JP2018119812A - Fault detection device, fault detection method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a fault detection device, a fault detection method and a program which make it easier to enhance the accuracy of fault detection of a target device and can easily compare the degree of faults.SOLUTION: Provided are an acquiring unit 11 which acquires output information representing a feature amount of a determination target used for determination, an analysis unit 12 which calculates an average value and a variation based on a standard deviation in a plurality of predetermined periods by acquiring a standard deviation of feature quantities in a predetermined period based on output information, a storage unit 13 which stores an average value and a variation of standard deviations obtained for each of a plurality of other determination targets determined not to be fault, and a determination unit 14 which determines whether the determination target is fault or not based on the relative position in the reference space based on the average values and variations of the standard deviations between the determination point determined by the average value and variance of the standard deviation of the determination target and the determination area composed of the average value and the variation of the standard deviations of the plurality of other determination targets.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a defect detection device, a defect detection method, and a program.

  A vehicle in which a pressure difference is generated between the inside and the outside of the vehicle may be equipped with a presser device that presses the side sliding door against the vehicle body. As the above-mentioned vehicle, a high-speed railway vehicle such as a Shinkansen in which a pressure difference is generated between the inside of the vehicle and the outside of the vehicle when traveling at high speed or passing a train can be exemplified. The side sliding door is a door provided at the side entrance of the vehicle and is a sliding door type side door.

  As an example of the above-mentioned presser device, the operation is started when the vehicle reaches a certain speed or more. The piston is moved by supplying compressed air into the piston cylinder, and the side sliding door is pressed with a predetermined force. A device that performs a presser motion is known. When releasing (relaxing) the presser action, for example, when stopping at the station and opening the side sliding door, first, an instruction to open the side sliding door is issued by handling (operating) the conductor. When the opening command is issued, the solenoid valve that controls the exhaust of the compressed air in the piston cylinder operates, and the compressed air in the piston cylinder is exhausted. Due to this exhaust, the force with which the piston presses the side sliding door becomes weak, and the pressing operation is released.

  When a problem that the pressing by the above-described presser device does not loosen occurs, there is a problem that the passenger (passenger) cannot get off because the problem also occurs in the opening of the side sliding door. In addition, when a problem that the side sliding door is not pressed occurs, it is difficult to keep the inside of the vehicle airtight, and there is a problem in that passengers feel uncomfortable such as ear hunting (significant service degradation).

  As a method for solving the above-described problem, as shown by a graph A in FIG. 10, when the loosening element of the presser device exceeds a reference range that is equal to or less than a predetermined threshold value for a certain number of times, the presser device is regarded as a defective product. A determination method is known. Here, the loose element is the time from when the side sliding door opening command is issued until the side sliding door starts to open. In addition, a method of using the Mahalanobis distance for determining defective products is also known (for example, see Patent Document 1).

JP, 2006-092909, A

  The above-described defect determination method can determine a defective product when the loosening element exceeds a predetermined reference range for a certain number of times, but there is a problem that determination may not be possible as illustrated below. .

  As an example of the case where the determination cannot be made, as shown by the graph B in FIG. 10, the loosening element is within the above-mentioned reference range, but there is a large variation and the pressing operation is unstable. As another example, as shown by a graph C in FIG. 10, there is a case where the momentary loose element exceeds the reference range described above. These have not yet reached a clear defect as in the case where the loosening element exceeds the above-mentioned reference range for a certain number of times or more, but it is highly possible that the degree of the defect will deteriorate if it continues to be used.

  Furthermore, since it is difficult to compare the degree of defects in the above-described defect determination method, there is also a problem that it is difficult to prioritize measures for eliminating defects such as replacement for a presser device having a plurality of defects.

  The present invention has been made to solve the above-described problem, and it is easy to improve the accuracy of detecting a failure of a target device, and it is possible to easily compare the degree of the failure, and the failure detection An object is to provide a method and a program.

In order to achieve the above object, the present invention provides the following means.
The defect detection apparatus according to the first aspect of the present invention includes an acquisition unit that acquires output information representing a feature quantity to be used for determination, and a standard deviation of the feature quantity in a predetermined period based on the output information. An analysis unit for calculating and calculating an average value and variation based on the standard deviation in a plurality of predetermined periods, and an average value of the standard deviation and the standard obtained for each of a plurality of other determination targets determined not to be defective And a determination point determined by the average value of the standard deviation and the variation of the standard deviation of the determination target in a reference space based on the average value of the standard deviation and the variation of the standard deviation. , And a relative position with respect to the determination area composed of the average value of the standard deviation and the variation of the standard deviation of the plurality of other determination targets. A determination section for determining whether or not the determination target is defect Zui, characterized in that is provided.

