to 4 approximately 90 seconds after ignition of the fire source was approximately 25°C. If the anomaly detection program of the present invention sets the probability threshold to 1.0 × 10 -4 ~1.0×10 -5 If set within this range, it is possible to detect a temperature anomaly caused by a fire when the temperature rises by approximately 5°C from the ambient temperature T0 of 20°C.

[0095] In contrast, typical fixed-temperature heat detectors are classified into three types based on their reaction time: "Special" > "Type 1" > "Type 2" > "Type 3". The "Special" fixed-temperature heat detector, which has the fastest reaction time, is configured to alert of a fire when it detects a temperature of 60°C or higher, meaning it will not activate at all with a temperature rise of approximately 5°C from an initial temperature of 20°C.

[0096] 3. Second embodiment of the fire monitoring system Figure 10 shows a fire monitoring system according to a second embodiment of the present invention. The fire monitoring system of this embodiment mainly consists of a plurality of sensors 1 to 4 installed in a plurality of monitoring areas, at least one repeater 20, and a receiver 10. The repeater 20 is electrically connected to the receiver 10 via a signal line 5 to the sensors 1 to 4 installed in any one of the monitoring areas. The repeater 20 has a receiving unit 11 and a control unit 12 similar to those of the first embodiment, and has an abnormality detection program similar to that of the first embodiment installed. On the other hand, the receiver 10 of this embodiment is an existing product and does not have an abnormality detection program installed in the receiver 10.

[0097] The receiving unit 11 of the repeater 20 is electrically connected to the multiple sensors 1 to 4 via the signal line 5. The receiving unit 11 receives digital signals output from the multiple sensors 1 to 4. These digital signals include physical quantities x measured by each of the multiple sensors 1 to 4. The receiving unit 11 transmits the received digital signals to the control unit 12. The control unit 12 is equipped with a processor for performing calculations.

[0098] Firstly, the processor of the control unit 12 transmits a digital signal containing a physical quantity x received from multiple sensors 1 to 4 to the receiver 10 and / or other repeaters, other receivers, or fire extinguishing equipment. For example, the receiver 10 performs a fire detection process similar to step S1 shown in Figure 2 based on the digital signal containing the physical quantity x, and determines whether or not a fire has occurred. If the receiver 10 determines that a fire has occurred, it causes the notification unit 13 to notify the receiver of the fire.

[0099] Secondly, the processor of the control unit 12 executes the processes of steps S2 to S10, S11 to S14 and S21 to S26 shown in Figure 2, according to an abnormality detection program similar to that of the first embodiment. If the control unit 12 determines in step S10 that an abnormality has occurred, it transmits an abnormality detection signal 5a to the receiver 10 and / or other repeaters, other receivers, or fire extinguishing equipment. For example, the receiver 10 causes the notification unit 13 to notify the occurrence of an abnormality based on the abnormality detection signal 5a.

[0100] In the fire monitoring system of this embodiment described above, an abnormality detection program is installed in the repeater 20. Therefore, by adding the repeater 20 to an existing fire monitoring system installed, for example, in an office building, tenant building, or logistics warehouse, it becomes possible to detect the occurrence of an abnormality in a predetermined monitoring area. In other words, the fire monitoring system of the present invention can be constructed using existing sensors 1 to 4, signal line 5, and receiver 10.

[0101] Furthermore, the repeater 20 can be selectively applied to multiple monitoring areas, allowing the user to detect anomalies in any monitoring area of ​​their choice. In addition, the repeater 20 can be easily installed by connecting signal lines 5 to its input and output terminals. Moreover, by installing an anomaly detection program on the repeater 20, the computational load on the receiver 10 can be reduced, and the amount of information stored in memory can be decreased. [Explanation of Symbols]

[0102] 1-4 sensors 5 signal lines 10 Receivers 11 Receiving unit 12. Control Unit (Processor) 13 Hochi Department 20 Repeaters 22 Control Unit (Processor) 5a Anomaly detection signal

Claims

1. An anomaly determination program that determines whether an anomaly has occurred based on signals output from a plurality of sensors arranged at a distance from each other in a single monitoring area for measuring a physical quantity x that changes due to a fire occurring in the monitoring area, Under normal conditions where no fire has occurred, multiple physical quantities x measured by multiple sensors are collected over a predetermined period of time or a predetermined number of times, each of the collected physical quantities x is converted into a physical quantity y normalized to a standard normal distribution, and the sum S of the physical quantities y is calculated. prev The first process for calculating, After the predetermined time has elapsed or after the predetermined number of times has been exceeded, multiple physical quantities x measured by multiple sensors are collected over a predetermined time or predetermined number of times, each of the collected physical quantities x is converted into a physical quantity y normalized to a standard normal distribution, and the sum S of the physical quantities y is calculated. new The second process for calculating, Sum S new average value Table 36 and standard deviation σ new A third process to calculate, average value Table 37 , standard deviation σ new and sum S prev average value Table 38 A fourth process involves determining whether the probability calculated by applying this to the cumulative distribution function is below a predetermined probability threshold, and if it is below the threshold, determining that an anomaly has occurred. An anomaly detection program characterized by causing a processor to execute a process that includes [a specific process].

2. A weight w corresponding to each of a plurality of sensors is set in advance, in the first process the sum S prev of physical quantities y multiplied by the weight w is calculated, and in the second process the sum S new of physical quantities y multiplied by the weight w is calculated, An anomaly detection program according to claim 1, which, upon receiving a signal that identifies a disturbance for each of the individual sensors, causes the processor to perform a fifth process of selecting the sensor affected by the disturbance and changing the weight w corresponding to the selected sensor.

