Diagnostic device and diagnostic method

Abstract

To provide a failure sign detection system that can be realized at low cost and can reduce false detection of failure signs in air conditioners caused by information that does not require measurement.SOLUTION: The diagnostic device includes a power diagnostic unit that detects signs of failure in any of the air conditioning units in a train set based on the power consumption in the train set, a data storage unit that stores information acquired from cars included in the train set, and a normal-state model generation unit that generates a normal-state model when the air conditioning units are operating normally. The power diagnostic unit estimates an estimated value of the power consumption in the train set under normal conditions using power change factors at diagnosis and the normal-state model, and determines that a sign of failure has occurred in any of the air conditioning units in the train set when a deviation obtained by subtracting the power consumption in the train set under diagnosis from the estimated value of the power consumption in the normal condition exceeds a predetermined value.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a diagnostic device and a diagnostic method. [Background technology]

[0002] Maintenance of railway vehicle air conditioning systems has traditionally been carried out using the TBM (Time Based Maintenance) method, which involves performing maintenance at regular intervals regardless of whether or not there is a malfunction. However, the CBM (Condition Based Maintenance) method has also been proposed, in which maintenance work is performed when signs of a malfunction are detected. With CBM, no maintenance is performed unless there is a malfunction, so it is expected to reduce the frequency of maintenance and the maintenance costs of railway vehicle air conditioning systems.

[0003] To achieve CBM in railway vehicle air conditioning systems, technology is needed to understand the status of the air conditioning system using information such as the interior temperature of the railway vehicle and the air conditioning system installed inside the system, and to determine signs of failure in the air conditioning system.

[0004] Although there are various types of failures in railway air conditioning, such as damage to refrigerant piping, failure of air conditioning compressors due to fouling of the indoor heat exchanger, and abnormal fan stoppage, this disclosure focuses on detecting signs of failure due to at least fouling of the indoor heat exchanger. If fouling of the indoor heat exchanger can be detected and maintenance to remove the fouling can be performed at the appropriate time, it will be possible to avoid events such as a decrease in comfort inside the car due to a decrease in cooling capacity caused by the fouling, and failure of the air conditioning compressor, and reduce the frequency of maintenance.

[0005] Furthermore, if the above-mentioned signs of failure could be detected using sensor information from sensors already installed in railway vehicles and air conditioning equipment, there would be no need to install new sensors specifically for detecting signs of failure, which would reduce the costs of installing sensors and maintaining the sensors themselves. Examples of sensor information that have traditionally been used to detect the performance of railway air conditioning and are considered effective for detecting signs of failure include in-car temperature sensor information used to control railway air conditioning, and air conditioning current and voltage information. Methods for detecting signs of failure in air conditioning using temperature sensors and power information have been proposed for railways and other fields.

[0006] Patent Document 1 discloses a method for determining the degree of contamination based on the difference between an actual temperature sensor measurement value inside a railway vehicle and a reference interior temperature associated with the outside air temperature and occupancy rate, and utilizing the determined degree of contamination for detecting signs of malfunction. Specifically, Patent Document 1 discloses the following as an invention relating to a vehicle air conditioning device, a vehicle air conditioning management system, and a vehicle air conditioning management method: "The present invention provides a vehicle air conditioning management system and a vehicle air conditioning management method that are capable of grasping changes in interior temperature from various environmental information serving as a reference for a comfortable interior environment and actual environmental information, early detection of abnormalities in components, and accurate determination of the need for and urgency of maintenance. The vehicle air conditioning management system 100 according to the present invention is characterized by including an air conditioning control device 20 that predicts abnormalities in components of the refrigeration cycle 10 by comparing an interior temperature T2 obtained when the refrigeration cycle 10 is actually operated with a preset reference interior temperature T1."

[0007] Patent Document 2 discloses a method for determining a decrease in cooling capacity based on the difference between the time it takes for the average interior temperature, calculated based on the interior temperatures of each car, to reach a set temperature after a railway air conditioning system receives a command to perform a test cooling operation and starts the test cooling operation, and the time it takes for the interior temperature of each car to reach the set temperature. Specifically, Patent Document 2 aims to provide a vehicle air conditioning system management system that can easily determine a decrease in the cooling capacity of an air conditioning system in a short time, and discloses the following content as an invention related to the vehicle air conditioning system management system: "The vehicle information control device 3 is configured to, when the air conditioning system 2 starts a test cooling operation, read the interior temperatures of each car 4, 5, and 6 detected by each temperature sensor 7 to calculate an average interior temperature, measure a first time until the average interior temperature reaches the set temperature, measure a second time until the interior temperatures of each car 4, 5, and 6 reach the set temperature, and subtract the first time from the second time to obtain deviations, and determine that the cooling capacity of the air conditioning system 2 has decreased when the obtained deviations are equal to or greater than a predetermined value."

[0008] Patent Document 3 discloses a method for diagnosing deterioration of an air conditioner based on the amount of power consumed by the air conditioner, targeting indoor air conditioning in a building, etc. Specifically, the method selects the pair with the closest similarity between the time rate of change in power consumption of multiple patterns acquired when the air conditioner was installed and the time rate of change in power consumption currently acquired, and performs deterioration determination based on the similarity. Specifically, Patent Document 3 aims to easily determine the degree of deterioration of an air conditioner, and discloses the following as inventions related to an air conditioner deterioration determination device, an air conditioning system, a deterioration determination method, and a deterioration determination program. "The air conditioner deterioration determination device 1 comprises a power consumption storage unit 11, a first change curve change rate calculation unit 12, a second change curve change rate calculation unit 13, a reference data setting unit 14, and a deterioration degree calculation unit 15. The first change curve change rate calculation unit 12 calculates the rate of change between each predetermined time for the first change curve, which is the change curve of the air conditioner's reference power consumption obtained at each predetermined time. The second change curve change rate calculation unit 13 calculates the rate of change between each predetermined time for the second change curve, which is the change curve of the air conditioner's judged power consumption obtained at the same predetermined time. The reference data setting unit 14 obtains the first change curve, the difference from the second change curve of which is within a predetermined range, based on the rate of change between each predetermined time, and sets the reference power consumption having the first change curve as reference data. The deterioration degree calculation unit 15 calculates the degree of deterioration of the air conditioner based on the difference between the reference data and the judged power consumption." [Prior art documents] [Patent documents]

[0009] [Patent Document 1] International Publication No. 2000 / 150724 [Patent Document 2] Japanese Patent Application Laid-Open No. 2013-248977 [Patent Document 3] Japanese Patent Application Laid-Open No. 2009-020721 Summary of the Invention [Problem to be solved by the invention]

[0010] The method of Patent Document 1 calculates the reference interior temperature by taking into account factors other than soiling that affect the interior temperature, such as outside temperature and occupancy rate, and is therefore expected to improve the performance of detecting signs of malfunction compared to methods that use a fixed value for the reference interior temperature. However, factors that cause changes in the interior temperature include not only outside temperature and occupancy rate, but also door opening and closing, the inflow of outside air while the vehicle is moving, window opening and closing, and solar radiation. For this reason, due to factors not taken into account in Patent Document 1, it is possible that the expected improvement in detection performance will not be achieved.