  The defect detection method according to the second aspect of the present invention includes an acquisition step of acquiring output information representing a feature amount to be used for determination, and a standard deviation of the feature amount in a predetermined period based on the output information. An analysis step for calculating and calculating an average value and variation based on the standard deviation in a plurality of predetermined periods; and an average of the standard deviation of the determination target in a reference space based on the average value of the standard deviation and the variation of the standard deviation From the determination point determined by the variation of the value and the standard deviation, and the average value of the standard deviation and the variation of the standard deviation obtained for each of a plurality of other determination objects determined not to be defective and stored in the storage unit A determination step for determining whether or not the determination target is defective based on a relative position between the determination region and the configured determination region. And wherein the Rukoto.

  The program according to the third aspect of the present invention provides a computer with an acquisition function for acquiring output information representing a feature quantity to be used for determination, and a standard deviation of the feature quantity over a predetermined period based on the output information. And an analysis function for calculating an average value and variation based on the standard deviation over a plurality of predetermined periods, and a reference space based on the average value of the standard deviation and variation of the standard deviation, the standard deviation of the determination target The average value of the standard deviation and the variation of the standard deviation obtained for each of the determination points determined by the variation of the average value and the standard deviation and the plurality of other determination objects determined not to be defective And a determination function for determining whether or not the determination target is defective based on a relative position between the determination region and the determination region. Characterized in that to.

  According to the defect detection device according to the first aspect of the present invention, the defect detection method according to the second aspect, and the program according to the third aspect, an average value and variation having a plurality of standard deviations as a population are obtained. In the reference space based on, the average value having the feature amount as a population by making a determination based on a relative position between a determination area to which another determination target that is determined not to be a defect belongs and a determination point of the determination target Compared with the case where the determination is performed in the reference space based on the variation, it becomes easier to increase the accuracy of the determination and to easily compare the degree of the defect.

  Specifically, when the determination is made in the reference space based on the average value and the variation having the feature amount as a population, the determination area may include the determination point of the determination target that caused the failure. Thus, when the determination is performed in the reference space based on the average value and the variation having the standard deviation as the population, the determination area does not include the determination point of the determination target causing the problem, and the determination area is separated from the determination area. Therefore, it becomes easy to improve the accuracy of determination, and it becomes easy to compare the degree of defects. Here, the standard deviation is a value obtained by using, as a population, the feature quantity to be determined obtained during a predetermined period.

  In the first aspect of the invention, the storage unit stores an average value of the standard deviation and a variation in the standard deviation obtained for each of a plurality of defect targets determined as defects, and the determination unit includes: Based on the relative position between the determination area and the determination point, and the relative position between the determination area and the defect area composed of the average value of the standard deviation and the standard deviation variation of the plurality of defect objects. It is preferable to determine whether or not the determination target is defective.

  By making a determination based on the relative position between the determination area and the determination point and the relative position between the defect area and the determination point as described above, the determination accuracy can be further improved, and the degree of the defect can be compared. It becomes easier.

  In the first aspect of the invention, the variation of the standard deviation includes a standard deviation in the plurality of standard deviations, a variance in the plurality of standard deviations, and an average value of the plurality of standard deviations and the plurality of standard deviations, respectively. It is preferable that the absolute value of the difference from the value is any summed value.

  In this way, the variation of the standard deviation is the standard deviation with multiple standard deviations in the population, the variance with multiple standard deviations in the population, and the average value and multiple standard deviations with multiple standard deviations in the population. By making the absolute value of the difference from each of these values one of the total values, the variation of the standard deviation can be obtained by calculation processing.

In the first aspect of the present invention, the determination based on the relative position is preferably determination based on Mahalanobis distance.
By making the determination based on the relative position in this way based on the Mahalanobis distance, the accuracy of the determination can be further improved and the degree of failure can be compared more easily. Specifically, the determination based on the Mahalanobis distance is a determination that reflects the shape of the determination area or the defect area, and therefore the accuracy of the determination is easily increased. Further, by comparing the Mahalanobis distance values obtained for the respective determination targets, it becomes easy to compare the degree of defects among the determination targets.