3. The anomaly detection program according to claim 2, wherein each of the physical quantities x collected in the first and second processes is converted into a physical quantity y normalized to a standard normal distribution by the following formula (1). [Number 25] however, 【number】 σ is the mean value of the physical quantity x, and σ is the standard deviation of the physical quantity x.

4. The sum S in the first and second processes described above. prev and sum S new The abnormality detection program according to claim 2, which is calculated by any of the following formulas (2) to (4). [Number 26] [Number 27] [Number 28] However, P and Q each represent a plurality of sensors used for measuring the physical quantity x. y (p) is the value of the physical quantity y of sensor P, and y (Q) is the value of the physical quantity y of sensor Q. w (Q,P) is the weight w set based on the correlation between sensor Q and sensor P.

5. The anomaly detection program according to claim 1, wherein the probability in the fourth process is calculated based on the value of t calculated by the following formula (5). [Number 29] However, n is the mean value. 【number】 The sum S used in the calculation new This is the number of data points.

6. An anomaly detection program according to claim 2, wherein when at least one of the multiple sensors is affected by the temperature of the air conditioner, the value of the weight w applied to at least one of the multiple sensors is changed based on information regarding the operation of the air conditioner.

7. An anomaly detection program according to claim 2, wherein when at least one of the multiple sensors is affected by temperature due to sunlight, the value of the weight w applied to at least one of the multiple sensors is changed based on at least one of the time of sunlight, amount of solar radiation, and temperature.

8. An anomaly detection program according to claim 1, which generates random numbers to increase the significant figures of a physical quantity x in the first process, adds the random numbers to each of the multiple physical quantities x measured by the multiple sensors, and then calculates a normalized physical quantity y.

9. A fire monitoring system that operates in accordance with an abnormality detection program described in any one of claims 1 to 8, Equipped with multiple sensors and receivers, The receiver comprises a receiving unit, a control unit, and a notification unit, and has the abnormality detection program installed. Multiple sensors are installed in the monitoring area and measure a physical quantity x that changes due to the occurrence of a fire, outputting a signal. The receiving unit receives signals output from multiple sensors, The processor of the control unit executes the first to fourth processes according to the anomaly detection program based on a plurality of physical quantities x measured by a plurality of sensors. The notification unit, when it determines that an abnormality has occurred in the fourth process, notifies that an abnormality has occurred. A fire monitoring system characterized by the following features.

10. The abnormality detection program includes a sixth process which determines whether a physical quantity x measured by at least one of the multiple sensors is equal to or greater than a preset threshold for the physical quantity x, and if it is equal to or greater than the threshold, it determines that a fire has occurred. The processor of the control unit executes the sixth process according to the anomaly detection program based on each of the multiple physical quantities x measured by the multiple sensors. The fire monitoring system according to claim 9, wherein the notification unit notifies that a fire has occurred when it is determined in the sixth process that a fire has occurred.

11. A fire monitoring system that operates in accordance with an abnormality detection program described in any one of claims 1 to 8, It comprises multiple sensors, at least one repeater, and a receiver. The relay unit comprises a receiving unit and a control unit, and the abnormality detection program is installed on it. Multiple sensors are installed in the monitoring area and measure a physical quantity x that changes due to the occurrence of a fire, outputting a signal. The receiving unit receives signals output from multiple sensors, The processor of the control unit executes the first to fourth processes according to the anomaly detection program based on a plurality of physical quantities x measured by a plurality of sensors, and if it is determined in the fourth process that an anomaly has occurred, it transmits a signal to another repeater or the receiver. A fire monitoring system characterized by the following features.

12. The abnormality detection program includes a sixth process which determines whether a physical quantity x measured by at least one of the multiple sensors is equal to or greater than a preset threshold for the physical quantity x, and if it is equal to or greater than the threshold, it determines that a fire has occurred. The fire monitoring system according to claim 11, wherein the processor of the control unit performs the sixth process according to the abnormality determination program based on each of the multiple physical quantities x measured by the multiple sensors, and when it is determined in the sixth process that a fire has occurred, it transmits a signal to another repeater or the receiver.

13. An anomaly detection program that determines whether an anomaly has occurred based on signals output from a plurality of sensors arranged at a distance from each other in a single monitoring area for measuring a physical quantity x, Multiple physical quantities x measured by multiple sensors are collected over a predetermined period of time or a predetermined number of times, each of the collected physical quantities x is converted into a physical quantity y normalized to a standard normal distribution, and the sum S of the physical quantities y is calculated. prev The first process for calculating, After the predetermined time has elapsed or after the predetermined number of times has been exceeded, multiple physical quantities x measured by multiple sensors are collected over a predetermined time or predetermined number of times, each of the collected physical quantities x is converted into a physical quantity y normalized to a standard normal distribution, and the sum S of the physical quantities y is obtained. new The second process for calculating, Sum S new average value Table 41 and standard deviation σ new A third process to calculate, average value Table 42 , standard deviation σ new and sum S prev average value Table 43 A fourth process involves determining whether the probability calculated by applying this to the cumulative distribution function is below a predetermined probability threshold, and if it is below the threshold, determining that an anomaly has occurred. An anomaly detection program characterized by causing a processor to execute a process that includes [a specific process].

14. A weight w corresponding to each of the multiple sensors is set in advance, and in the first process, the sum S of the physical quantities y multiplied by the weight w is used. prev The following is calculated, and in the second process, the sum S of the physical quantities y multiplied by the weight w is calculated. new The following was calculated: An anomaly detection program according to claim 13, which, upon receiving a signal that identifies a disturbance for each of the individual sensors, causes the processor to perform a fifth process of selecting the sensor affected by the disturbance and changing the weight w corresponding to the selected sensor.

Images