Anomaly prediction detection system for mobile air-powered systems

Abstract

To detect an abnormality sign of an air compressor mounted on a railway vehicle and of an air unit system including the same.SOLUTION: An abnormality sign detection system detecting an abnormality sign of an air unit system for a mobile body comprising an air compressor making compressed-air, an air tank storing the compressed-air, and a compressed-air use device using the compressed-air stored in the air tank, is provided with: a data collection unit collecting historical data for the air unit system for the mobile body; and an abnormality sign detection unit detecting the abnormality sign of the air unit system for the mobile body by comparing estimated working time of the air compressor with actual working time of the air compressor based on a compressed-air use model representing a relation expression of working time of the air compressor calculated through the historical data collected with the data collection unit and an air use amount of a compressed-air use device.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an abnormal sign detection system for an air system including an air compressor for railway vehicles.

Background Art

[0002] From the viewpoint of improving the efficiency and labor saving of maintenance work of railway vehicles, the development of abnormal sign detection technology using data obtained from on-vehicle equipment has been promoted. In particular, air compressors mounted on railway vehicles and the air systems including them are devices for filling compressed air into an air tank for supplying compressed air to an air spring, an air brake, etc. Since an abnormality in the device affects the operation, there is a high need for abnormal sign detection.

[0003] As a technology for solving this, as shown in Patent Document 1, from the relationship between the pressure accumulation information during the operation of the compressor (air compressor) and the absence of operation of the brake device, and the pressure displacement of the air spring, the occupancy rate, the outside air temperature, and the vehicle speed within a predetermined time, a model formula is created for the pressure increase amount of the pressure accumulation system by multiple regression analysis, and a means for determining the state of the compressor (air compressor) from the predicted value of the pressure increase amount of the pressure accumulation system obtained from the model formula and the actual pressure increase amount of the pressure accumulation system is provided.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, since the method described in Patent Document 1 uses data of the brake device that uses compressed air during the operation of the air compressor, although an abnormality related to the pressure accumulation performance of the air compressor is detected, an abnormality cannot be extracted when there is an abnormality in the air system including the air compressor. [Means for solving the problem]

[0006] To solve the above problems, the present invention provides an abnormality prediction detection system for a mobile air system comprising an air compressor that produces compressed air, an air tank that stores the compressed air, and a compressed air usage device that uses the compressed air stored in the air tank, wherein the system includes a data collection unit that collects historical data in the mobile air system, The operating time and compressed air usage of the air compressor are calculated from the historical data collected by the data collection unit. Device An abnormality prediction unit detects an abnormality in the mobile air system by comparing the estimated operating time of the air compressor with the actual operating time of the air compressor based on a compressed air usage model that shows the relationship between the amount of air used, This is an anomaly prediction detection system characterized by having the following features. [Effects of the Invention]

[0007] According to the present invention, it is possible to extract not only abnormalities related to the pressure accumulation performance of the air compressor, but also signs of abnormalities in the air system. Furthermore, by extracting the location where the signs of abnormality occur, prompt maintenance becomes possible. Other issues, configurations, and effects not mentioned above will be clarified by the following description of the embodiments. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 shows an abnormality prediction system for a mobile air system in Embodiment 1 of the present invention. [Figure 2] Figure 2 shows a configuration in which the abnormality prediction detection system shown in Figure 1 is further modified by adding a mobile air system, which is the device targeted for abnormality prediction diagnosis, and a compressed air usage model estimation unit that estimates the compressed air usage model. [Figure 3] Figure 3 shows an example of the configuration of a data registration device. [Figure 4]Figure 4 shows the operating status of the air compressor and the instantaneous air consumption and total air consumption of the devices that use compressed air. [Figure 5] Figure 5 shows the operating status of the air compressor and the amount of compressed air used by each compressed air user. [Figure 6] Figure 6 shows an example of registered data. [Figure 7] Figure 7 shows an example where the horizontal axis represents elapsed time and the vertical axis represents the operating interval. It compares the estimated operating interval calculated by the anomaly prediction detection unit (dotted line) with the actual operating interval (solid line), and detects an anomaly when the estimated operating interval becomes shorter than the actual operating interval by a threshold. [Figure 8] Figure 8, like Figure 4, shows the operating status of the air compressor and the air consumption of the equipment that uses compressed air. [Figure 9] Figure 9 shows an abnormality prediction detection system for a mobile air system in Embodiment 2 of the present invention. [Figure 10] Figure 10 shows an example of the configuration of a data registration device. [Figure 11] Figure 11 shows an example of segmented data. [Figure 12] Figure 12 shows an example of registered data. [Figure 13] Figure 13 shows an example of the configuration of the compressed air usage model estimation unit. [Figure 14] Figure 14 shows an example of the ambient temperature segmentation process. [Figure 15] Figure 15 shows a group of compressed air usage models. [Figure 16] Figure 16 shows an example of the configuration of the anomaly prediction detection unit. [Figure 17] Figure 17 shows an abnormality prediction detection system for a mobile air system in Embodiment 3 of the present invention. [Figure 18] Figure 18 shows an example of the configuration of a data registration device. [Figure 19] Figure 19 shows an example of segmented data. [Figure 20]FIG. 20 is a diagram showing an example of registered data. [Figure 21] FIG. 21 is a diagram showing a configuration example of a compressed air usage model estimation unit. [Figure 22] FIG. 22 is a diagram showing an example of a formation characteristic information division process. [Figure 23] FIG. 23 is a diagram showing a group of compressed air usage models. [Figure 24] FIG. 24 is a diagram showing a configuration example of an abnormality omen detection unit. [Figure 25] FIG. 25 is a diagram showing an abnormality omen detection system of a mobile body air system in Example 4 of the present invention. [Figure 26] FIG. 26 is a diagram showing a configuration example of a data registration device. [Figure 27] FIG. 27 is a diagram showing an example of divided data. [Figure 28] FIG. 28 is a diagram showing an example of registered data. [Figure 29] FIG. 29 is a diagram showing a configuration example of a compressed air usage model estimation unit. [Figure 30] FIG. 30 is a diagram showing an example of a group of compressed air usage models. [Figure 31] FIG. 31 is a diagram showing a configuration example of an abnormality omen detection unit. [Figure 32] FIG. 32 is a diagram showing an abnormality omen detection system of an air system in Example 5 of the present invention. [Figure 33] FIG. 33 is a diagram showing a configuration example of a data registration device. [Figure 34] FIG. 34 is a diagram showing the operating state of an air compressor, the instantaneous air usage amount of a device using compressed air, and their total. [Figure 35] FIG. 35 is a diagram showing the operating state of an air compressor and the usage amounts of respective compressed air usage devices. [Figure 36] FIG. 36 is a diagram showing an example of data for abnormality omen determination. [Figure 37]Figure 37 shows an example where the horizontal axis represents elapsed time and the vertical axis represents operating time. It compares the estimated operating time (dotted line) calculated by the anomaly prediction detection unit 3102 with the actual operating time (solid line), and detects an anomaly when the actual operating time exceeds a threshold compared to the estimated operating time. [Figure 38] Figure 38 shows an abnormality prediction detection system for a mobile air system in Embodiment 6 of the present invention. [Modes for carrying out the invention]

[0009] Embodiments of the present invention will be described below with reference to the drawings. However, the present invention is not limited to these embodiments. Furthermore, in the drawings, identical parts are denoted by the same reference numerals.

[0010] [Example 1] In this embodiment, in order to solve the problem that abnormalities related to the pressure accumulation performance of an air compressor can be detected, but abnormalities in the air system including the air compressor cannot be extracted, we focus on the compressed air usage during the time intervals when the air compressor is not operating (hereinafter referred to as the operating interval). Based on the brake cylinder pressure (hereinafter referred to as BC pressure), air spring pressure (hereinafter referred to as AS pressure) output by the brake device which affects air consumption, and the operating interval, a model is created to predict air consumption. Based on the created air consumption model, the predicted operating interval is calculated, and when the actual operating interval is measured to be more than a threshold shorter than the predicted operating interval a certain number of times, it is determined that there is a sign of an abnormality. Furthermore, from the above air consumption model, the operating time, including cases where compressed air is used during operation, is predicted, and when the actual operating time is measured to be more than a threshold longer than the predicted operating time a certain number of times, it is determined that there is a sign of an abnormality. In addition, the location of the abnormality is extracted based on the air consumption model using the air consumption model based on the determination results of the operating interval and operating time and the abnormality prediction data.

[0011] Figure 1 shows an abnormality prediction detection system for a mobile air system in Embodiment 1 of the present invention. The abnormality prediction detection system comprises a data acquisition device 106, a data registration device 107, a history database 108, and an abnormality prediction detection unit 110.

[0012] The data collection device 106 collects compressed air usage data 154A, 154B, 154C, ... (hereinafter referred to collectively as compressed air usage data 154) which is the amount of compressed air used by the compressed air usage devices 103A, 103B, 103C, ... (hereinafter referred to collectively as compressed air usage devices 103), and operating status data 151 of the air compressor 101, which will be described later.

[0013] The data registration device 107 converts the collected data 155, which consists of the operating status data 151 of the air compressor 101 and the compressed air usage data 154 of the compressed air user device 103 stored in the data acquisition device 106, into registration data 156 and abnormality prediction data 157 based on predetermined rules. The history database 108 stores the registration data 156.

[0014] The data registration device 107 converts the collected data 155, which consists of the operating status data 151 of the air compressor 101 and the compressed air usage data 154 of the compressed air user device 103 stored in the data acquisition device 106, into registration data 156 and abnormality prediction data 157 based on predetermined rules.