  According to the defect detection device according to the first aspect of the present invention, the defect detection method according to the second aspect, and the program according to the third aspect, a reference based on an average value and variation having a plurality of standard deviations as a population In the space, since the determination is made based on the relative position between the determination area to which another determination target determined to be not defective belongs and the determination point of the determination target, it is easy to improve the accuracy of the defect detection of the target device. There is an effect that it is possible to easily compare the degree of defects.

It is the schematic explaining the structure of the rail vehicle to which the malfunction detection apparatus and malfunction detection apparatus which concern on the 1st Embodiment of this invention are applied. It is a block diagram explaining the structure of the malfunction detection apparatus of FIG. It is a flowchart explaining the outline of the detection method by the malfunction detection apparatus of FIG. FIG. 4A is a graph showing the time transition of the loosening element and the standard deviation when the presser is normal, and FIG. 4B is the graph of the loosening element and the standard deviation when the presser is defective. It is a graph showing a time transition. FIG. 5A is a graph for explaining normal regions, defect regions and determination points in the reference space based on the average value and variation of the standard deviation, and FIG. 5B shows the average value and variation of the loosening element. It is a graph explaining the normal area | region and malfunction area | region in the reference space based. It is the schematic explaining the structure of the rail vehicle to which the malfunction detection apparatus and malfunction detection apparatus which concern on the 2nd Embodiment of this invention are applied. It is a block diagram explaining the structure of the malfunction detection apparatus of FIG. It is a graph showing the time transition of the control temperature difference in two air conditioners. FIG. 9A is a graph for explaining the positional relationship between the control temperature difference data before the treatment and the normal region and the failure region, and FIG. 9B is the control temperature difference data after the treatment, It is a graph explaining the positional relationship with a normal area | region and a malfunction area | region. It is a graph explaining the example where it is difficult to determine the defect in the conventional determination method.

[First Embodiment]
Hereinafter, a malfunction detection apparatus 10 according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 5. In the present embodiment, as shown in FIG. 1, the present invention will be described by applying the present invention to a failure detection device 10 that detects a failure in a presser device (determination target) 111 that presses a side sliding door 110 of a railway vehicle 100 against a vehicle body. As the railway vehicle 100, a vehicle in which a pressure difference is generated between the inside and the outside of the vehicle, specifically, a high-speed railway vehicle such as a Shinkansen can be exemplified.

  The side sliding door 110 is a door provided at the side entrance of the railway vehicle 100 and is a sliding door type side door. The presser device 111 starts the operation of pressing the side sliding door 110 against the vehicle body when the railway vehicle 100 reaches a certain speed or higher. Examples of the presser device 111 include an apparatus having a configuration in which the piston is moved by supplying compressed air into the piston cylinder and the side sliding door 110 is pressed against the vehicle body with a predetermined force. A thing can be used and the structure is not particularly limited.

  The defect detection device 10 is installed in a place different from the railcar 100, for example, in a facility capable of information communication with the railcar 100, and an arithmetic processing device such as a server or a computer that detects a defect in the presser device 111. It is. In the present embodiment, description will be made by applying to an example in which the failure detection apparatus 10 is an information processing apparatus having a CPU (Central Processing Unit), ROM, RAM, an input / output interface, and the like. As shown in FIG. 2, the control program stored in the storage device such as the ROM described above is mainly obtained by cooperating the CPU, ROM, RAM, input / output interface, etc. The storage unit 13, the determination unit 14, the comparison unit 15, and the output unit 16 are made to function.

  The acquisition unit 11 is connected to the recording unit 121 of the railway vehicle 100 so as to be able to perform information communication. The acquisition unit 11 receives data (output information) from the recording unit 121 that represents a loose element (feature amount) used for determining a failure. To get. The communication method between the acquisition unit 11 and the recording unit 121 may be a known communication method, and the communication frequency may be a known frequency and is not particularly limited.

  The recording unit 121 stores data representing the looseness of the presser device 111 detected by the detection unit 122 and is mounted on the railway vehicle 100. The detection unit 122 detects an opening command that is output when the conductor of the railway vehicle 100 is handled (operated), and a detection signal that indicates the start of the opening of the side sliding door 110 that is output from the sensor 112 disposed on the side sliding door 110. Is detected.