[0011] It is also possible to develop a method for setting a reference interior temperature by taking into account all of the above factors, by applying the method described in Patent Document 1. However, it is not easy to measure solar radiation information and window opening and closing status on trains, and there are concerns that it will be difficult to develop a method that takes these factors into account.

[0012] The method of Patent Document 2 makes it possible to detect signs of failure in a specific air conditioning unit by comparing the average temperature change within the train with the temperature change in each car. However, in cases where the problem is thought to progress uniformly in each car within the train, such as soiling of the indoor heat exchanger, it may be difficult to detect a difference between the average temperature change within the train and the temperature change in each car.

[0013] Furthermore, in the method of Patent Document 2, it is assumed that the test cooling operation will be carried out when entering a railcar depot, etc., so it is easy to fix information that cannot be measured by on-board sensors, such as window opening / closing and sunlight, to specific conditions. For example, with regard to sunlight, the test cooling operation can be limited to inside the pit, and conditions with low sunlight can be set. By fixing conditions that cannot be sensed in this way, it is expected that detection accuracy will be improved when combined with the method of Patent Document 1.

[0014] However, Patent Document 2 does not mention when or how often inspections should be carried out. For this reason, it has become necessary to reduce the amount of power consumed during inspections by conducting test runs at appropriate times and at appropriate frequencies, and it is necessary to propose a separate method for determining the appropriate timing and frequency of inspections.

[0015] The method of Patent Document 3 performs deterioration determination based on the power of each individual air conditioning unit. However, in railways, it is rare to obtain the power of each individual air conditioning unit; instead, the total power, including the power of the air conditioning unit, is often obtained from the current and voltage of equipment that supplies power to the air conditioning unit, such as a static inverter (SIV). For this reason, when applying the method of Patent Document 3 to railways, it is possible to perform deterioration determination based on the power of the SIV instead of the air conditioning equipment. However, Patent Document 3 does not consider the use of an SIV. Regardless of which prior art was used, it was difficult to realize a failure prediction and detection system using measurable sensor information from existing sensors.

[0016] An object of the present invention is to propose a failure sign detection system that can be implemented at low cost and that reduces false detections of failure signs in air conditioners that occur due to unmeasurable information. [Means for solving the problem]

[0017] In order to solve the above-mentioned problems, one representative diagnostic device of the present invention is a diagnostic device that includes a power diagnosis unit that detects signs of failure in one of the air conditioning units in a train based on the power consumption in the train, a data storage unit that stores information acquired by the cars included in the train, and a normal-state model generation unit that generates a normal-state model for when the air conditioning units are operating normally, wherein the power diagnosis unit estimates an estimated value of the power consumption in the train under normal conditions by using power change factors at diagnosis and the normal-state model, and determines that a sign of failure has occurred in one of the air conditioning units in the train if the deviation obtained by subtracting the power consumption in the train under diagnosis from the estimated value of the power consumption in the train under normal conditions exceeds a predetermined value. [Effects of the Invention]

[0018] According to the present invention, by using the power consumption within a train, which can be estimated using information available on the railway, it is possible to reduce false detection of signs of air conditioning unit failure due to information that cannot be measured, and to construct a detection system that can be implemented at low cost. Problems, configurations, and effects other than those described above will become apparent from the following description of the preferred embodiment of the invention. [Brief explanation of the drawings]

[0019] [Figure 1] FIG. 1 is a diagram showing the configuration of a formation and a diagnosis device in the first embodiment. [Figure 2] FIG. 2 is a graph showing information necessary for generating a normal state model in the first embodiment. [Figure 3] FIG. 3 is a diagram showing an example of the power consumption d32 in the formation during learning and the power change factors d31 during learning. [Figure 4] FIG. 34 is a graph showing a method for calculating the deviation and a specific example of the deviation. [Figure 5] FIG. 5 is a diagram showing the configuration of a formation and a diagnosis device in the second embodiment. [Figure 6] FIG. 6 is a graph summarizing the estimation targets of the normal model in the second embodiment and information used for estimation. [Figure 7] FIG. 7 is a table showing an example of a data set. [Figure 8] FIG. 8 is a flowchart of the first and second embodiments. [Figure 9] FIG. 9 is a diagram showing the configuration of a knitting and diagnosis device in the third embodiment. [Figure 10] FIG. 10 is a flowchart of the failure sign detection in the third embodiment. [Figure 11] FIG. 11 is a graph showing an example of an air conditioning control method and a cooling performance evaluation method in Example 3. DETAILED DESCRIPTION OF THE INVENTION

[0020] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings to help understand the present invention. Note that the following embodiments are examples of specific embodiments of the present invention and do not limit the technical scope of the present invention. In addition, in the description of the drawings, the same parts are designated by the same reference numerals.

[0021] (Internal power output device) This disclosure focuses on the power supplied by an in-set power output device configured with an SIV or the like. The in-set power output device supplies power to multiple devices, such as air compressors, lighting, and doors, in addition to air conditioning devices. The power waveforms of the in-set power output device tend to match the shapes of the power waveforms of the air conditioning devices and the like, and it is believed that by examining the power waveforms of the in-set power output device, it may be possible to predict the operation of the air conditioning devices and the like.

[0022] Furthermore, unlike the temperature inside the train, the power consumed by the power output devices in the train set (power consumption inside the train set) can be estimated using information currently available on railways. Specifically, it can be estimated using information indicating the operating state of each device to which power is supplied. For example, power consumption can be estimated by analyzing the power consumed when the device is on or off. Furthermore, the power consumed by air conditioning devices, which consume most of the power supplied from the power output devices in the train set, depends on the outside temperature and the temperature inside the train in addition to the operating state, and information on these temperatures is also available.

[0023] In this disclosure, in particular, a description will be given of obtaining information (specifically, power consumption and power change factors) related to the power supplied from the in-train power output devices to detect signs of failure in the air conditioning devices. [Example]

[0024] In the first embodiment, a diagnostic device that estimates a failure sign of an air conditioner based on the difference between the estimated value of the power consumption in the train set during normal operation and the power consumption in the train set during diagnosis will be described.

[0025] (Configuration of diagnostic equipment) 1A is a diagram illustrating the configuration of a train set 2 and a diagnostic device 3a in Example 1. The train set 2 is a group of cars consisting of one or more cars, and includes an in-train information transmitter 21, an air conditioning device 22, an in-train power output device 23, and a power change factor device 24.