[0015] The history database 108 stores the registered data 156 and a compressed air usage model 159 of the mobile air system 105, which will be described later, estimated using a portion of the registered data 156.

[0016] The abnormality prediction detection unit 110 determines an abnormality 160 of the mobile air system 105 by comparing the operating interval estimated from the abnormality prediction determination data 157 based on the compressed air usage model 159 with the actual operating interval included in the abnormality prediction determination data 157, and outputs the result.

[0017] Figure 2 shows a configuration in which the abnormality prediction detection system shown in Figure 1 is modified by adding a compressed air usage model estimation unit 109 that estimates the mobile air system 105 and the compressed air usage model 159, which are the devices targeted for abnormality prediction diagnosis.

[0018] The mobile air system 105, which is an additional component to the aforementioned abnormality prediction detection system, consists of an air compressor 101 that produces compressed air, an air tank 102 that stores the compressed air, a plurality of compressed air users 103A, 103B, 103C, ... (hereinafter, "compressed air users 103" refers collectively to these) that use the compressed air stored in the air tank 102, and an air compressor control device 104 that determines a control command 153 for the air compressor 101 based on the operating status data 151 of the air compressor 101 and the pressure information 152 of the air tank 102. Furthermore, the compressed air usage model estimation unit 109 estimates the compressed air usage model 159 of the mobile air system 105 based on history data 158, which is part of the registered data 156 contained in the history database 108.

[0019] In Figure 2, solid lines represent mechanical connections, and dotted lines represent the flow of information. Regarding the flow of information, communication may be used in addition to mechanical or electrical connections. For example, the pressure information 152 of the air tank 102 may be measured using sensors within the air tank 102 and transmitted to the air compressor control device 104 via communication, or the air tank 102 and the air compressor control device 104 may be mechanically connected, allowing the air compressor control device 104 to directly measure the pressure information 152 of the air tank 102.

[0020] For the sake of explanation, the following explanation will be based on the configuration shown in Figure 2. However, the abnormality prediction detection system shown in Figure 1 may also be configured with either the mobile air system 105 or the compressed air usage model estimation unit 109 added in Figure 2. In the configuration shown in Figure 1, the operating status data 151 and compressed air usage data 154 can be transmitted to and received by the data acquisition device 106. Furthermore, the compressed air usage model estimation unit 109 can be located externally, and the estimated compressed air usage model 159 can be stored in the history database 108.

[0021] Furthermore, in Figures 1 and 2, the data acquisition device 106, data registration device 107, history database 108, compressed air usage model estimation unit 109, and abnormality prediction detection unit 110 are not included in the mobile air system 105. However, even if any or all of the data acquisition device 106, data registration device 107, history database 108, compressed air usage model estimation unit 109, and abnormality prediction detection unit 110 are included in the mobile air system 105, the invention will not be hindered.

[0022] Furthermore, the mobile air system 105, data acquisition device 106, data registration device 107, history database 108, compressed air usage model estimation unit 109, and anomaly prediction detection unit 110 do not need to be independent devices; the invention can function even if they are integrated into a single system. When integrated into a single system, it becomes possible to process everything with a single CPU, enabling an integrated software configuration, allowing for wired data transmission, and simplifying manufacturing. On the other hand, if each device is independent, it can be used as an add-on, making it easier to add other devices later.

[0023] When the moving object is a railway vehicle, the compressed air-using devices 103 include brake systems, air springs, and pneumatic doors. These devices may also use compressed air when the railway vehicle is running on inclines or curves. For this reason, the operating status data 151, air tank 102 pressure information 152, and compressed air usage data 154 acquired from a stationary air compressor will have different characteristics from the operating status data 151, air tank 102 pressure information 152, and compressed air usage data 154 acquired while the railway vehicle is running.

[0024] The data registration device 107 will be explained using Figure 3. Figure 3 shows an example of the configuration of the data registration device 107. The data registration device 107 consists of a section division process 201 that divides the collected data 155 into sections based on a predetermined rule to form sectioned data 251, a cumulative value calculation process 202 that calculates cumulative values ​​252 (252A, 252B, 252C…) based on compressed air usage data 154 (154A, 154B, 154C…) used by compressed air users 103 (103A, 103B, 103C…) included in the sectioned data 251, and a registration data creation process 203 that combines the time included in the sectioned data 251 and the cumulative values ​​252 to form registration data 156 and abnormality prediction data 157. The registration data 156 and abnormality prediction data 157 are identical in data format. The collected data 155 is distributed to the registration data 156 and abnormality prediction data 157 by any method. For example, registration data 156 could be data from the start of use up to one month, while abnormality prediction data 157 could be data from thereafter. Furthermore, historical data 158, which is part of the registration data 156, is used by the compressed air usage model estimation unit 109 to create a compressed air usage model, whereas abnormality prediction data 157 is not used to create a compressed air usage model, but is used to determine abnormalities in the mobile air system 105. A point that must be observed when considering these methods is that the historical data 158 used to create the compressed air usage model must not include the abnormality prediction data 157. By not including the abnormality prediction data 157 in the historical data 158, data indicating abnormalities will not be mixed with the model creation data, making it possible to create model equations and diagnose abnormalities more accurately. However, it is also possible to record the abnormality prediction data 157 in the historical database after it has been used to determine abnormalities, or to record it in the historical database along with the distribution while keeping it separate from the historical data. Furthermore, if it has been used to determine an anomaly, it may be included as part of the historical data 158. The specific details of registration data 156 will be described later.

[0025] In this embodiment, the interval division process 201 divides the collected data 155 into intervals based on the operating interval of the air compressor. Using this interval division rule, it becomes possible to extract the collected data 155 based on the operating interval, reducing the influence of air compressor operation and making it easier to detect abnormalities in the air system. The operating interval will be explained using Figure 4.

[0026] Figure 4 shows the operating state of the air compressor and the instantaneous state of the equipment using compressed air. compression This diagram shows air consumption and its total consumption. The horizontal axis represents time, and the vertical axis, from top to bottom, represents the operating status of the air compressor, instantaneous compressed air consumption, and the operating interval. compression This shows the cumulative value of air consumption. The operating interval is the time from the time the air compressor finishes operating (t1) to the time it starts operating again (t2). Therefore, the value obtained at t2, which marks the end of the operating interval, represents the cumulative value during that operating interval. compression This represents the amount of air used. Note that once this value is registered... compression The cumulative value of air consumption is reset, so at t2, it changes to 0.

[0027] An example of segmentation is shown using Figure 5. Figure 5 is a diagram showing the operating status of the air compressor and the amount of compressed air used by each compressed air user. Figure 5 shows the case where there are three compressed air users. The collected data 155 and segmented data 251 consist of time-series data and include at least the amount of compressed air used by compressed air user 1, the amount of compressed air used by compressed air user 2, the amount of compressed air used by compressed air user 3, and the operating status of the air compressor. The collected data 155 is data that includes the entire time being measured, whereas the segmented data 251 is data divided according to the operating interval of the air compressor, and the data for one operating interval is considered to be from the point in time when the operating status of the air compressor changed from ON to OFF to the point in time when it changed from OFF to ON. In this figure, three sets of segmented data 251 are obtained.

[0028] The cumulative value calculation process 202 obtains a cumulative value 252 by summing the usage amounts of each compressed air user during the operating interval. In the case of the segmented data 251 shown in Figure 5, the cumulative value for compressed air user 1 is 15+25+25=65, the cumulative value for compressed air user 2 is 20+15+30+10=75, and the cumulative value for compressed air user 3 is 20+20+45+10+20+10=125.

[0029] The registration data 156, which is created in the registration data creation process 203 and registered in the history database 108, will be explained using Figure 6.

[0030] Figure 6 shows an example of registered data 156. Registered data 156 consists of the cumulative value of compressed air user 1, the cumulative value of compressed air user 2, the cumulative value of compressed air user 3, and the operating interval obtained when the data is divided into intervals by the interval division process 201. In the case of the divided data 251 shown in Figure 5, the cumulative value of compressed air user 1 is 65, the cumulative value of compressed air user 2 is 75, and the cumulative value of compressed air user 3 is 125, as calculated above. If this data is recorded every second, then in Figure 5 there are 10 intervals where "Air compressor operating status" is recorded as OFF, so the operating interval can be calculated as 10 seconds.

[0031] The compressed air usage model estimation unit 109 will now be explained. The compressed air usage model estimation unit 109 calculates the compressed air usage model 159 using historical data 158 which is part of the registered data 156 included in the aforementioned historical database 108 and does not include the abnormality prediction judgment data 157. The method for calculating the compressed air usage model 159 will now be explained. Based on the historical data 158, [Formula 1] Yj = X[1,j]·a1+X[2,j]·a2+X[3,j]·a3+…+X[n,j]·an (Equation 1) This is created according to the number of Yj(j=1,2,…,m). Yj(j=1,2,…,m) represents the operating interval, which is the time width of the interval division shown in Figure 3, and X[i,j](i=1,2,…,n)(j=1,2,…,m) is a parameter relating to the cumulative amount in each compressed air user during the operating interval Yj(j=1,2,…,m). The multiple obtained (Equation 1) are treated as a system of simultaneous equations, and the coefficient ai(i=1,2,…,n) for each compressed air user is calculated using methods such as the least squares method. This allows us to calculate the coefficient ai(i=1,2,…,n) for each compressed air user. [Formula 2] Y = a1·X1+a2·X2+a3·X3+…+an·Xn (Formula 2) This is obtained. Let's call this compressed air usage model 159. Note that Y represents the estimated operating interval.