  Here, the loose time is the time required from the time when the opening command of the side sliding door 110 is issued by the handling (operation) of the conductor of the railway vehicle 100 to the time when the side sliding door 110 starts to open. Specifically, it is possible to exemplify a method for obtaining a loose time based on time information when an opening command is detected and time information when a detection signal of the sensor 112 is detected.

  The analysis unit 12 performs a calculation process for obtaining a standard deviation using a loose time element acquired in a predetermined period as a population, and then calculates a plurality of standard deviations obtained in each of a plurality of predetermined periods as a population. An arithmetic process for calculating an average value and variation of the standard deviation is performed. The standard deviation obtained here, the average value of the standard deviation, and the variation of the standard deviation are calculated based on the data of the loose time element obtained from one presser device 111.

  In the present embodiment, the predetermined period is described as applied to an example in which the predetermined period is 5 days (for example, Monday to Friday). However, the predetermined period depends on the number of loose time elements required for obtaining the standard deviation. It can be changed as appropriate. For example, when the number of loose time data that can be acquired per day is large, the predetermined period can be shortened. When the number of loose time data that can be acquired per day is small, the predetermined time period can be decreased. It will be longer. In addition, if the number of loose time element data required for obtaining the standard deviation is large, the predetermined period will be lengthened, and the loose time element data required for obtaining the standard deviation will be When the number is small, the predetermined period can be shortened.

  A plurality of predetermined periods used when calculating the average value and variation of the standard deviation in the present embodiment will be described by applying to an example in which a part of the periods overlap each other. The plurality of predetermined periods may not overlap with each other, and there may be vacancy between the periods.

  The average value of the standard deviation in the present embodiment will be described by applying to an example in which a plurality of standard deviations used when obtaining the average is an average value obtained by performing the same weighting. Note that an average value that performs different weighting on a plurality of standard deviations may be used, and is not particularly limited.

  The variation of the standard deviation in the present embodiment will be described by applying to an example that is a standard deviation (standard deviation of standard deviation) using a plurality of standard deviations used when obtaining the variation as a population. Note that the variation of the standard deviation may be a standard deviation of the standard deviation, or a variance having a plurality of standard deviations used for obtaining the variation (a variance of the standard deviation). Furthermore, it may be the sum of the absolute values of the difference between the average value using a plurality of standard deviations used when determining variation and the plurality of standard deviations.

  The storage unit 13 has an average value of standard deviations and variations of standard deviations obtained for a plurality of pressers 111 (hereinafter also referred to as “failure groups”) that have been determined to have a defect in advance, in other words, a defect region NR. The average value of the standard deviation and the variation of the standard deviation obtained for a plurality of presser devices 111 (hereinafter also referred to as “normal group”) determined to be normal (not defective) in advance, In other words, information on the normal area GR is stored.

  In the reference space based on the standard deviation average value and the standard deviation variation, the determination unit 14 belongs to the determination point JP determined by the standard deviation average value and the standard deviation variation of the presser device 111 that is the determination target, and the normal group. Based on the normal area (judgment area) GR composed of the standard deviation average value and the standard deviation variation, and the defect area NR composed of the standard deviation average value and standard deviation variation belonging to the defect group, In addition to performing a process for determining whether or not the presser device 111 as a determination target is defective, a process for evaluating is performed. Details of the determination method and the evaluation method will be described later.

  The comparison unit 15 compares the degree of failure in the presser device 111 that has already been determined to be defective and the degree of failure in the presser device 111 that has been determined to be the latest failure, and ranks the degree of failure in these presser devices 111. It is a thing to do. Details of the ranking will be described later.

  The output unit 16 outputs the result determined by the determination unit 14 and the evaluated result to the outside. As an output partner, for example, a terminal of a department that performs maintenance management of the railway vehicle 100 can be exemplified.

  Next, a defect detection method in the defect detection apparatus 10 having the above configuration will be described with reference to FIGS. In the present embodiment, the defect detection device 10 will be described as applied to an example of determining a defect of the presser device 111 at a predetermined timing. Examples of the predetermined timing include an example in which the determination is made in consideration of factors that are considered to have an influence on the frequency of occurrence of defects, such as the date of manufacture and usage of the presser device 111. Further, the determination of the defect may be performed discretely at a predetermined timing or may be performed continuously, and the determination timing is not limited.