[0026] In the formation 2, the formation information transmitter 21 transmits information output by the air conditioner 22, the formation power output device 23, and the power change factor device 24 to the diagnostic device 3.

[0027] The diagnostic device 3a is a device that detects signs of failure in the air conditioning device 22 based on information transmitted by the intra-train information transmission device 21, and includes a data accumulation unit 31, a normal state model generation unit 32, a normal state model 33, and a power diagnosis unit 34. The diagnostic device 3a may be located within the train set 2 as shown in FIG. 1 , or may be installed in ground equipment outside the train set 2.

[0028] In the diagnostic device 3a, a data storage unit 31 stores information transmitted by the intra-component information transmission device 21 for a predetermined period of time, and a normal state model generation unit 32 generates a normal state model 33 based on the learning-time power change factors d31 and learning-time intra-component power consumption d32 contained in the data storage unit 31.

[0029] The power diagnosis unit 34 has the functions of normal-state power estimation 341, power difference calculation 342, and failure sign determination 343 based on the difference. The power diagnosis unit 34 detects a failure sign for any of the air conditioning units in the train set based on the in-train power consumption (learning-time in-train set power consumption d32 and diagnosis-time in-train set power consumption d34). Specifically, the power diagnosis unit 34 estimates normal-state power by performing normal-state power estimation 341 using the normal-state model 33 and diagnosis-time power change factors d33 included in the data accumulation unit 31. The power diagnosis unit 34 further calculates the difference between the diagnosis-time in-train set power consumption d34 included in the data accumulation unit 31 and the normal-state power in power difference calculation 342, and performs failure sign determination 343 based on the difference. The data included in the data accumulation unit 31 (learning-time power change factors d31, learning-time in-train set power consumption d32, diagnosis-time power change factors d33, and diagnosis-time in-train set power consumption d34) will be described later.

[0030] The intra-set information transmitter 21 is a device that transmits information output by the air conditioning unit 22, the intra-set power output unit 23, and the power change factor device 24 that are mounted on the set 2 to the diagnostic device 3a, and is, for example, a vehicle information control device. However, the intra-set information transmitter 21 may be mounted on each of the air conditioning unit 22, the intra-set power output unit 23, and the power change factor device 24 and transmit information to the data storage unit 31, and is not limited to being a single vehicle information control device as shown in the present disclosure.

[0031] The air conditioner 22 is a device that controls the number of operating internal air conditioning compressors based on temperature information and set temperature information inside the vehicle, and maintains the temperature inside the vehicle at a predetermined temperature, and has a cooling operation function.

[0032] The in-formation power output device 23 is a device that outputs power consumed within the formation 2. The power consumed by each device within the formation 2 is supplied by, for example, a VVVF (variable voltage variable frequency) inverter or a static inverter (SIV). However, since the power consumed by the air conditioning unit 22 is supplied by, for example, an SIV, hereafter in this specification, the in-formation power refers to the power output by a device such as an SIV that supplies power to the air conditioning unit 22. In other words, if the air conditioning unit 22 receives power from an SIV, the in-formation power output device 23 refers to the SIV.

[0033] The in-formation power output device 23 has the function of supplying power to devices such as the air conditioning device 22, as well as the function of transmitting the output power of the in-formation power output device 23 or information (current value and voltage value) required to calculate the output power to the in-formation information transmitter 21.

[0034] The power change factor device 24 is a device to which the in-set power output device 23 supplies power, and although only one is shown in Figure 1 for ease of understanding, it includes multiple devices. For example, if the in-set power output device 23 supplies power to the air conditioning device 22, compressor, door opening and closing device, lighting, fans that circulate air inside the car, etc., these are included in the power change factor device 24. Also, the arrows connecting the blocks in Figure 1 represent the input and output of information, and electrical and mechanical connections are not shown. For example, although the in-set power output device 23 and the power change factor device 24 are electrically connected, this connection will not be shown in this specification.

[0035] The data storage unit 31 stores information acquired by the cars included in the formation. Specifically, the data storage unit 31 has a function of storing information transmitted from the formation information transmission device 21 for a predetermined period of time and outputting a portion of the stored data to the normal state model generation unit 32 and the power diagnosis unit 34.

[0036] The normal state model generation unit 32 generates a normal state model for the case where the air conditioners 22 are normal. Specifically, the normal state model generation unit 32 has a function of generating a normal state model 33 used to estimate the power consumption in the train set (normal state power) for the case where the air conditioners 22 are normal, based on the learning-time power change factors d31 and the learning-time power consumption in the train set d32 extracted from the data accumulation unit 31.

[0037] The processing by the normal state model generation unit 32 is performed before a diagnosis is performed by the power diagnosis unit 34, which will be described later, and a normal state model 33 is generated in advance. The timing of the processing by the normal state model generation unit 32 is when learning data has been accumulated. For example, one possible method is to set the learning data accumulation period to 0 to 3 months after maintenance work on the air conditioner 22, and perform the processing when a sufficient amount of data has been accumulated to generate the normal state model 33.

[0038] The learning-time power change factors d31 are a group of information that are factors that affect the power consumption within the formation. For example, the learning-time power change factors d31 include at least information on the number of operating air conditioning compressors of the air conditioners and operation information on compressors used for opening and closing doors, etc. Furthermore, because the power consumption of the air conditioners 22 tends to increase as the outside temperature increases, the learning-time power change factors d31 may also include outside temperature information.

[0039] The air conditioning compressor is in the refrigeration cycle of the air conditioner 22 and has the function of compressing the refrigerant that has passed through the indoor heat exchanger. On the other hand, the compressor is a device that generates compressed air for braking, opening and closing doors, etc., and is a different device from the air conditioning compressor. The outside air temperature is temperature information obtained from a temperature sensor installed outside the vehicle.

[0040] The learning-time in-formation power consumption d32 is information indicating the in-formation power consumption corresponding to the learning-time power change factor d31. The learning-time in-formation power consumption d32 indicates the in-formation power consumption at the same time as the learning-time power change factor d31.

[0041] The learning-time power change factors d31 and the learning-time in-train power consumption d32 are acquired within a learning data accumulation period, and are time-series data having a plurality of values ​​within a predetermined time range.

[0042] (Generating a normal model) The normal state model 33 is a model for estimating normal state power. The normal state model generation unit 32 uses the learning state power change factor d31 and the learning state power consumption in the train set d32 to calculate an increase ΔE i Calculate the increase in power consumption during a specific learning session, ΔE i and the learning-time power change factor at a certain time, and the sum of these products is calculated for all learning-time power change factors d31, and the result is generated as the normal-time model 33. This will be specifically described below.