[0032] The abnormality prediction detection unit 110 calculates the estimated operating interval (dotted line) for a given time interval by substituting the cumulative amount Xi (i=1,2,…,n) of each compressed air user during the operating interval, which is determined based on Figure 4, into the compressed air usage model 159 (Equation 2). By comparing this with the actual operating interval, the unit determines the abnormality prediction 160. The processing results of the abnormality prediction detection unit 110 will be explained using Figure 7.

[0033] Figure 7 shows an example where the horizontal axis represents elapsed time and the vertical axis represents the operating interval. It compares the estimated operating interval (dotted line) calculated by the anomaly prediction detection unit 110 with the actual operating interval (solid line), and detects an anomaly when the estimated operating interval becomes shorter than the actual operating interval by a threshold. By making this determination, the presence or absence of an anomaly can be determined. Note that an anomaly may be determined not by a single detection, but by multiple detections within a certain time range, or by multiple consecutive detections.

[0034] By using the method described above, it becomes possible to estimate the operating interval based on a compressed air consumption model created by focusing on the period during which compressed air is used in the mobile vehicle (operating interval), and then compare it with the actual operating interval to determine if an anomaly has occurred.

[0035] In this embodiment, there are three compressed air users 103, but the present invention is not limited to three; any configuration including one or more compressed air users will not impede the content of this invention.

[0036] Furthermore, as a modification of this embodiment, in (Equation 1) using historical data 158 for creating the compressed air usage model 159, it is assumed that the amount of air leakage in the air piping is a constant amount b with respect to time, [Formula 3] Yj = X[1,j]·c1+X[2,j]·c2+X[3,j]·c3+…+X[n,j]·cn+b·Yj (Formula 3) A compressed air usage model 159 may be created. By anticipating a certain amount of air leakage in advance, false detection of abnormalities can be prevented. Furthermore, if (Equation 3) is rearranged... [Formula 4] Yj = (X[1,j]·c1+X[2,j]·c2+X[3,j]·c3+…+X[n,j]·cn) / (1-b) (Equation 4) Yj(j=1,2,…,m) represents the operating interval, which is the time width of the interval division determined based on Figure 4, and X[i,j] (i=1,2,…,n) (j=1,2,…,m) is a parameter relating to the cumulative amount in each air-using device during the operating interval Yj(j=1,2,…,m). By considering the multiple obtained (Equation 4) as a system of simultaneous equations and using methods such as the least squares method, the coefficient ci / (1-b)(i=1,2,…,n) for each air-using device is calculated. Furthermore, from a comparison of (Equation 4) and (Equation 2), c1 / a1=c2 / a2=…cn / an, so b can be calculated by comparing the obtained ci / (1-b)(i=1,2,…,n) with the coefficient ai(i=1,2,…,n) of (Equation 2). Thus, [Formula 5] Y = (c1·X1+c2·X2+c3·X3+…+cn·Xn) / (1-b) (Equation 5) This can be obtained. This may be used as compressed air model 159. The content of the present invention remains unchanged even when using this compressed air model 159.

[0037] Furthermore, as a further modification of this embodiment, it is conceivable to change the time width of the interval division. An example of this is shown in Figure 8. Figure 8 is a diagram that, like Figure 4, shows the operating status of the air compressor and the air consumption of the equipment that uses compressed air. The horizontal axis represents time, and the vertical axis, from top to bottom, shows the operating status of the air compressor, the instantaneous amount of compressed air used, and the total amount of air used in the given interval. The difference in definition from Figure 4 is that the definition of the interval width is from the time when the air compressor stops operating (t1) to the time when it starts operating again and stops operating (t3). In other words, while Figure 4 is based on the operating interval of the air compressor, Figure 8 includes the operating time in addition to the operating interval of the air compressor. Operation The difference lies in the use of a period. Based on this time range, in order to implement the content shown in Example 1, the explanation above refers to the operating interval plus the operating time, rather than the operating interval. Operation Create a table as a period, and the indicator on the vertical axis of Figure 7 Operation By using a periodic cycle, it is possible to determine whether or not there is a malfunction in the air compressor using the method explained in Figure 7.

[0038] Furthermore, in this embodiment, the operating interval was estimated from the compressed air usage model 159 and compared with the measured value to detect abnormal signs, but the compressed air usage model 159 Device It is also possible to estimate the amount of compressed air used. Device Compressed air usage and actually measured compressed air usage Device By comparing the amount of compressed air used, it is also possible to detect signs of abnormalities.

[0039] [Example 2] Figure 9 shows an abnormality prediction system for a mobile air system in Embodiment 2 of the present invention. Components identical to those in Embodiment 1 are indicated with the same numbers as in Figure 2, and their explanations are omitted. As in Figure 2, solid lines represent mechanical connections, and dotted lines represent the flow of information.

[0040] An air compressor control device that determines a control command 153 for the air compressor 101 based on the operating status data 151 of the air compressor 101 and the pressure information 152 of the air tank 102, and the air compressor control device that determines a control command 153 for the air compressor 101 based on the operating status data 151 of the air compressor 101 and the pressure information 152 of the air tank 102. A mobile air system system 802 consisting of 104 and a temperature sensor 801 that measures ambient temperature data 851, compressed air usage data 154A, 154B, 154C, ... of the compressed air usage devices 103A, 103B, 103C, ... (hereinafter, compressed air usage data 154 refers collectively to these), a data acquisition device 803 that collects operating status data 151 of the air compressor 101 and ambient temperature data 851, and data acquisition device The system comprises a data registration device 804 that converts collected data 852, which consists of operating status data 151 of the air compressor 101, compressed air usage data 154 of the compressed air user 103, and ambient temperature data 851 stored in a device 803, into registered data 853 and abnormality prediction data 854 based on predetermined rules; a history database 108 that stores the registered data 853; a compressed air usage model estimation unit 805 that estimates a group of compressed air usage models 856 for the mobile air system 802 based on history data 855, which is part of the registered data 853 stored in the history database 108; and an abnormality prediction detection unit 806 that determines an abnormality 160 of the mobile air system 802 by comparing the operating interval estimated from the abnormality prediction data 854 based on the compressed air usage model group 856 with the actual operating interval included in the abnormality prediction data 854.

[0041] Furthermore, similar to Example 1, the invention is not hindered even if any or all of the data acquisition device 803, data registration device 804, history database 108, compressed air usage model estimation unit 805, and anomaly prediction detection unit 806 are included in the mobile air system 802. Also, the invention is not hindered even if the data acquisition device 803, data registration device 804, history database 108, compressed air usage model estimation unit 805, and anomaly prediction detection unit 806 are not independent devices but are included in a single integrated system.

[0042] The data registration device 804 will be explained using Figure 10. Figure 10 shows an example of the configuration of the data registration device 804. It consists of a section division process 201 that divides the collected data 852 into sections based on a predetermined rule to create sectioned data 951; a cumulative value calculation process 202 that calculates cumulative values ​​252 (252A, 252B, 252C…) based on compressed air usage data 154 (154A, 154B, 154C…) used by compressed air users 103 (103A, 103B, 103C…) included in the sectioned data 951; an outside temperature extraction process 901 that extracts a representative outside temperature 952 based on the outside temperature included in the sectioned data 951; and a registration data creation process 902 that combines the time, cumulative value 252, and representative outside temperature 952 included in the sectioned data 951 to create registration data 853 or abnormality prediction data 854. The registration data 853 and the abnormality prediction data 854 are identical in data format. Some of the historical data 855 from the registered data 853 is used by the compressed air usage model estimation unit 805 to create a compressed air usage model, whereas the abnormality prediction data 854 is not used to create a compressed air usage model, but is used to determine abnormalities in the mobile air system 802.

[0043] The collected data 852 is distributed to the registration data 853 and the anomaly prediction data 854 by any method. For example, the registration data 853 could be data from the first month of use, and the anomaly prediction data 854 could be data from thereafter. When considering these methods, it is important to note that the historical data 855 mentioned above must not include the anomaly prediction data 854. By not including the anomaly prediction data 854 in the historical data 855, data indicating anomalies will not be mixed with the model creation data, allowing for more accurate model equation creation and anomaly prediction diagnosis. However, it is also possible to record the anomaly prediction data 854 in the historical database after it has been used to determine an anomaly, or to record it in the historical database along with the distribution while keeping it separate from the historical data. Alternatively, if it has been used to determine an anomaly, it may be included as part of the historical data 855.

[0044] The specific details of registration data 853 will be described later. The collected data 852 consists of time-series data and shall include at least the compressed air consumption of compressed air device 1, compressed air consumption of compressed air device 2, compressed air consumption of compressed air device 3, the operating status of the air compressor, and the ambient temperature. The recording time for each data point may also be included.

[0045] The division method performed in the interval division process 201 is the same as in Figure 5 of Example 1, if the collected data 155 is replaced with collected data 852 and the divided data 251 is replaced with divided data 951. However, the data content differs from Example 1 in that it includes outside temperature. Note that the data format of collected data 852 and divided data 951 is the same. Figure 11 shows an example of divided data 951.

[0046] The process for extracting ambient temperature 901 will now be explained. Since one divided data set 951 contains N data points, there are also N ambient temperature data points. It is possible to use the average temperature of these N data points as the representative ambient temperature. Alternatively, methods such as using the highest temperature, lowest temperature, or the temperature with the most frequency can also be used. In the case of the divided data set 951 shown in Figure 11, if the average temperature is used as the representative ambient temperature, it will be (17 × 3 + 18 × 7) / 10 = 17.7 degrees. If it is the highest temperature, it will be 18 degrees. If it is the lowest temperature, it will be 17 degrees. If it is the temperature with the most frequency, it will be 18 degrees.