  When the defect detection process by the defect detection apparatus 10 is started, as shown in FIG. 3, the acquisition unit 11 performs a process of acquiring the loose data recorded in the recording unit 121 (S10: acquisition step). ). The acquired data includes information specifying the presser device 111, information related to the loose time element linked to the information specifying the presser device 111, and the time when the detection unit 122 acquired information related to the loose time element (year) Month and day) information.

  When the loose element data is acquired, the analysis unit 12 performs the loose element analysis process (S11: analysis step). Specifically, for the presser device 111 to be determined, a calculation process is performed to obtain a standard deviation with a loose time element acquired in a predetermined period as a population. Thereafter, arithmetic processing is performed for calculating an average value and variation of standard deviations using a plurality of standard deviations obtained in each of a plurality of predetermined periods as a population.

  FIG. 4A is a graph showing the time course of the looseness element and the standard deviation when the presser device 111 is normal. FIG. 4B is a graph showing the time course of looseness and standard deviation when the presser device 111 is defective.

  From the graph YJ representing the time transition of the loose time element, it can be seen that the value of the loose time element fluctuates within a predetermined range when the presser device 111 is normal. On the other hand, when the presser device 111 is defective, it can be seen that the value of the loosening element fluctuates greatly. Further, it can be seen from the graph SD representing the time transition of the standard deviation that when the presser device 111 is normal, the fluctuation of the standard deviation value is relatively small and the average value is small. On the other hand, when the presser device 111 is defective, it can be seen that the variation of the standard deviation value is relatively large and the average value is large.

  When the average value and the variation of the standard deviation are calculated, the determination unit 14 determines whether or not the presser device 111 that is the determination target is defective based on the reference space based on the average value of the standard deviation and the variation of the standard deviation. Is performed (S12: determination step).

The contents of the arithmetic processing to be determined will be described as follows.
The determination unit 14 varies the average value and the standard deviation of the standard deviation obtained for the failure group stored in the storage unit 13, in other words, information on the failure region NR and the standard deviation obtained for the normal group. Variation of the average value and standard deviation, in other words, information on the normal region GR is acquired.

  Then, the defect distance LN, which is the Mahalanobis distance between the determination point JP determined by the average value of the standard deviation of the presser device 111 that is the determination target obtained in the calculation process of S11 and the variation of the standard deviation, and the defect area NR, and The normal distance LG, which is the Mahalanobis distance between the determination point JP and the normal region GR, is obtained by calculation. As a calculation method for obtaining the Mahalanobis distance, a known calculation method can be used, and the calculation method is not limited to a specific calculation method.

  The obtained defect distance LN is compared with the normal distance LG. For example, if the value of the defect distance LN is less than the value of the normal distance LG (the value of the defect distance LN <the value of the normal distance LG), the determination is made. It is determined that the presser device 111 has a problem. On the other hand, if the value of the defect distance LN is equal to or greater than the value of the normal distance LG (the value of the defect distance LN ≧ the value of the normal distance LG), the presser device 111 that is the determination target is determined to be normal.

  Here, the contents of the determination process in the determination unit 14 will be described with reference to the graph shown in FIG. The graph shown in FIG. 5A represents a reference space in which the horizontal axis is an average value of standard deviations, and the vertical axis is variation of standard deviations. In the graph, a white square represents the presser device 111 belonging to the normal group, and a white triangle represents the presser device 111 belonging to the failure group. The white circle represents the presser device 111 (determination point JP) that is an evaluation target.

  For example, as shown in FIG. 5A, even if the determination point JP is not included in the defect area NR or the normal area GR, and the defect of the presser device 111 cannot be determined simply, the defect distance By comparing the LN and the normal distance LG, it is possible to easily determine whether the determination point JP is close to the defective area NR or the normal area GR, and it is possible to determine whether or not there is a defect. In the case shown in FIG. 5A, the value of the defect distance LN ≧ the value of the normal distance LG, and it is determined that the presser device 111 that is the determination target is normal.

  When the defect area NR composed of defect groups and the normal area GR composed of normal groups are displayed on the graph, the defect area NR and the normal area GR are separated as shown in FIG. I know that. As a comparison, when the failure area NR and the normal area GR are displayed on a graph (reference space) in which the horizontal axis is the average value of the loose time elements and the vertical axis is the variation of the loose time elements, the result is as shown in FIG. . In FIG. 5B, it can be seen that the defective area NR and the normal area GR partially overlap each other and are not separated.