[0043] 2 is a graph 4 showing information necessary for generating a normal state model in Example 1. The upper part of graph 4 is a graph showing a time series transition 41 of power consumption y in a train set during learning, and the lower part of graph 4 is a graph showing a value xi corresponding to the i-th (i is an integer between 1 and N, and N is a predetermined integer) learning-time power change factor 42 among the learning-time power change factors d31. To facilitate understanding, graph 4 omits the display of all but the i-th of the multiple types of learning-time power change factors.

[0044] In the first embodiment, for all power change factors including the i-th power change factor, the increase amount ΔE of the power consumption y in the learning configuration when a specific learning power change factor 42 transitions from the first state to the second state at time t1 (43) is calculated. i (44) is calculated. Furthermore, ΔE i and the state of the power change factor at a certain time t, u i Using (t), a normal state model 33 is constructed by the following equation (1): N is the total number of types of power change factors d31 during learning.

number

[0045] ΔE i Specifically, the calculation method of (44) is to subtract the power consumption y in the learning train set at time t1 (43) from the power consumption y in the learning train set at time t2 (45), a predetermined time after time t1 (43). For time t2 (45), for example, select the time when the value of the power consumption y in the learning train set stops increasing or decreasing after the second state is reached.

[0046] u i Regarding (t), if the first state of a certain power change factor device 24 is its operation off and its operation on, then the time when it becomes the first state is u i (t) is 0, and u is the time when the second state occurs. i Let (t) be 1.

[0047] The normal state model 33 may be a model that takes into consideration the transient response in which the power in the train set gradually increases due to power change factors. In this case, the normal state model generation unit 32 calculates the time constant T i Calculate the increase in power consumption d32 during a specific learning session ΔE i The sum of the products of the learning power change factor at a certain time and the term representing the unit step transient response obtained from the time constant is calculated for all learning power change factors, and the sum is generated as a normal state model 33. For example, the time when the ith learning power change factor d31 transitions from the first state to the second state is defined as t i , the time constant in the transient response of the power consumption d32 in the corresponding learning time configuration is T i Then, the normal state model 33 may be expressed as the following equation (2).

number

[0048] In equation (2), time t i After that, E changes over time by ΔE i represents a step response that increases by ΔE i and u i The term in parentheses between (t) represents the unit step transient response. Time constant T i is the time when the state transitions from the first state to the second state. i Regarding the calculation method, during learning, after transitioning from the first state to the second state, the increase in E is ΔE i This can be roughly calculated by taking the time it takes for the temperature to reach 63.2% of the steady-state value as the time constant.

[0049] FIG. 3 is a diagram showing an example of learning-time power consumption in a train set d32 and learning-time power change factors d31. The learning-time power consumption in a train set d32 and learning-time power change factors d31 are values ​​acquired at times t1, t2, t3, etc. For the learning-time power consumption in a train set d32, the acquired values ​​are a at time t1, a2 at time t2, and a3 at time t3. The i-th learning-time power change factor d31 is the number of operating compressors, which is o1 at time t1, o2 at time t2, and o3 at time t3. The (i+1)th learning-time power change factor d31 is the compressor open / close signal, which is r1 at time t1, r2 at time t2, and r3 at time t3. The (i+2)th learning-time power change factor d31 is the outside air temperature, which is d1 at time t1, d2 at time t2, and d3 at time t3.

[0050] Equations (1) and (2) can be derived based on the learning-time in-formation power consumption d32 and learning-time power change factor d31. In deriving the equations, the "first state" is set to, for example, a state in which the compressor is running (on state), and the "second state" is set to a state in which the compressor is not running (off state). In the case of the number of operating compressors, for example, it is possible to set a state in which the number of operating compressors is equal to or greater than a predetermined number as the first state, and a state in which the number of operating compressors is less than the predetermined number as the second state.

[0051] While the learning-time in-formation power consumption d32 and learning-time power change factors d31 have been described, data is acquired in a similar format in the case of diagnosis-time in-formation power consumption d34 and diagnosis-time power change factors d33, which will be described later, except for the data acquisition period. Also, when using outside air temperature as a power change factor, it may be difficult to create a model that includes changes in outside air temperature, so it may be possible to create a normal-time model for each outside air temperature or set a correction term.

[0052] The above is the description of how the normal state model 33 is generated in the first embodiment.

[0053] (Failure prediction detection) Next, we will describe the elements necessary for diagnosis and the diagnostic means using the normal state model 33. The diagnosis is the detection of signs of failure in the air conditioner 22, and is carried out at a time other than the learning data accumulation period.

[0054] The diagnostic power change factors d33 are a group of information that are factors that affect the power consumption within the formation. For example, the diagnostic power change factors d33 include at least information on the number of operating air conditioning compressors of the air conditioning units and operation information on compressors used for opening and closing doors, etc. Furthermore, because the power consumption of the air conditioning units 22 tends to increase as the outside temperature increases, the diagnostic power change factors d33 may also include outside temperature information.

[0055] The in-train-of-car-car power consumption at diagnosis d34 is information indicating the in-train-of-car-car power consumption corresponding to the in-train-of-car-car power change factor d33. The in-train-of-car-car power consumption at diagnosis d34 indicates the in-train-of-car-car power consumption at the same time as the in-train-of-car-car power change factor d33.

[0056] The power change factor at diagnosis d33 and the power consumption in the train set at diagnosis d34 are acquired outside the learning data accumulation period, and are time-series data having a plurality of values ​​within a predetermined time range.

[0057] The power diagnosis unit 34 has a function of detecting signs of failure of the air conditioners 22 at the time of diagnosis, based on the power change factors at diagnosis d33, the in-train-of-train power consumption at diagnosis d34, and the normal state model 33. Specifically, the power diagnosis unit 34 estimates an estimated value of in-train-of-train power consumption during normal times (normal state power E) using the power change factors at diagnosis d33 and the normal state model 33, and determines that a sign of failure has occurred in one of the air conditioners 22 in the train set 2 if the deviation obtained by subtracting the in-train-of-train power consumption during diagnosis d34 from the estimated value of in-train-of-train power consumption during normal times exceeds a predetermined value.

[0058] In the normal state power estimation 341, the diagnostic power change factor d33 is input to the normal state model 33, and the normal state power at the time of diagnosis is estimated by the formula (1) or (2). Specifically, in the formula (1) or (2), the i-th diagnostic power change factor included in the diagnostic power change factor d33 is expressed as u i(t) and calculate the normal power consumption E.

[0059] The power difference calculation 342 calculates the deviation by subtracting the power consumption d34 in the formation at the time of diagnosis from the normal power E.