[0047] The registration data 853, which is created in the registration data creation process 902 and registered in the history database 108, will be explained using Figure 12. Figure 12 shows an example of registered data 853. Registered data 853 consists of the cumulative value of compressed air user 1, the cumulative value of compressed air user 2, the cumulative value of compressed air user 3, the representative outside temperature 952, and the operating interval obtained when the data is divided into intervals by the interval division process 201. In the case of the divided data 951 shown in Figure 11, the cumulative value of compressed air user 1 is 15+25+25=65, the cumulative value of compressed air user 2 is 20+15+30+10=75, the cumulative value of compressed air user 3 is 20+20+45+10+20+10=125, and if the average temperature is used as the representative outside temperature, it is 17.7 degrees. Also, if this data is recorded every second, the operating interval will be 10 seconds.

[0048] The compressed air usage model estimation unit 805 will be explained using Figure 13. Figure 13 shows an example of the configuration of the compressed air usage model estimation unit 805. The compressed air usage model estimation unit 805 consists of an outside temperature division process 1201 that creates history data 158A, 158B, and 158C for outside temperature division range setting values ​​1251A, 1251B, and 1251C within a predetermined range, based on history data 855 which is part of the registered data 853 included in the history database 108 and does not include abnormality prediction judgment data 854, and outside temperature division range setting values ​​1251; a compressed air usage model estimation unit 109 that calculates compressed air usage models 159A, 159B, and 159C using the history data 158A, 158B, and 158C as inputs; and a compressed air usage model group generation unit 1202 that outputs the compressed air usage models 159A, 159B, and 159C, which have been created to correspond to the outside temperature division range setting values ​​1251A, 1251B, and 1251C, as a compressed air usage model group 856. The historical data 158A, 158B, and 158C are in the same format as the historical data 158 described in Figure 1 of Example 1. Furthermore, the compressed air usage models 159A, 159B, and 159C are the same as the compressed air usage model 159 described in Figure 1 of Example 1. The ambient temperature division range setting value 1251 is a value that can be set by any method. For example, one could set the range from -10 degrees to 40 degrees in 1-degree or 0.5-degree increments, or one could arbitrarily define the temperature range and set it as 10 degrees or more to less than 12 degrees, 12 degrees or more to less than 15 degrees, or 15 degrees or more to less than 20 degrees.

[0049] Figure 14 illustrates the results of performing the outside temperature division process 1201 assuming the outside temperature division range setting value 1251 is set to 10 degrees or more and less than 12 degrees, 12 degrees or more and less than 15 degrees, and 15 degrees or more and less than 20 degrees, and the representative outside temperature values ​​included in the history data 855 are 16 degrees, 19 degrees, 12 degrees, 17 degrees, 18 degrees, and 11 degrees. Figure 14 shows an example of the outside temperature division process. When the outside temperature division range setting value 1251A is set to 10 degrees or more and less than 12 degrees, the history data 158A includes data for 12 degrees and 11 degrees. When the outside temperature division range setting value 1251B is set to 12 degrees or more and less than 15 degrees, the history data 158B will have no data. If the outside temperature division range setting value 1251C is set to 15 degrees Celsius or higher and less than 20 degrees Celsius, the historical data 158C will be 16 degrees Celsius, 19 degrees Celsius, 17 degrees Celsius, and 18 degrees Celsius. Note that in the examples in Figures 13 and 14, there are three outside temperature division range settings 1251, but it is not necessary to have three; you can set as many as you need.

[0050] The compressed air usage model estimation unit 109 is the same as the process described in Example 1, so it will be omitted here.

[0051] The compressed air usage model group 856 obtained by the compressed air usage model group generation unit 1202 will be explained using Figure 15. Figure 15 is a diagram showing the compressed air usage model group 856. A compressed air usage model 159 is stored for each ambient temperature division range setting value. In the example in Figure 15, the ambient temperature division range from 12°C to 13°C is: [Formula 6] Y = d1·X1+d2·X2+d3·X3+…+dn·Xn (Formula 6) The range from 13°C to 14°C is, [Formula 7] Y = e1·X1+e2·X2+e3·X3+…+en·Xn (Formula 7) It will become.

[0052] The anomaly prediction detection unit 806 will be explained using Figure 16. Figure 16 shows an example of the configuration of the abnormality prediction detection unit 806. The abnormality prediction detection unit 806 consists of a compressed air usage model selection unit 1501 that selects a compressed air usage model 159 based on a compressed air usage model group 856 and abnormality prediction judgment data 854, and an abnormality prediction detection unit 110 that determines an abnormality 160 based on the compressed air usage model 159 and the abnormality prediction judgment data 854.

[0053] The compressed air usage model selection unit 1501 extracts a compressed air usage model 159 corresponding to the ambient temperature division range of the compressed air usage model group 856, based on the ambient temperature information contained in the abnormality prediction data 854.

[0054] The abnormality prediction detection unit 110 is the same as the method described in Example 1.

[0055] By using the method described above, and by creating a compressed air consumption model that takes outside temperature into consideration and is less affected by the driving environment, it is possible to accurately estimate the operating interval based on the compressed air consumption model created by focusing on the period during which compressed air is used in the mobile vehicle (operating interval), and by comparing it with the actual operating interval, it becomes possible to identify abnormalities.

[0056] [Example 3] Figure 17 illustrates the abnormality prediction detection system for a mobile air system according to the present invention. Components identical to those in Example 1 or Example 2 are indicated with the same numbers as in Figure 2 or Figure 9, and their explanations are omitted. As with Figure 2, solid lines represent mechanical connections, and dotted lines represent the flow of information.

[0057] The system includes an air compressor 101 that produces compressed air, an air tank 102 that stores the compressed air, a plurality of compressed air users 103A, 103B, 103C, ... (hereinafter, "compressed air users 103" refers collectively to these) that use the compressed air stored in the air tank 102, an air compressor control device 104 that determines a control command 153 for the air compressor 101 based on the operating status data 151 of the air compressor 101 and the pressure information 152 of the air tank 102, and an air compressor control device 104 that measures ambient temperature data 851. A mobile air system system 1602 consists of a temperature sensor 801 and a formation characteristic database 1601 that stores formation characteristic information 1651, and a data collection device 16 that collects compressed air usage data 154A, 154B, 154C, ... of the compressed air usage devices 103A, 103B, 103C, ... (hereinafter, compressed air usage data 154 will be referred to collectively as these), operating status data 151 of the air compressor 101, ambient temperature data 851, and formation characteristic information 1651. 03, a data registration device 1604 that converts data 1652, which consists of operating status data 151 of the air compressor 101, compressed air usage data 154 of the compressed air user 103, ambient temperature data 851 and organization characteristic information 1651 stored in the data acquisition device 1603, into registration data 1653 and abnormality prediction data 1654 based on predetermined rules, a history database 108 that stores the registration data 1653, and the registration data 16 stored in the history database 108 The abnormality prediction detection system comprises a compressed air usage model estimation unit 1605 that estimates a group of compressed air usage models 1656 for the mobile air system 1602 based on historical data 1655, which is part of 53, and an abnormality prediction detection unit 1606 that determines an abnormality prediction 160 for the mobile air system 1602 by comparing the operating interval estimated from abnormality prediction determination data 1654 based on the compressed air usage model group 1656 with the actual operating interval included in the abnormality prediction determination data 1654.

[0058] Furthermore, similar to Example 1, the invention is not hindered even if any or all of the data acquisition device 1603, data registration device 1604, history database 108, compressed air usage model estimation unit 1605, and abnormality prediction detection unit 1606 are included in the mobile air system 1602. Also, the invention is not hindered even if the data acquisition device 1603, data registration device 1604, history database 108, compressed air usage model estimation unit 1605, and abnormality prediction detection unit 1606 are not independent devices but are included in a single integrated system.

[0059] The data registration device 1604 will be explained using Figure 18. Figure 18 shows an example of the configuration of the data registration device 1604. The system consists of: a section division process 201 that divides data 1652 into sections based on a predetermined rule to create divided data 1751; a cumulative value calculation process 202 that calculates cumulative values ​​252 (252A, 252B, 252C…) based on compressed air usage data 154 (154A, 154B, 154C…) used by compressed air users 103 (103A, 103B, 103C…) included in the divided data 1751; an outside temperature extraction process 901 that extracts representative outside temperatures 952 based on outside temperatures included in the divided data 1751; a train formation vehicle information process 1701 that extracts train formation characteristic information 1752 based on train formation characteristic information included in the divided data 1751; and a registration data creation process 1702 that combines the time, cumulative value 252, representative outside temperature 952, and train formation characteristic information 1752 included in the divided data 1751 to create registration data 1653 and abnormality prediction data 1654. The registration data 1653 and the anomaly prediction data 1654 are identical in data format.

[0060] Some of the historical data 1655 from the registered data 1653 is used by the compressed air usage model estimation unit 1605 to create a compressed air usage model, whereas the abnormality prediction data 1654 is not used to create a compressed air usage model, but is used to determine abnormalities in the mobile air system 1602.