  In other words, in the determination of the malfunction of the presser device 111 based on the reference space composed of the average value of the loose time elements and the variation of the loose time elements, as shown in FIG. 5B, the malfunction area NR and the normal area GR It can be seen that it is difficult to determine the defect because the two are partially overlapped and not separated. That is, by performing the determination based on the reference space composed of the average value of the standard deviation and the variation of the standard deviation, the normal presser device 111 is determined to be defective, or the defective presser device 111 is determined to be normal. It becomes difficult to suppress this.

  When it is determined in the determination process of S12 that the presser device 111 that is the determination target is not defective, in other words, is normal (in the case of NO), the detection of the defect in the defect detection device 10 ends, and the next defect is detected. The process of waiting until is performed.

  When it is determined in the determination process of S12 that the presser device 111 that is the determination target is defective (in the case of YES), calculation processing for calculating the rank of the presser device 111 that is the determination target by the comparison unit 15 is performed (S13). .

  Specifically, a calculation process for storing the failure distance LN and normal distance LG in the presser device 111 that has been determined to be the latest failure in the storage unit 13 and a presser device 111 that has already been determined to be a failure and stored in the storage unit 13. The failure distance LN and the normal distance LG at the current position are compared with the failure distance LN and the normal distance LG at the presser device 111 that has been determined to be the latest failure, and the order of the degree of the failure of the presser device 111 determined as the failure is calculated. The arithmetic processing to perform is performed.

  As a method for calculating the rank of the degree of defects, a method of comparing values obtained by a function using the defect distance LN and the normal distance LG as parameters, for example, obtaining a difference value between the defect distance LN and the normal distance LG, A method of calculating the rank of the degree of failure according to the magnitude of the value can be exemplified. In addition, a method for calculating the rank of the degree of defect based on the magnitude of the defect distance LN and a method for calculating the rank of the degree of defect based on the magnitude of the normal distance LG can be exemplified.

  In the present embodiment, the failure distance LN and the normal distance LG in the presser device 111 stored in the storage unit 13 are excluded from those of the presser device 111 in which the failure has been dealt with by work such as repair, maintenance, or replacement. This will be explained by applying to the example. That is, the presser device 111 in which the problem is dealt with will be described by applying to an example in which a calculation process for removing the defect distance LN and the normal distance LG from the storage unit 13 is performed.

  When the ranks of the presser devices 111 determined to be defective are calculated, a list that is ranked based on the degree of defects of the plurality of presser devices 111 determined to be defective, in other words, an evaluated result is created. The The created list is output by the output unit 16 to the outside such as a terminal of a department that performs maintenance management of the railway vehicle 100.

  According to the defect detection device 10 having the above-described configuration, the normal region GR to which the presser device 111 determined to be non-defective (normal) belongs in the reference space based on the average value and variation having a plurality of standard deviations as a population. When the determination is made based on the Mahalanobis distance (relative position) of the determination point JP of the determination target presser device 111, and the determination is performed in the reference space based on the average value and the variation with the loose time element as the population Compared to the above, it becomes easier to improve the accuracy of determination, and it becomes easier to compare the degree of defects. Specifically, when the determination is performed in the reference space based on the average value and variation using the loose time element as a population, the determination point JP of the presser device 111 that has caused a problem may be included in the normal region GR. On the other hand, when the determination is made in the reference space based on the average value and the variation with the standard deviation as the population, the determination point JP of the presser device 111 causing the defect in the normal region GR is not included, and the defect The determination point of the presser device 111 that has caused the separation from the normal region GR. Therefore, it becomes easy to improve the accuracy of determination, and it becomes easy to compare the degree of defects.

  Further, by performing the determination based on the Mahalanobis distance between the normal region GR and the determination point JP and the Mahalanobis distance between the defective region NR and the determination point JP, the normal region GR and the determination point JP Compared with the case where the determination is made only with the Mahalanobis distance between the two, or the case where the determination is made only with the Mahalanobis distance between the defective area NR and the determination point JP, the accuracy of the determination can be further improved. It becomes easier to compare the degree of defects.