[0060] 4 is a graph 5 showing a method for calculating the deviation and a specific example of the deviation. The upper part of graph 5 is a graph showing the time progression of the power consumption y in the train set at the time of diagnosis, with the solid line showing the time progression of the power consumption 51 in normal operation and the dashed line showing the time progression of the power consumption 52 in the train set at the time of diagnosis. The lower part of graph 5 is a graph showing the time progression of the deviation 54 obtained by subtracting the power consumption 52 in the train set at the time of diagnosis from the power consumption in normal operation 51, with the dotted line showing the time progression of the deviation 54 and the solid line showing the smoothed deviation 55 obtained by smoothing the deviation 54.

[0061] When the time trends of normal power 51 and power consumption in the train at diagnosis 52 are obtained as shown in the upper part of graph 5, the time trends of deviation 54 obtained by subtracting power consumption in the train at diagnosis 52 from power consumption in normal power 51 are obtained as shown in the lower part of graph 5.

[0062] Furthermore, the deviation 54 may temporarily increase due to model errors, which may result in false detection of a failure sign. Therefore, instead of the deviation 54, the deviation 54 may be smoothed using a moving median, a moving average, or the like to obtain a smoothed deviation 55, which may be used in subsequent processing.

[0063] In the failure sign determination 343 based on the difference, it is determined that a failure sign has occurred when the deviation 54 or the smoothed deviation 55 exceeds a predetermined threshold. This disclosure focuses particularly on failure sign detection by detecting dirt in the air conditioning indoor heat exchanger, and the method of determination differs depending on whether the rotation speed of the air conditioning compressor is constant regardless of the degree of dirt or whether the rotation speed increases due to dirt.

[0064] This utilizes the fact that the power consumption of the air conditioning compressor may decrease when the rotation speed of the air conditioning compressor is constant.

[0065] Specifically, when the indoor heat exchanger becomes dirty, the amount of heat exchanged between the refrigerant and the air in the indoor heat exchanger decreases, resulting in a decrease in the amount of refrigerant evaporation and a decrease in the flow rate of gaseous refrigerant flowing into the air conditioning compressor. When the rotation speed of the air conditioning compressor is constant, the refrigerant flow rate is reduced compared to when the indoor heat exchanger is clean, and it is thought that the energy required to compress the refrigerant is reduced, which may result in a decrease in the power consumption of the air conditioning compressor. The majority of the power consumption of the air conditioning unit 22 is consumed by the air conditioning compressor, and the air conditioning unit 22 also accounts for the majority of the power consumption within the train. Therefore, it is thought that the decrease in power consumption of the air conditioning compressor due to the dirt on the indoor heat exchanger will significantly reduce the power consumption within the train.

[0066] That is, when dirt is present, the power consumption in the train decreases compared to normal times, so deviation 54, which is obtained by subtracting power consumption in the train at diagnosis 52 from normal power 51, increases in a positive direction. Therefore, if deviation 54 exceeds a predetermined value, it is possible to determine that the dirt has increased and that a fault symptom has occurred.

[0067] On the other hand, if the compressor rotation speed is not constant, there is a possibility that power consumption will increase due to dirt, so in such a precondition, if the deviation 54 falls below a predetermined threshold, it is determined to be a sign of failure.

[0068] (Actions and Effects) The above is the failure sign diagnosis method for the air conditioner 22 in Example 1. Example 1 makes it possible to diagnose failure signs of the air conditioner 22, particularly to detect contamination of the indoor heat exchanger, by estimating the power consumption within the formation using measurable information. Therefore, compared to failure sign detection of the air conditioner 22 using a temperature sensor, the influence of reduced detection performance due to information that does not need to be measured is minimized, making it possible to reduce false detections compared to conventional failure sign detection methods using temperature sensors.

[0069] Furthermore, since the information used in the first embodiment can be acquired relatively easily from the vehicle, the addition of equipment for acquiring data is minimized, and a low-cost detection system can be constructed. [Example]

[0070] In the second embodiment, an example is shown in which, for the purpose of performing a more accurate normal power estimation, normal power estimation 341 uses the power consumption in the formation at diagnosis in addition to the factors of power change at diagnosis.

[0071] (Configuration of diagnostic equipment) Fig. 5 is Fig. 1b showing the configuration of a train set 2 and a diagnostic device 3b in Example 2. Example 2 differs from Example 1 in that in addition to the power change factors at diagnosis d33, the power consumption in the train set at diagnosis d34 is also used in the normal-state power estimation 341. Also, unlike Example 1, in the normal-state power estimation 341, the power change factors at diagnosis d33 and the power consumption in the train set at diagnosis d34 are used in the normal-state model 33, and therefore the operation of the normal-state model generation unit 32 is also different.

[0072] (Operation of normal model generation unit) The operation of the normal state model generating unit 32 in the second embodiment will be described. 6 is a graph 6 summarizing the estimation targets of the model 33 in normal operation according to the second embodiment and information used for estimation. The upper part of the graph 6 shows the time transition of the power consumption in the train set during learning 66 (power consumption y of the first train set during learning), and the lower part of the graph 6 shows the time transition of the power change factors during learning 65 (power change factors x i ) over time.

[0073] The normal state model generation unit 32 uses the learning-time power change factors d31 and the learning-time in-train-set power consumption d32, and, assuming that a time a predetermined time before the first time 62 is defined as a second time 63 and a time before the second time 63 is defined as a third time 64, generates a normal state model 33 that can calculate an estimate of the in-train-set power consumption under normal conditions at the first time 62, using the diagnosis-time power change factors d33 from the third time 64 to the first time 62 and the diagnosis-time in-train-set power consumption d34 from the third time 64 to the second time 63. This will be explained in detail below.

[0074] In the second embodiment, the normal state model 33 is used to estimate the power consumption 61 in the formation at a first time 62. Here, a time a predetermined time before the first time 62 is defined as a second time 63, and a third time 64 further before the second time is defined as a third time.

[0075] At this time, a regression problem is created in which the in-formation power consumption 61 at the first time 62 is used as the dependent variable, and the learning-time power change factors 65 obtained between the third time 64 and the first time 62 and the learning-time in-formation power consumption 66 obtained between the third time 64 and the second time 63 are used as explanatory variables, to create a normal state model 33. In other words, when the data group of the explanatory variables is x, the dependent variable is y, and the model is f, the problem is reduced to finding a function f such that y=f(x).

[0076] The explanatory variable x includes a response variable that is older than the first time (power consumption in the train set during learning 66) because the power consumption in the train set depends on the power consumption in the train set in the past. By including a past response variable in the explanatory variables, it is possible to express a transient response, such as when a specific signal is input and the response variable gradually changes and reaches a certain value.

[0077] The model can be estimated by using a dataset that compiles (x, y) for multiple first times and solving a regression problem using a method classified as machine learning, such as a gradient boosting algorithm for decision trees or an autoencoder.

[0078] Figure 7 is Table 7, which shows an example of a data set. Table 7 is a data set in which (x, y) is organized for each index i1 corresponding to the first time 62. An index is integer information introduced in a program to perform array or matrix processing. There is a one-to-one correspondence between an index and a time, and the index increases as the time increases.