[0061] The registration data 1653 and the anomaly prediction data 1654 are distributed by any method. For example, registration data 1653 could be data from the first month of use, and anomaly prediction data 1654 could be data from thereafter. When considering these methods, it is important to note that the history data 1655 mentioned above must not include the anomaly prediction data 1654. By not including the anomaly prediction data 1654 in the history data 1655, data indicating anomalies will not be mixed with the model creation data, allowing for more accurate model equation creation and anomaly prediction diagnosis. However, it is also possible to record the anomaly prediction data 1654 in the history database after it has been used to determine an anomaly, or to record it in the history database along with the distribution while ensuring it is not mixed with the history data. Alternatively, if it has been used to determine an anomaly, it may be included as part of the history data 1655. The specific details of registration data 1653 will be described later.

[0062] Data 1652 consists of time-series data and shall include at least the compressed air consumption of compressed air device 1, compressed air consumption of compressed air device 2, compressed air consumption of compressed air device 3, the operating status of the air compressor, the outside temperature, and information on the composition characteristics. The recording time for each data point may also be included.

[0063] The division method performed in the segment division process 201 is the same as in Example 1, if the collected data 155 is replaced with data 1652 and the divided data 251 is replaced with divided data 1751. However, the data content differs from Example 1 in that data 1652 and divided data 1751 include train formation characteristic information and outside temperature. Note that the data format of data 1652 and divided data 1751 is the same. Figure 19 shows an example of divided data 1751.

[0064] The train formation information processing 1701 will now be explained. If the train formation characteristic information in the time-series data included in the divided data 1751 is all the same and assigned to A, then the train formation characteristic information 1752 will output A, which is this identical train formation characteristic information. On the other hand, if the data is not the same, then the train formation characteristic information 1752 will output "no train formation information". In the example in Figure 19, the train formation characteristic information is the same 3M2T, so the train formation characteristic information 1752 will output 3M2T. If any of the data in Figure 19 were not 3M2T, then the train formation characteristic information 1752 will output "no data".

[0065] The registration data 1653, which is created in the registration data creation process 1702 and registered in the history database 108, will be explained using Figure 20. Figure 20 shows an example of registered data 1653. Registered data 1653 consists of the cumulative value of compressed air user 1, the cumulative value of compressed air user 2, the cumulative value of compressed air user 3, the representative outside temperature 952, the train configuration characteristic information 1651, and the operating interval obtained when the data is divided into sections by the section division process 201. In the case of the divided data 1751 shown in Figure 19, the cumulative value of compressed air user 1 is 15+25+25=65, the cumulative value of compressed air user 2 is 20+15+30+10=75, the cumulative value of compressed air user 3 is 20+20+45+10+20+10=125, the average temperature is 17.7 degrees if the representative outside temperature is used, and the train configuration characteristic information is uniformly 3M2T, so it becomes 3M2T. Also, if this data is recorded every second, the operating interval will be 10 seconds.

[0066] The compressed air usage model estimation unit 1605 will be explained using Figure 21. Figure 21 shows an example of the configuration of the compressed air usage model estimation unit 1605. The compressed air usage model estimation unit 1605 consists of a configuration characteristic information splitting process 2001 that creates history data 855A, 855B, and 855C for configuration characteristic information 2051A, 2051B, and 2051C based on history data 1655, which is part of the registered data 1653 included in the history database 108, and configuration characteristic information 1752 included in the history data 1655; a compressed air usage model estimation unit 805 that calculates compressed air usage model groups 856A, 856B, and 856C using the history data 855A, 855B, and 855C as input; and a compressed air usage model group generation unit 2002 that outputs the compressed air usage model groups 856A, 856B, and 856C for configuration characteristic information 2051A, 2051B, and 2051C together as a compressed air usage model group 1656.

[0067] The organization characteristic information partitioning process 2001 will be explained using Figure 22. Figure 22 shows an example of the train formation characteristic information division process. Figure 22 illustrates the results of performing the division of train formation characteristic information 2051 assuming that the train formation characteristic information 2051 is set to 3M2T, 4M4T, and 2M2T, and that the train formation characteristics included in the history data 1655 are 3M2T, 4M4T, 2M2T, 3M2T, 4M4T, and 2M2T. When train formation characteristic information 2051A is set to 3M2T, history data 158A includes data corresponding to 3M2T. When train formation characteristic information 2051B is set to 4M4T, history data 158B includes data corresponding to 4M4T. When train formation characteristic information 2051C is set to 2M2T, history data 158C includes data corresponding to 2M2T. Note that in the examples in Figures 21 and 22, there are three train formation characteristic information entries, but it is not necessary to have three; you can set as many as needed.

[0068] The ambient temperature division process 1201 is the same as the process described in Example 2, and the compressed air usage model estimation unit 109 is the same as the process described in Example 1, so these are omitted.

[0069] The compressed air usage model group 1656 obtained by the compressed air usage model group generation unit 2002 will be explained using Figure 23. Figure 23 is a diagram showing the compressed air usage model group. A compressed air usage model 159 is stored for each set of organization characteristic information and ambient temperature division range setting value. In the example in Figure 23, the set of organization characteristic information is 3M2T, and the ambient temperature division range is from 12°C to 13°C. [Formula 8] Y = d1·X1+d2·X2+d3·X3+…+dn·Xn (Formula 8) The formation characteristics information is 3M2T, and the temperature range is from 13°C to 14°C. [Formula 9] Y = e1·X1+e2·X2+e3·X3+…+en·Xn (Equation 9) It will become.

[0070] The anomaly prediction detection unit 1606 will be explained using Figure 24. Figure 24 shows an example of the configuration of the abnormality prediction detection unit. The abnormality prediction detection unit 1606 consists of a compressed air usage model selection unit 2301 that selects a compressed air usage model 159 based on a group of compressed air usage models 1656 and abnormality prediction judgment data 1654, and an abnormality prediction detection unit 110 that determines an abnormality 160 based on the compressed air usage model 159 and the abnormality prediction judgment data 1654. The compressed air usage model selection unit 2301 extracts the compressed air usage model 159 corresponding to the train configuration characteristics information and outside temperature division range from the compressed air usage model group 1656, based on the train configuration characteristics information and outside temperature information contained in the abnormality prediction data 1654.

[0071] The abnormality prediction detection unit 110 is the same as the method described in Example 1.

[0072] Using the method described above, a compressed air consumption model that takes into account different train configurations can be created. Based on the compressed air consumption model created by focusing on the period during which compressed air is used in the mobile unit (operating interval), the operating interval can be estimated and compared with the actual operating interval, making it possible to determine abnormalities appropriate for each train configuration.

[0073] [Example 4] Figure 25 shows an abnormality prediction system for a mobile air system in Embodiment 4 of the present invention. Components identical to those in Embodiment 1, Embodiment 2, or Embodiment 3 are indicated with the same numbers as in Figure 2, Figure 9, or Figure 17, and their explanations are omitted. As with Figure 2, solid lines represent mechanical connections, and dotted lines represent the flow of information.

[0074] The system comprises an air compressor 101 that produces compressed air, an air tank 102 that stores the compressed air, a plurality of compressed air users 103A, 103B, 103C, ... (hereinafter, "compressed air users 103" refers collectively to these) that use the compressed air stored in the air tank 102, an air compressor control device 104 that determines a control command 153 for the air compressor 101 based on the operating status data 151 of the air compressor 101 and the pressure information 152 of the air tank 102, and a temperature sensor that measures ambient temperature data 851. A mobile air system system 1602 consists of a SA 801 and a formation characteristics database 1601 that stores formation characteristics information 1651, and a data acquisition device 240 that collects compressed air usage data 154A, 154B, 154C, ... of the compressed air usage devices 103A, 103B, 103C, ... (hereinafter, compressed air usage data 154 will be a collective term for these), operating status data 151 of the air compressor 101, ambient temperature data 851, formation characteristics information 1651, and pressure information 152. 1, a data registration device 2402 that converts data 2451, which consists of operating status data 151 of the air compressor 101, compressed air usage data 154 of the compressed air user 103, ambient temperature data 851, organization characteristic information 1651, and pressure information 152 stored in the data acquisition device 2401, into registration data 2452 and abnormality prediction data 2453 based on predetermined rules, a history database 108 that stores the registration data 2452, and the registered data stored in the history database 108 The abnormality prediction detection system comprises a compressed air usage model estimation unit 2403 that estimates a group of compressed air usage models 2455 for the mobile air system 1602 based on historical data 2454, which is part of data 2452, and an abnormality prediction detection unit 2404 that determines an abnormality prediction 160 for the mobile air system 1602 by comparing the operating interval estimated from abnormality prediction determination data 2453 based on the compressed air usage model group 2455 with the actual operating interval included in the abnormality prediction determination data 2453.

[0075] Furthermore, similar to Example 1, the invention is not hindered even if any or all of the data acquisition device 2401, data registration device 2402, history database 108, compressed air usage model estimation unit 2403, and abnormality prediction detection unit 2404 are included in the mobile air system 1602. Also, the invention is not hindered even if the data acquisition device 2401, data registration device 2402, history database 108, compressed air usage model estimation unit 2403, and abnormality prediction detection unit 2404 are not independent devices but are included in a single integrated system.

[0076] The data registration device 2402 will be explained using Figure 26. Figure 26 shows an example of the configuration of the data registration device 2402. Based on a predetermined rule, the data 2451 is divided into sections, and divided into sections 2551 (section division process 201). Based on the compressed air usage data 154 (154A, 154B, 154C…) used by the compressed air usage device 103 (103A, 103B, 103C…), a cumulative value calculation process 202 calculates a cumulative value 252 (252A, 252B, 252C…) for the divided data 2551. Based on the outside temperature information contained in the divided data 2551, a representative outside temperature 952 is extracted (outside temperature extraction process 901). The system consists of a train formation vehicle information processing 1701 that extracts train formation characteristic information 1752 based on the train formation characteristic information contained in Ta2551, a pressure fluctuation calculation processing 2501 that extracts pressure fluctuations 2552 in the data based on the pressure information 152 contained in the divided data 2551, and a registration data creation processing 2502 that combines the time, cumulative value 252, representative outside temperature 952, train formation characteristic information 1752, and pressure fluctuations 2552 contained in the divided data 2551 to create registration data 2452 and abnormality prediction judgment data 2453.