  By determining the defect based on the Mahalanobis distance, the accuracy of the determination can be further enhanced and the degree of the defect can be compared more easily. Specifically, the determination based on the Mahalanobis distance is a determination that reflects the shape of the normal region GR and the defect region NR, and therefore the accuracy of the determination is easily improved. Furthermore, by comparing the Mahalanobis distance values obtained for the presser devices 111 that are the respective determination targets, it is possible to easily compare the degree of defects among the determination targets.

[Second Embodiment]
Next, a failure detection apparatus 10 according to the second embodiment of the present invention will be described with reference to FIGS. The basic configuration of the defect detection device 10 of the present embodiment is the same as that of the first embodiment, but the target for detecting the occurrence of a defect is different from that of the first embodiment. Therefore, in the present embodiment, only the relationship with the target for detecting a defect will be described using FIGS. 6 to 9, and description of other contents will be omitted.

  In the present embodiment, as shown in FIG. 6, the present invention will be described by applying the present invention to a failure detection device 10 that detects a failure in an air conditioner (determination target) 210 of a railway vehicle 200. The air conditioner 210 sucks air in the vehicle of the railway vehicle 200 and blows out air whose temperature is adjusted to a desired temperature into the vehicle. As the air conditioner 210, one having a known configuration can be used, and the specific configuration of the air conditioner 210 is not particularly limited.

  In the present embodiment, description will be made by applying to an example in which two air conditioners 210 are arranged in one vehicle. As will be described later, the temperature difference of the intake air detected in a plurality of air conditioners 210 is obtained. If possible, the number of air conditioners 210 arranged in one vehicle is not particularly limited.

  As shown in FIGS. 6 and 7, the air conditioner 210 includes a sensor 212 and a detection unit that detect the temperature of the intake air (hereinafter also referred to as “control temperature”) in the vicinity of the intake port for sucking air in the vehicle. 122 is provided. Detection signals indicating the control temperature detected by the sensor 212 are collected in the detection unit 122 and recorded in the recording unit 121.

  Examples of information recorded in the recording unit 121 include a control temperature in the air conditioner 210 and a difference in control temperature between the two air conditioners 210 arranged in one vehicle. When the temperature of the control air in the air conditioner 210 is recorded, a difference in control temperature in the air conditioner 210 arranged in one vehicle is calculated in the defect detection device 10. In the present embodiment, description will be made by applying to an example in which a difference in control temperature is calculated in the air conditioner 210 arranged in one vehicle in the detection unit 122 or the recording unit 121.

  Since the configuration of the defect detection device 10 is the same as that described in the first embodiment, the description of the configuration is omitted. Further, the contents of the arithmetic processing for detecting a defect by the defect detection device 10 of the present embodiment are intended for the average value of the standard deviation of the loose time element and the variation of the standard deviation of the loose time element in the first embodiment. However, the only difference is that the average value of the standard deviation of the control temperature difference and the variation of the standard deviation of the control temperature difference are targeted. That is, the content of the calculation process for detecting a defect by the defect detection apparatus 10 of the present embodiment is the same as that of the first embodiment except that only the data to be calculated is different, and thus the description thereof is omitted.

  Next, a detection example of the air conditioner 210 that has caused a defect by the defect detection apparatus 10 of the present embodiment and a detection example of the air conditioner 210 that has become normal after treating the defect will be described. The graph shown in FIG. 8 shows the time transition of the control temperature difference in the two air conditioners 210. The left part in the graph shows a state before the air conditioner 210 has a problem and before the trouble is treated (hereinafter also referred to as “before treatment”), and the right part has been treated. The state after (hereinafter also referred to as “after treatment”) is shown.

  The positional relationship between the control temperature difference data (black triangle in the graph) before the treatment and the normal region GR and the defective region NR is as shown in FIG. As shown in the figure, the data of the control temperature difference before the treatment is located inside the defective area NR. Therefore, it is determined that a defect has occurred even in the defect detection calculation process performed by the defect detection apparatus 10.

  On the other hand, the positional relationship between the control temperature difference data after treatment (black circles in the graph) and the normal region GR and the defective region NR is as shown in FIG. 9B. As shown in the figure, the data of the control temperature difference after the treatment has moved to the inside of the normal region GR. For this reason, it is determined that the malfunction detection calculation process by the malfunction detection apparatus 10 is normal.