[0079] In Table 7, y(i1) is the objective variable y corresponding to index il, i.e., the power consumption in the train during learning, and xj(i) (i = i1, ..., i2, ..., i3, j = 1, 2, ..., N) is the value of the j-th learning power change factor j corresponding to index i. i2 is the index corresponding to the second time 63, and i3 is the index corresponding to the third time 64.

[0080] As shown in the table, for each value of index i1, the explanatory variables (x1(i3), ..., x1(i1), ..., xN(i3), ..., xN(i1), y(i3), ..., y(i2)) corresponding to the objective variable y(i1) at i1 are organized. By using this format, it becomes possible to apply the above-mentioned machine learning techniques.

[0081] Furthermore, for a predetermined time interval set between the first time 62 and the second time 63, the normal state model generation unit 32 sets a value in the normal state model 33 such that the in-train-piece power consumption at diagnosis at the second time 63 is not duplicated (copied) into the in-train-piece power consumption at diagnosis at the first time 62. Specifically, when the estimation method in the second embodiment is adopted, if the time interval between the first time 62 and the second time 63 is short, such as one or two seconds, an inaccurate model may be generated in which the in-train-piece power consumption obtained immediately before the first time 62 is copied and output as the estimation result for the first time 62. This is because, in the process of model estimation, the amount of change in in-train-piece power consumption is small over a short time, and therefore the method of copying the value at the second time 63 may be selected as the most appropriate solution.

[0082] In order to avoid the above-mentioned copy event, it is important to devise a method such that the interval between the first time 62 and the second time 63 is equal to or greater than a predetermined time interval. By doing so, the power consumption in the formation changes by a certain amount from the second time 63 to the first time 62, and it is possible to suppress the phenomenon in which a method of copying the value at the second time 63 is selected as the optimal solution. The predetermined time interval can be, for example, 20 seconds, 30 seconds, or the like, but other values ​​are also acceptable. Multiple time intervals are tried and a value that does not cause a copy event is selected.

[0083] The normal state model generation unit 32 creates a predetermined data set from the power change factors at diagnosis d33 and the in-unit power consumption at diagnosis d34 so as not to copy data for the generated normal state model 33. The power diagnosis unit 34 performs normal state power estimation 341 using the normal state model 33 and the predetermined data set to estimate the normal state power.

[0084] (Flowcharts of Examples 1 and 2) FIG. 8 is a flowchart of the first and second embodiments. The normal state model generation unit generates a normal state model 33 using the learning-time in-train set power consumption d32 and learning-time power change factors d31 acquired during the learning data accumulation period (step S1). Next, it is determined whether the normal state model 33 satisfies the conditions for use (step S2). The conditions for use include the fact that the diagnosis-time power change factors d33 and the diagnosis-time in-train set power consumption d34 are acquired at discrete times, and the increase amount ΔE i If the use condition is satisfied (Yes in step S2), the power diagnosis unit 34 estimates the normal power using the normal model 33 (step S10) and determines whether there is a sign of failure based on the difference (step S11). The process when the use condition is satisfied corresponds to the diagnosis method of the first embodiment.

[0085] On the other hand, if the usage conditions are not met (No in step S2), the normal state model generation unit 32 creates a predetermined data set from the diagnosis-time power change factors d33 and the diagnosis-time in-set power consumption d34 (step S20). The power diagnosis unit 34 estimates the normal state voltage and power using the normal state model 33 and the predetermined data set (step S21), and determines the margin for failure based on the difference (step S22). The processing when the usage conditions are not met corresponds to the diagnosis method of the second embodiment.

[0086] The conditions for using the normal state model 33 are not limited to the above-mentioned cases, and the conditions can be set according to the type of power change factor, the change in power consumption, and the type of normal state model.

[0087] (Actions and Effects) To construct a normal state model using the method of Example 1, the time series changes of the power change factors during learning d31 and the power consumption in the learning configuration d32 are analyzed, and the power change amount (increase amount) ΔE i and time constant T i It is necessary to estimate the power consumption of the air conditioner 22, and the more learning-time power change factors d31 there are, the greater the labor required for estimation. Furthermore, since the power consumption of the air conditioner 22 tends to increase as the outdoor temperature rises, it is actually necessary to add a correction term to equations (1) and (2) that increases in accordance with the outdoor temperature. Further consideration is required as to what value this correction term should have and what kind of function it should be, which poses a challenge in building an accurate model.

[0088] In the method of the second embodiment, all power change factors are organized into a data frame as explanatory variables, and a normal model can be constructed. Therefore, in addition to the advantages of the method of the first embodiment, the method of the second embodiment has the advantage of reducing the power change amount ΔE i and time constant T i This has the advantage of reducing the amount of analysis required to estimate the power consumption within the train and the amount of work required to analyze the effect of outside temperature on the power consumption within the train. [Example]

[0089] The first and second embodiments are inventions relating to the detection of signs of failure of the air conditioner 22 based on the power consumption within the train. On the other hand, the third embodiment describes a method for detecting signs of failure of individual air conditioners only when inspection is required, by further performing the detection of signs of failure using a temperature sensor only when the train 2 enters the depot, in addition to the contents of the first and second embodiments.

[0090] (composition) FIG. 9 is a diagram showing the configuration of the knitting 2 and the diagnostic device 3c in the third embodiment. In addition to the configurations of Examples 1 and 2, Example 3 adds a warehousing detection unit 35, an inspection control command unit 36, and a temperature diagnosis unit 37. The warehousing detection unit 35 detects the warehousing of train set 2. Furthermore, when the inspection control command unit 36 ​​detects a sign of failure using the power diagnosis unit and detects warehousing, it issues an inspection air-conditioning control command to train set 2. Furthermore, the temperature diagnosis unit 37 detects a sign of failure in the air conditioning equipment of each car using the in-car temperature transition obtained during the implementation period of the inspection air-conditioning control. Although FIG. 7 omits the input information of the normal state model generation unit 32, the input information of the power diagnosis unit 34, and internal processing (learning-time power change factors d31 and learning-time in-train power consumption d32), these are the same as those of Example 1 or Example 2.