[0077] Some of the historical data 2454 from the registered data 2452 is used by the compressed air usage model estimation unit 2403 to create a compressed air usage model, whereas the abnormality prediction data 2453 is not used to create a compressed air usage model, but is used to determine abnormalities in the mobile air system 1602.

[0078] The registration data 2452 and the anomaly prediction data 2453 are distributed by any method. For example, registration data 2452 could be data from the start of use up to one month, and anomaly prediction data 2453 could be data from thereafter. When considering these methods, it is important to note that the history data 2454 mentioned above must not include the anomaly prediction data 2453. By not including the anomaly prediction data 2453 in the history data 2454, data indicating anomalies will not be mixed with the model creation data, allowing for more accurate model equation creation and anomaly prediction diagnosis. However, it is also possible to record the anomaly prediction data 2453 in the history database after it has been used to determine an anomaly, or to record it in the history database along with the distribution while keeping it separate from the history data. Alternatively, if it has been used to determine an anomaly, it may be included as part of the history data 2454. The specific contents of registration data 2452 will be described later.

[0079] Data 2451 consists of time-series data and shall include at least the compressed air consumption of compressed air device 1, compressed air consumption of compressed air device 2, compressed air consumption of compressed air device 3, the operating status of the air compressor, ambient temperature, composition characteristics information, and pressure fluctuations. The recording time for each data point may also be included.

[0080] The division method performed in the interval division process 201 is the same as in Figure 5 of Example 1, if the collected data 155 is replaced with data 2451 and the divided data 251 is replaced with divided data 2551. However, the data content differs from Example 1 in that data 2451 and divided data 2551 include pressure fluctuations, formation characteristic information, and outside temperature. Note that the data format of data 2451 and divided data 2551 is the same. Figure 27 shows an example of divided data 2551.

[0081] The pressure fluctuation calculation process 2401 is explained below. The pressure fluctuation 2552 is calculated using the absolute value of the difference between the pressure at the start (when the air compressor has finished operating) and the pressure at the end (just before the air compressor starts operating), which are contained in one data point 2451.

[0082] The registration data 2452, which is created in the registration data creation process 2502 and registered in the history database 108, will be explained using Figure 28. Figure 28 shows an example of registered data 2452. Registered data 2452 consists of the cumulative value of compressed air user 1, the cumulative value of compressed air user 2, the cumulative value of compressed air user 3, the representative outside temperature 952, the configuration characteristic information 1651, the pressure fluctuation 2552, and the operating interval obtained when the data is divided into sections by the section division process 201. In the case of the divided data 2551 shown in Figure 26, the cumulative value of compressed air user 1 is 15+25+25=65, the cumulative value of compressed air user 2 is 20+15+30+10=75, the cumulative value of compressed air user 3 is 20+20+45+10+20+10=125, the average temperature is 17.7 degrees if the representative outside temperature is used, the configuration characteristic information is 3M2T as it is uniformly 3M2T, and the pressure fluctuation is 820-655=165. Furthermore, if this data is recorded every second, the operating interval would be 10 seconds.

[0083] The compressed air usage model estimation unit 2403 will be explained using Figure 29. Figure 29 shows an example of the configuration of the compressed air usage model estimation unit 2403. The compressed air usage model estimation unit 2403 consists of a train formation characteristic information division process 2001 that creates data 2851A, 2851B, and 2851C for train formation characteristic information 2051A, 2051B, and 2051C based on train formation characteristic information 1752 contained in history data 2454, which is part of the registered data 2452 contained in history database 108; compressed air usage model estimation units 2801A, 2801B, and 2801C that calculate compressed air usage models 2852A, 2852B, and 2852C as their respective outputs, using data 2851A, 2851B, and 2851C as their respective inputs; and a compressed air usage model group generation unit 2802 that outputs the compressed air usage models 2852A, 2852B, and 2852C for train formation characteristic information 2051A, 2051B, and 2051C together as a compressed air usage model group 2455.

[0084] The formation characteristic information division process 2001 is similar to that shown in Figure 22, but differs from the historical data 855A, 855B, and 855C in that the historical data 2454 provided as input includes pressure fluctuations 2552, and the data 2851A, 2851B, and 2851C obtained by the formation characteristic information division process 2001 also include pressure fluctuations 2552.

[0085] The compressed air usage model estimation unit 2801 will be explained. Based on the historical data 2454, [Formula 10] Zj = X[1,j] · G1+X[2,j] · G2+X[3,j] · G3+…+X[n,j] · Gn+Yj · b (Equation 10) This is created according to the number of Zj(j=1,2,…,m). Zj(j=1,2,…,m) represents the pressure fluctuation during the operating interval, and Yj(j=1,2,…,m) represents the operating interval. b represents the air leakage rate per unit time in the air piping. X[i,j](i=1,2,…,n)(j=1,2,…,m) is a parameter relating to the cumulative amount in each compressed air user during the operating interval Yj(j=1,2,…,m). The multiple obtained (Equation 10) are treated as a system of simultaneous equations, and the coefficient Gi(i=1,2,…,n) for each compressed air user is calculated using methods such as the least squares method. This allows us to calculate the coefficient Gi(i=1,2,…,n) for each compressed air user. Z = G1·X1+G2·X2+G3·X3+…+Gn·Xn + b·Y (Formula 11) This is obtained. Transforming this so that the operating interval Y is on the left side (Equation 12) results in compressed air usage model 2852. [Formula 12] Y= (Z - (G1·X1+G2·X2+G3·X3+…+Gn·Xn )) / b (Equation 12)

[0086] The compressed air usage model group 2455 obtained by the compressed air usage model group generation unit 2802 will be explained using Figure 30. Figure 30 is a diagram showing an example of the compressed air usage model group 2455. A compressed air usage model 2852 is stored for each set of organization characteristic information and ambient temperature division range setting value. In the example of Figure 30, the set of organization characteristic information is 3M2T, and the ambient temperature division range is from 12°C to 13°C. [Formula 13] Y =(Z - (d1·X1+d2·X2+d3·X3+…+dn·Xn)) / b (Equation 13) The formation characteristics information is 3M2T, and the temperature range is from 13°C to 14°C. [Formula 14] Y =(Z - (e1·X1+e2·X2+e3·X3+…+en·Xn)) / b (Equation 14) It will become.

[0087] The anomaly prediction detection unit 2404 will be explained using Figure 31. Figure 31 shows an example of the configuration of the anomaly prediction detection unit 2404. The anomaly prediction detection unit 2404 consists of a compressed air usage model selection unit 3001 that selects a compressed air usage model 159 based on a compressed air usage model group 2455 and anomaly prediction judgment data 2452, and an anomaly prediction detection unit 2404 that determines an anomaly 160 based on a compressed air usage model 2852 and anomaly prediction judgment data 2453.

[0088] The compressed air usage model selection unit 3001 extracts the compressed air usage model 2852 corresponding to the train configuration characteristics information and outside temperature division range from the compressed air usage model group 2455, based on the train configuration characteristics information and outside temperature information contained in the abnormality prediction data 2453.

[0089] The abnormality prediction detection unit 110 is the same as the method described in Example 1.

[0090] Using the method described above, a compressed air consumption model can be created that takes into account conditions with different pressure fluctuations. Based on the compressed air consumption model created by focusing on the period during which compressed air is used in the mobile vehicle (operating interval), the operating interval can be estimated and compared with the actual operating interval to determine if an anomaly has occurred.

[0091] [Example 5] In Examples 1 to 4, the time interval for the interval division, i.e., the time related to operation, is defined as the operating interval, which is the time when the air compressor is not operating, and the time from the end of operation of the air compressor to the end of operation for the next time, or from the start of operation of the air compressor to the start of operation for the next time. Operation While the period was used as an example in the previous explanation, in this embodiment, the operating time, which is the time the air compressor is running, will be used as an example. Note that components identical to those in Example 1 are indicated with the same numbers as in Figure 1, and their explanations are omitted.

[0092] Figure 32 illustrates the abnormality prediction detection system for a mobile air system according to the present invention. A mobile air system 105 is comprised of an air compressor 101 that produces compressed air, an air tank 102 that stores the compressed air, a plurality of compressed air users 103A, 103B, 103C, ... (hereinafter, "compressed air users 103" refers to these collectively) that use the compressed air stored in the air tank 102, an air compressor control device 104 that determines a control command 153 for the air compressor 101 based on the operating status data 151 of the air compressor 101 and the pressure information 152 of the air tank 102, compressed air usage data 154A, 154B, 154C, ... (hereinafter, "compressed air usage data 154" refers to these collectively) of the compressed air users 103A, 103B, 103C, ..., a data acquisition device 106 that collects the operating status data 151 of the air compressor 101, and the air stored in the data acquisition device 106 The abnormality prediction detection system comprises a data registration device 3101 that converts collected data 155, consisting of operating status data 151 of the compressor 101 and compressed air usage data 154 of the compressed air user 103, into registered data 156 and abnormality prediction data 3151 based on predetermined rules; a history database 108 that stores the registered data 156; a compressed air usage model estimation unit 109 that estimates the compressed air usage model 159 of the mobile air system 105 based on the history data 158, which is part of the registered data 156 stored in the history database 108; and an abnormality prediction detection unit 3102 that determines an abnormality 160 of the mobile air system 105 by comparing the operating time estimated from the abnormality prediction data 3151 based on the compressed air usage model 159 with the actual operating time included in the abnormality prediction data 3151. In Figure 32, solid lines represent mechanical connections, and dotted lines represent the flow of information.