  The technical scope of the present invention is not limited to the above embodiment, and various modifications can be made without departing from the spirit of the present invention. For example, in the present embodiment, the defect detection device 10 is described as applied to an example in which the failure detection device 10 is installed in a facility different from the railcar 100. However, the defect detection device 10 may be mounted on the railcar 100, The installation position of the defect detection device 10 is not particularly limited.

  In the present embodiment, when determining whether or not the presser device 111 is defective, the distance between the determination point JP and the defective area NR and the distance between the determination point JP and the normal area GR. However, the determination may be made using only the distance between the determination point JP and the normal region GR. In this case, if the distance between the determination point JP and the normal region GR is smaller than a threshold value determined based on the shape or spread of the normal region GR, the presser device 111 is determined to be normal, and if the distance is larger than the threshold value, the presser device is determined. 111 is determined to have a problem.

  In the present embodiment, the description is applied to an example in which the Mahalanobis distance is used to determine whether or not the presser device 111 is defective. However, the distance between the determination point JP and the defective region NR Any evaluation method may be used as long as the distance between the determination point JP and the normal region GR can be evaluated, and the evaluation method is not particularly limited.

  DESCRIPTION OF SYMBOLS 10 ... Defect detection apparatus, 11 ... Acquisition part, 12 ... Analysis part, 13 ... Memory | storage part, 14 ... Determination part, 111 ... Presser device (determination object), 210 ... Air-conditioner (determination object), JP ... Determination point, GR ... normal area (determination area), NR ... defective area, S10 ... acquisition step, S11 ... analysis step, S12 ... determination step

Claims (6)

An acquisition unit for acquiring output information representing a feature amount of a determination target used for determination;
Obtaining a standard deviation of the feature amount in a predetermined period based on the output information, and calculating an average value and variation based on the standard deviation in a plurality of predetermined periods;
A storage unit for storing the average value of the standard deviation and the variation of the standard deviation obtained for each of a plurality of other determination objects determined not to be a defect;
In a reference space based on the average value of the standard deviation and the variation of the standard deviation, a determination point determined by the average value of the standard deviation of the determination target and the variation of the standard deviation, and the standard of the plurality of other determination targets A determination unit that determines whether or not the determination target is defective based on a relative position between an average value of deviation and a determination region configured by variations in the standard deviation;
A defect detection device characterized in that is provided. The storage unit stores an average value of the standard deviation and a variation of the standard deviation obtained for each of a plurality of defect targets determined as defects,
The determination unit includes a relative position between the determination area and the determination point, and an average value of the standard deviations of the plurality of defect targets and variations in the standard deviation and the determination point. The defect detection device according to claim 1, wherein it is determined whether the determination target is a defect based on a relative position.   The variation of the standard deviation includes a standard deviation in a plurality of the standard deviations, a variance in the plurality of standard deviations, and an absolute value of a difference between an average value of the plurality of standard deviations and a value of each of the plurality of standard deviations. The defect detection device according to claim 1, wherein the defect detection device is any one of a total value of the two.   4. The defect detection device according to claim 1, wherein the determination based on the relative position is a determination based on a Mahalanobis distance. 5. An acquisition step of acquiring output information representing a feature amount of a determination target used for determination;
An analysis step for obtaining a standard deviation of the feature amount in a predetermined period based on the output information and calculating an average value and variation based on the standard deviation in a plurality of predetermined periods;
In a reference space based on the average value of the standard deviation and the variation of the standard deviation, a determination point determined by the average value of the standard deviation and the variation of the standard deviation of the determination target, and a plurality of other points determined not to be defective It is determined whether or not the determination target is defective based on a relative position between an average value of the standard deviation obtained for each determination target and stored in the storage unit, and a determination region configured by variations in the standard deviation. A determination step for determining;
A defect detection method characterized by comprising: On the computer,
An acquisition function for acquiring output information representing a characteristic amount of a determination target used for determination;
An analysis function for obtaining a standard deviation of the feature amount in a predetermined period based on the output information, and calculating an average value and variation based on the standard deviation in a plurality of predetermined periods;
In a reference space based on the average value of the standard deviation and the variation of the standard deviation, a determination point determined by the average value of the standard deviation and the variation of the standard deviation of the determination target, and a plurality of other points determined not to be defective It is determined whether or not the determination target is defective based on a relative position between an average value of the standard deviation obtained for each determination target and stored in the storage unit, and a determination region configured by variations in the standard deviation. A judgment function to judge,
A program characterized by realizing.