[0091] (Failure prediction detection) FIG. 10 is a flowchart of the failure sign detection in the third embodiment. First, the inspection control command unit 36 ​​acquires the failure sign diagnosis result from the power diagnosis unit 34 (S101), and then acquires the warehousing determination result from the warehousing detection unit 35 (S102). The failure sign diagnosis result can be acquired, for example, by the failure sign diagnosis of the first embodiment (step S11) or the failure sign diagnosis of the second embodiment (step S22) shown in FIG. 8. Next, the inspection control command unit 36 ​​determines whether or not there is a failure sign from the power diagnosis unit 34, and whether or not the warehousing detection unit 35 determines whether or not the train set 2 has been warehousing when it has stopped at a predetermined position, for example, in a pit at a rolling stock depot (S103). If it is determined in S103 that there is a failure sign and that the train set has been warehousing (Yes in S103), the inspection control command unit 36 ​​issues an inspection air conditioning control command to the train set 2 (S104). After train set 2 has performed inspection air-conditioning control (S105), the temperature diagnosis unit 37 performs an air-conditioning performance evaluation using the interior temperature for each car in train set 2, and outputs the failure sign detection results for each car (S106). Note that if no failure signs are detected or if train set 2 is not determined to have entered the warehouse (No in S103), the failure sign detection ends.

[0092] Each functional block of the diagnosis device 3c will be described with reference to Figure 9. The warehousing detection unit 35 has a function of detecting that the train set 2 has stopped at a predetermined position. The predetermined position is, for example, a stopping position in a pit at a rolling stock depot. The warehousing detection unit 35 also acquires the current kilometer position of the train set 2 from the data accumulation unit 31, and detects that the train set has entered the warehousing when the kilometer position becomes close to a predetermined value, for example, the kilometer position of the stopping position at the rolling stock depot.

[0093] If the track number information of train set 2 can be acquired from the data accumulation unit 31, it is possible to determine whether or not train set 2 is in a pit from the track number information, and one method is to detect that train set 2 has entered warehousing only when it has entered a pit track number. Since the amount of solar radiation in pits is small, if a criterion is adopted that detects that train set 2 has entered warehousing when it is in the pit, it is possible to suppress the rise in the temperature inside the car due to the amount of solar radiation and the resulting decline in the detection performance of the temperature diagnosis unit 37.

[0094] The inspection control command unit 36 ​​has a function of issuing an inspection air conditioning control command to the formation 2 when the power diagnosis unit 34 detects a sign of failure and the warehousing detection unit 35 detects warehousing. If air conditioning control is possible using the formation information transmission device 21, one method of implementing inspection air conditioning control is to send an air conditioning control command to the formation information transmission device 21, but if individual air conditioners 22 are capable of receiving inspection air conditioning control commands, the inspection air conditioning control command may be sent to the individual air conditioners.

[0095] Here, the inspection air conditioning control is a control that causes a phenomenon in which differences in interior temperatures due to a decrease in cooling performance are likely to appear, in order to detect a decrease in cooling performance associated with signs of a malfunction using an interior temperature sensor.

[0096] Methods for controlling air conditioning for inspection have already been proposed, and existing methods can be used. For example, a method has been proposed in which air conditioning control is started when train set 2 begins operation, and the degree of deterioration in cooling performance is evaluated from the degree of temperature drop during air conditioning control.

[0097] In the third embodiment, attention is focused on the fact that a decline in cooling performance slows the rate at which the interior temperature of a vehicle decreases, and an example of detecting signs of failure based on the interior temperature is shown. After receiving an air-conditioning control command for inspection, train set 2 estimates the interior temperature under normal conditions by using the diagnostic interior temperature when the air-conditioning compressor is operating, which is obtained when the air-conditioning compressor is stopped for a predetermined time and then operated for a predetermined time, and a normal temperature model. If the temperature deviation obtained by subtracting the estimated interior temperature under normal conditions from the diagnostic interior temperature at a certain time when the air-conditioning compressor is operating exceeds a predetermined value, it is determined that a sign of failure has occurred in the air-conditioning device of the vehicle corresponding to the diagnostic interior temperature at the certain time. This will be explained in detail below.

[0098] 11 is a graph 9 showing an example of an air-conditioning control method and a cooling performance evaluation method in Example 3. Graph 9 shows data corresponding to a specific car in formation 2, with the upper graph of graph 9 showing the time transition of the car interior temperature 72 and the lower graph 9 showing the number of operating air-conditioning compressors (CPs). It is also assumed that each air-conditioning device is equipped with four air-conditioning CPs.

[0099] In graph 9, when formation 2 receives an inspection air conditioning control command at time tf, all air conditioning CPs are shut down until time ts. After all units are shut down, the air conditioning CPs are not operating, meaning that no air conditioning is being performed, and therefore if the outside temperature is higher than the interior temperature 72, the interior temperature 72 will rise. Next, after time ts, the number of operating air conditioning CPs is changed to four, starting air conditioning control, and after waiting until time te, data acquisition for the inspection is completed. In other words, the inspection period 71 is defined as the time from time tf to time te, and inspection air conditioning control is carried out.

[0100] Next, a temperature deviation 74 is calculated by subtracting the normal interior temperature 73 from the acquired interior temperature 72. If the temperature deviation 74 at time te exceeds a predetermined value, it is determined that the temperature drop is slowing down due to a decline in cooling performance, and a malfunction is considered to have occurred.

[0101] One method for estimating the normal interior temperature 73 is to obtain data by performing test air conditioning control multiple times during the learning data accumulation period, and then estimate the temperature based on a normal temperature model constructed based on the obtained data.

[0102] Regarding the method of generating the normal temperature model, for example, it is possible to adopt a method of constructing the normal model generation unit 32 of Example 2 by replacing the power change factor during learning d31 with the in-car temperature change factor during learning, and the power consumption within the train during learning d32 with the in-car temperature during learning.

[0103] The factors that cause changes in the vehicle interior temperature during learning are a group of information that affect the vehicle interior temperature 72. For example, they include at least information on the number of operating air conditioning compressors in the air conditioner, door opening / closing information, and outside air temperature.

[0104] The interior temperature 72 during learning is the interior temperature corresponding to the temperature change factor during learning. The interior temperature during learning is the interior temperature at the same time as the time of the temperature change factor during learning.

[0105] In addition, the temperature inside the vehicle under normal conditions can be estimated by replacing the power change factor at diagnosis d33 with the temperature change factor at diagnosis, the power consumption in the train at diagnosis d34 with the temperature inside the vehicle under diagnosis, and the normal condition model 33 with the normal condition temperature model in the normal condition power estimation 341 of Example 2.

[0106] The factors that cause changes in the vehicle interior temperature during diagnosis are a group of information that affect the vehicle interior temperature 72. For example, they include at least information on the number of operating air conditioning compressors in the air conditioner, door opening / closing information, and outside air temperature.

[0107] The vehicle interior temperature at the time of diagnosis is the vehicle interior temperature 72 corresponding to the temperature change factor at the time of diagnosis. The vehicle interior temperature 72 at the time of diagnosis is the vehicle interior temperature at the same time as the temperature change factor at the time of diagnosis.

[0108] The normal temperature model is a model for estimating the normal interior temperature when the air conditioner 22 is operating normally at a certain time, based on at least the interior temperature change factors at the time of diagnosis. The normal temperature model may estimate the normal interior temperature not only using the interior temperature change factors at the time of diagnosis, but also using the interior temperature at a time period prior to the time of diagnosis.