[0093] The data registration device 3101 will be explained using Figure 33. Figure 33 shows an example of the configuration of the data registration device 3101. It consists of a section division process 3201 that divides the collected data 155 into sections based on a predetermined rule, resulting in divided data 251 and divided data 3251; a cumulative value calculation process 3202 that calculates cumulative values ​​252 (252A, 252B, 252C…) based on the compressed air usage data 154 (154A, 154B, 154C…) used by the compressed air usage device 103 (103A, 103B, 103C…) included in the divided data 251 and divided data 3251; and a registration data creation process 3203 that combines the time included in the divided data 251 and the cumulative value 252 to create registration data 156, and combines the time included in the divided data 3251 and the cumulative value 3252 to create abnormality prediction judgment data 3151. The registration data 156 is data created based on the operating interval, while the abnormality prediction judgment data 3151 is data created based on the operating time.

[0094] The definition of operating time is shown using Figure 34. Figure 34 shows the operating status of an air compressor and the instantaneous and total air consumption of devices using compressed air. The horizontal axis represents time, and the vertical axis, from top to bottom, shows the operating status of the air compressor, the instantaneous compressed air consumption, and the total air consumption during the operating interval. The operating time is the time from when the air compressor starts operating (t2) to when it stops operating (t3). Therefore, at t3, which marks the end of the operating time, the total air consumption is 0.

[0095] An example of segmentation is shown using Figure 35. Figure 35 is a diagram showing the operating status of the air compressor and the amount of compressed air used by each compressed air user. Figure 35 shows the case where there are three compressed air users. The collected data 155 and segmented data 251 and segmented data 3251 consist of time-series data and include at least the amount of compressed air used by compressed air user 1, the amount of compressed air used by compressed air user 2, the amount of compressed air used by compressed air user 3, and the operating status of the air compressor. Collected data 155 is data that includes the entire measured time, while segmented data 251 is data that is segmented by the operating interval of the air compressor, and segmented data 3251 is data that is segmented by the operating time of the air compressor. In this figure, three sets of segmented data 251 and three sets of segmented data 3251 are obtained.

[0096] The cumulative value calculation process 3202 obtains a cumulative value 252 by summing the usage amounts of each compressed air user during the operating intervals in the divided data 251, and a cumulative value 3252 by summing the usage amounts of each compressed air user during the operating time in the divided data 3251.

[0097] The abnormality prediction data 3151, created in the registration data creation process 3203 and registered in the history database 108, will be explained using Figure 36. Note that the registration data 156 is the same as that described in Example 1.

[0098] Figure 36 shows an example of abnormality prediction data 3151. The abnormality prediction data 3151 consists of the cumulative value of compressed air user 1, the cumulative value of compressed air user 2, the cumulative value of compressed air user 3, and the operating time obtained when the data is divided into intervals by the interval division process 3201. In the case of the divided data 3251 shown in Figure 35, the cumulative value of compressed air user 1 is 110, the cumulative value of compressed air user 2 is 15, and the cumulative value of compressed air user 3 is 20. Furthermore, if this data is recorded every second, the operating time will be 10 seconds.

[0099] The abnormality prediction detection unit 3102 calculates the operating time based on the compressed air usage model 159 and determines the abnormality prediction 160 by comparing it with the actual operating time. The method for calculating the operating time based on the compressed air usage model 159 will be explained below.

[0100] The estimated operating time W is calculated using the coefficient ai (i=1,2,…,n) of the compressed air usage model 159 described in (Equation 2) of Example 1. [Formula 15] W = S + a1·X1+a2·X2+a3·X3+…+an·Xn (Formula 15) This is the result. Note that S represents the average operating time when compressed air is not being used in a compressed air-using device. This value can be quantitatively calculated using the following formula. [Formula 16] S=60×MP×TL / (0.1013×Q) (Equation 16) MP represents the maximum pressure of the air compressor [MPa], TL represents the capacity of the air tank [L], and Q represents the compressed air discharge rate [L / min]. Also, 0.1013 represents atmospheric pressure [MPa].

[0101] The processing results of the anomaly prediction detection unit 3102 will be explained using Figure 37. Figure 37 shows an example where the horizontal axis represents elapsed time and the vertical axis represents operating time. It compares the estimated operating time (dotted line) calculated by the anomaly prediction detection unit 3102 with the actual operating time (solid line), and detects an anomaly when the actual operating time exceeds a threshold compared to the estimated operating time. By making this determination, an anomaly can be identified. Note that an anomaly may be judged not only based on a single detection, but also if multiple detections occur within a certain time range, or if multiple detections occur consecutively.

[0102] By using the method described above, it becomes possible to predict abnormalities by estimating the operating time based on a compressed air consumption model created by focusing on the period during which compressed air is used in the mobile vehicle (operating interval), and comparing it with the actual operating time.

[0103] Although this embodiment was explained based on the compressed air usage model described in Example 1, the content of the present invention is not hindered even if the compressed air usage models described in Examples 2 to 4 are used.

[0104] [Example 6] The methods described in Examples 1 to 5 are methods for detecting anomalies. In addition, a method for identifying the location of anomalies is presented.

[0105] If an abnormality warning is detected in any of the compressed air usage models from Examples 1 to 5, the compressed air usage model is recreated using the data containing the abnormality warning. This is achieved by comparing the coefficients of the recreated compressed air usage model with those of the original compressed air usage model, thereby setting the priority of the abnormality warning. A specific example is described below.

[0106] Figure 38 shows an abnormality prediction detection system for a mobile air system in Embodiment 6 of the present invention. In addition to the mobile air system in Figure 2 described in Embodiment 1, the system includes a data synthesis unit 3701 that synthesizes abnormality prediction determination data 157 and history data 158 to create data 3751, a compressed air usage model re-estimation unit 3702 that takes data 3751 as input and outputs a compressed air usage model 3752, and an abnormality prediction device presentation unit 3703 that takes compressed air usage model 3752, compressed air usage model 159, and abnormality prediction 160 as input and outputs an abnormality prediction device sequence 3753. The function of the compressed air usage model re-estimation unit 3702 is the same as that of the compressed air usage model estimation unit 109, so its explanation is omitted.

[0107] The abnormality prediction device display unit 3703 will now be explained. If the abnormality prediction 160 is normal, the abnormality prediction device display unit 3703 will not output anything as the abnormality prediction device sequence 3753. If the fault prediction 160 is an abnormality prediction, the compressed air usage model 159, which is the model before creation, will be displayed. [Formula 17] Y = a1·X1+a2·X2+a3·X3+…+an·Xn (Equation 17) This is the recreated compressed air-using model 3752. [Formula 18] Y = k1·X1+k2·X2+k3·X3+…+kn·Xn (Equation 18) The coefficient ratios ki / ai (i=1,2,…,n) are compared, and the order of the abnormality prediction devices for compressed air users is output in descending order of this value (3753). Alternatively, the failure probability for each compressed air user can be used, and the order of the abnormality prediction devices (3753) can be output in descending order of probability. Furthermore, the order of the abnormality prediction devices for compressed air users can be output in descending order of the coefficient ratio ki / ai × failure probability of compressed air user i (i=1,2,…,n) (3753).

[0108] Based on the above, when an abnormality is detected, it becomes possible to indicate the sequence of compressed air-using devices that are most likely to be causing the abnormality.

[0109] Although this embodiment was described based on the mobile air system for mobile bodies described in Example 1, the use of the mobile air system for mobile bodies described in Examples 2 to 5 does not impede the content of the present invention. [Explanation of Symbols]

[0110] 101: Air compressor, 102: Air tank, 103, 103A, 103B, 103C: Compressed air usage device, 104: Air compressor control device, 105: Mobile air system, 106: Data acquisition device, 107: Data registration device, 108: History database, 109: Compressed air usage model estimation unit, 110: Anomaly prediction detection unit, 151: Air compressor operating status data, 152: Air tank pressure information, 153: Control command for the air compressor, 154, 154A, 154B, 154C: Compressed air usage data for the compressed air usage device, 155 :Collected data, 156:Registered data, 157:Data for anomaly prediction determination, 158, 158A, 158B, 158C:History data, 159, 159A, 159B, 159C:Compressed air usage model, 160:Anomaly prediction, 201:Interval division processing, 202:Cumulative value calculation processing, 203:Registered data creation processing, 251:Divided data, 252, 252A, 252B, 252C:Cumulative value, 801:Temperature sensor, 802:Mobile air system, 803:Data acquisition device, 804:Data registration device, 805:Compressed air usage model estimation unit, 806: Anomaly prediction detection unit, 851: outside temperature data, 852: collected data, 853: registered data, 854: data for anomaly prediction determination, 855: history data, 856: compressed air usage model group, 901: outside temperature extraction process, 902: registered data creation process, 951: divided data, 952: representative outside temperature, 1201: outside temperature division process, 1202: compressed air usage model group generation unit, 1251, 1251A, 1251B, 1251C: outside temperature division range setting value, 1501: compressed air usage model selection unit, 1601: organization characteristics database, 1602: mobile air system Unified system, 1603: Data acquisition device, 1604: Data registration device, 1605: Compressed air usage model estimation unit, 1606: Anomaly prediction detection unit, 1651: Train formation characteristic information, 1652: Data, 1653: Registered data, 1654: Data for anomaly prediction determination, 1655: History data, 1656: Compressed air usage model group, 1701: Train formation vehicle information processing, 1702: Registered data creation processing, 1751: Divided data, 1752: Train formation characteristic information, 2001: Train formation characteristic information division processing, 2002: Compressed air usage model group generation unit, 2051A, 2051B,2051C: Formation characteristics information, 2301: Compressed air usage model selection unit, 2401: Data acquisition device, 2402: Data registration device, 2403: Compressed air usage model estimation unit, 2404: Anomaly prediction detection unit, 2451: Data, 2452: Registered data, 2453: Data for anomaly prediction determination, 2454: History data, 2455: Compressed air usage model group, 2501: Pressure fluctuation calculation processing, 2502: Registered data creation processing, 2551: Divided data, 2552: Pressure fluctuation, 2801A, 2801B, 2801C: Compressed air usage model estimation unit, 2802: Compressed air usage model group generation The order of operations is as follows: 2851A, 2851B, 2851C: Data, 2852A, 2852B, 2852C: Compressed air usage model, 3001: Compressed air usage model selection unit, 3101: Data registration device, 3102: Anomaly prediction detection unit, 3151: Data for anomaly prediction determination, 3201: Interval division processing, 3202: Cumulative value calculation processing, 3203: Registration data creation processing, 3251: Divided data, 3701: Data synthesis unit, 3702: Compressed air usage model re-estimation unit, 3703: Anomaly prediction device presentation unit, 3751: Data, 3752: Compressed air usage model, 3753: Anomaly prediction device.

Claims

1. An abnormality prediction detection system for detecting abnormalities in a mobile air system, comprising an air compressor that produces compressed air, an air tank that stores the compressed air, and a compressed air usage device that uses the compressed air stored in the air tank, A data collection unit that collects historical data in a mobile air system, An abnormality prediction detection unit detects an abnormality in the mobile air system by comparing the estimated operating time of the air compressor with the actual operating time of the air compressor, based on a compressed air usage model that shows a relationship between the operating time of the air compressor and the amount of air used by the compressed air user, calculated from the historical data collected by the data acquisition unit. An anomaly prediction detection system characterized by having the following features.

2. An anomaly prediction detection system according to claim 1, The historical data includes at least time data relating to the operation of the air compressor and compressed air consumption data of the compressed air user. An anomaly prediction detection system characterized by the following features.

3. An anomaly prediction detection system according to claim 2, The aforementioned time data relating to operation includes at least one of the following: data relating to the operating interval from the end of operation of the air compressor to the start of operation again, data relating to the operating cycle from the end of operation of the air compressor to the end of operation again, or data relating to the operating time of the air compressor. The compressed air usage data is data relating to the amount of compressed air used by the compressed air user during the operating time. An anomaly prediction detection system characterized by the following features.

4. An abnormality prediction detection system for detecting abnormalities in a mobile air system, comprising an air compressor that produces compressed air, an air tank that stores the compressed air, and a compressed air usage device that uses the compressed air stored in the air tank, A data collection unit that collects historical data in a mobile air system, An abnormality prediction detection unit detects an abnormality in the mobile air system by comparing the amount of compressed air used by the compressed air user, which is estimated based on a compressed air usage model that shows a relationship between the operating interval of the air compressor and the amount of compressed air used by the compressed air user calculated from the historical data collected by the data acquisition unit, with the actual amount of compressed air used. An anomaly prediction detection system characterized by having the following features.

5. An anomaly prediction detection system according to claim 1, The system further includes an air usage model creation unit that uses the historical data collected by the data acquisition unit to create a compressed air usage model for the mobile air system. An anomaly prediction detection system characterized by the following features.

6. An anomaly prediction detection system according to claim 3, The aforementioned historical data further includes ambient temperature data for ambient temperature measured by the mobile air system, The abnormality prediction detection unit detects abnormalities in the mobile air system based on the compressed air usage model of the mobile air system calculated from the historical data, including the outside temperature data. An anomaly prediction detection system characterized by the following features.

7. An anomaly prediction detection system according to claim 3, The historical data further includes organizational characteristics data relating to the organizational characteristics of the mobile air system, The abnormality prediction detection unit detects abnormalities in the mobile air system based on the compressed air usage model of the mobile air system calculated for each configuration characteristic using the historical data, including the configuration characteristic data. An anomaly prediction detection system characterized by the following features.

8. An anomaly prediction detection system according to claim 3, The historical data further includes pressure information data of the air tank mounted on the mobile air system, The abnormality prediction detection unit detects abnormalities in the mobile air system based on the compressed air usage model of the mobile air system calculated from the historical data, including the pressure information data of the air tank. An anomaly prediction detection system characterized by the following features.

9. An anomaly prediction detection system according to any one of claims 3, 5, 6, 7, or 8, The abnormality prediction unit detects abnormalities in the mobile air system based on a compressed air usage model calculated by assuming a constant amount of air leakage in the mobile air system. An anomaly prediction detection system characterized by the following features.

10. An anomaly prediction detection system according to claim 1, The abnormality prediction detection unit outputs an abnormality prediction when it has measured a predetermined number of times an event occurs where the difference between the estimated value estimated based on the compressed air usage model and the actual measured value exceeds a threshold. An anomaly prediction detection system characterized by the following features.

11. An anomaly prediction detection system according to claim 10, The abnormality prediction detection unit estimates the operating time of the air compressor based on the compressed air usage model of the mobile air system, and outputs an abnormality prediction when the difference between the actual operating time and the estimated operating time exceeds a threshold a predetermined number of times. An anomaly prediction detection system characterized by the following features.

12. An anomaly prediction detection system according to claim 11, When outputting an anomaly prediction, the compressed air usage model re-estimation unit recreates the compressed air usage model, including the data in which the anomaly prediction was detected. An abnormality prediction device display unit extracts and outputs the sequence of compressed air usage devices that are highly likely to have abnormality indicators by comparing the recreated compressed air usage model with the compressed air usage model before recreation. An anomaly prediction detection system characterized by having the following features.

13. An anomaly prediction detection system according to claim 12, The abnormality prediction device display unit extracts a sequence of compressed air users that are highly likely to exhibit abnormality by considering the probability of failure of the compressed air users. An anomaly prediction detection system characterized by the following features.

14. An air system system comprising an air system that is installed in a railway vehicle to create and use compressed air, and an abnormality prediction detection system that detects signs of abnormality in the air system, The aforementioned air system is An air compressor that produces compressed air, An air tank for storing the compressed air, A brake device that uses compressed air stored in the aforementioned air tank, a compressed air usage device including an air spring, The air piping connecting the air compressor, the air tank, and the compressed air usage device, An air compressor control device controls the air compressor based on the operating information of the air compressor and the pressure information of the air tank. Composed of, The aforementioned abnormality prediction detection system is A data acquisition device that collects operating information of the air compressor, compressed air consumption of the compressed air user, ambient temperature, configuration characteristics, and pressure fluctuation information of the air tank, A data registration device that separates the information collected by the data acquisition device into registration data used for creating model equations and anomaly detection data used for comparing with estimated values ​​as measured values, in order to avoid duplication, and converts it into a user-specified data format. A history database for storing the aforementioned registration data, A compressed air usage model estimation unit creates a compressed air usage model that shows a relationship between the time related to operation, including the operating interval, operating cycle, or operating time of the air compressor, and the compressed air usage of the compressed air usage device, from registered data including operation information, compressed air usage, ambient temperature, configuration characteristics, and tank pressure fluctuation information stored in the aforementioned history database. An abnormality prediction unit determines that there is an abnormality in the air system and outputs the determination result if the difference between the time / compressed air consumption for predicted operation estimated from the compressed air usage model and the time / compressed air consumption for actual operation exceeds a threshold. An entire air system characterized by being composed of the following.

15. An anomaly prediction detection system according to claim 4, The system further includes an air usage model creation unit that uses the historical data collected by the data acquisition unit to create a compressed air usage model for the mobile air system. An anomaly prediction detection system characterized by the following features.

16. An anomaly prediction detection system according to claim 6, The historical data further includes organizational characteristics data relating to the organizational characteristics of the mobile air system, The abnormality prediction detection unit detects abnormalities in the mobile air system based on the compressed air usage model of the mobile air system calculated for each configuration characteristic using the historical data, including the configuration characteristic data. An anomaly prediction detection system characterized by the following features.

17. An anomaly prediction detection system according to claim 6, The historical data further includes pressure information data of the air tank mounted on the mobile air system, The abnormality prediction detection unit detects abnormalities in the mobile air system based on the compressed air usage model of the mobile air system calculated from the historical data, including the pressure information data of the air tank. An anomaly prediction detection system characterized by the following features.

18. An anomaly prediction detection system according to claim 7, The historical data further includes pressure information data of the air tank mounted on the mobile air system, The abnormality prediction detection unit detects abnormalities in the mobile air system based on the compressed air usage model of the mobile air system calculated from the historical data, including the pressure information data of the air tank. An anomaly prediction detection system characterized by the following features.

19. An anomaly prediction detection system according to any one of claims 15, 16, 17, or 18, The abnormality prediction unit detects abnormalities in the mobile air system based on a compressed air usage model calculated by assuming a constant amount of air leakage in the mobile air system. An anomaly prediction detection system characterized by the following features.

Images