[0109] (Actions and Effects) In Examples 1 and 2, failure signs are detected based on the power consumption within the formation, and therefore, failure signs are detected for one of the air conditioning units within the formation, and it is difficult to detect failure signs for individual air conditioning units. On the other hand, in Example 3, failure signs are detected based on information from interior temperature sensors installed in each car within formation 2, and therefore failure signs can be detected for each of the air conditioning units 22 installed in each car within formation 2.

[0110] Furthermore, by limiting the inspection period to specific conditions, such as inside a pit, it is possible to minimize the impact on the temperature inside the vehicle caused by unmeasurable information such as sunlight and the opening and closing of windows, thereby maintaining a high level of performance in detecting signs of failure based on the temperature inside the vehicle.

[0111] Furthermore, by implementing inspection air conditioning control after entry into storage only when a failure sign is detected by failure sign detection based on power consumption within the train, it is possible to avoid performing unnecessary inspection air conditioning control when there are no failure signs, thereby reducing inspection costs.

[0112] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the gist of the present invention.

[0113] The following are examples of possible embodiments of the present invention, but the present invention is not limited to these. (Aspect 1) a power diagnosis unit that detects a sign of failure in any of the air conditioning units in the train set based on the power consumption in the train set; a data storage unit that stores information acquired by vehicles included in the formation; a normal state model generation unit that generates a normal state model when the air conditioner is normal; A diagnostic device comprising: The power diagnosis unit An estimated value of the power consumption in the formation under normal conditions is estimated by using the diagnostic power change factors and the normal condition model; determining that a failure symptom has occurred in one of the air conditioning units in the train set when a deviation obtained by subtracting the power consumption in the train set at the time of diagnosis from the estimated value of the power consumption in the train set under normal conditions exceeds a predetermined value; A diagnostic device comprising: (Aspect 2) 2. The diagnostic device of embodiment 1, comprising: The normal state model generation unit Using the factors that change power during learning and the power consumption within the train during learning, determining an increase in power consumption in the learning configuration when a specific learning power change factor transitions from a first state to a second state; generating a normal-state model by calculating the sum of the products of the increase in power consumption in a specific learning-time configuration and the learning-time power change factors at a certain time for all learning-time power change factors; A diagnostic device comprising: (Aspect 3) The diagnostic device according to aspect 1 or aspect 2, The normal state model generation unit Using the factors that change power during learning and the power consumption within the train during learning, determining an increase in power consumption in the learning configuration when a specific learning power change factor transitions from a first state to a second state; Furthermore, the time constant is calculated from the time transition of the power consumption in the learning configuration. generating a normal-state model by calculating the sum of the products of the increase in power consumption in a specific learning-time configuration, the learning-time power change factors at a certain time, and a term representing a unit step transient response obtained from the time constant, for all learning power change factors; A diagnostic device comprising: (Aspect 4) A diagnostic device according to any one of aspects 1 to 3, comprising: The power diagnosis unit estimating an estimated value of power consumption in a train set during normal operation by using the power change factors during diagnosis, the power consumption in the train set during diagnosis, and the normal operation model; A diagnostic device comprising: (Aspect 5) A diagnostic device according to any one of aspects 1 to 4, comprising: The normal state model generation unit Using the factors that change power during learning and the power consumption within the train during learning, a second time being a time a predetermined time before a first time, and a third time being a time before the second time, generating a normal-state model capable of calculating an estimated value of the power consumption in the train set under normal conditions at the first time, using the factors of power change during diagnosis from the third time to the first time and the power consumption in the train set during diagnosis from the third time to the second time; A diagnostic device comprising: (Aspect 6) A diagnostic device according to any one of aspects 1 to 5, comprising: The normal state model generation unit Regarding the predetermined time interval set between the first time point and the second time point, In the normal state model, the power consumption in the diagnosis unit at the second time is set to a value that is not duplicated with the power consumption in the diagnosis unit at the first time; A diagnostic device comprising: (Aspect 7) A diagnostic device according to any one of aspects 1 to 6, comprising: The diagnostic power change factors are: Including air conditioning compressor operation information, compressor operation information, and outside air temperature information; A diagnostic device comprising: (Aspect 8) A diagnostic device according to any one of aspects 1 to 7, comprising: a warehousing detection unit that detects the warehousing of the train set; an inspection control command unit that issues an inspection air conditioning control command to the train set when a failure symptom is detected by the power diagnosis unit and when entering storage is detected; a temperature diagnosis unit that detects signs of failure in the air conditioning device of each vehicle using changes in the vehicle interior temperature obtained during the implementation period of the test air conditioning control; The diagnostic device further comprises: (Aspect 9) A diagnostic device according to any one of aspects 1 to 8, comprising: After receiving the inspection air conditioning control command, The normal interior temperature is estimated by using the diagnostic interior temperature when the air conditioning compressor is operating, which is obtained by operating the air conditioning compressor for a predetermined period of time after stopping it for a predetermined period of time, and the normal temperature model. When a temperature deviation obtained by subtracting an estimated value of the vehicle interior temperature under normal conditions from the vehicle interior temperature at diagnosis at a certain time when the air conditioning compressor is operating exceeds a predetermined value, it is determined that a malfunction symptom has occurred in the air conditioning device of the vehicle corresponding to the vehicle interior temperature at diagnosis at the certain time. A diagnostic device comprising: (Aspect 10) a power diagnosis unit that detects a sign of failure in any of the air conditioning units in the train set based on the power consumption in the train set; a data storage unit that stores information acquired by vehicles included in the formation; a normal state model generation unit that generates a normal state model when the air conditioner is normal; A diagnostic method for a diagnostic device comprising: An estimated value of the power consumption in the formation under normal conditions is estimated by using the diagnostic power change factors and the normal condition model; determining that a failure symptom has occurred in one of the air conditioning units in the train set when a deviation obtained by subtracting the power consumption in the train set at the time of diagnosis from the estimated value of the power consumption in the train set under normal conditions exceeds a predetermined value; A diagnostic method characterized by: (Aspect 11) 11. A diagnostic method according to embodiment 10, comprising: In the normal state model generation unit, Using the factors that change power during learning and the power consumption within the train during learning, a second time being a time a predetermined time before a first time, and a third time being a time before the second time, generating a normal-state model capable of calculating an estimated value of the power consumption in the train set under normal conditions at the first time, using the factors of power change during diagnosis from the third time to the first time and the power consumption in the train set during diagnosis from the third time to the second time; A diagnostic method characterized by: (Aspect 12) 12. The diagnostic method of embodiment 10 or embodiment 11, comprising: The diagnostic device comprises: The system further includes an entry detection unit, an inspection control command unit, and a temperature diagnosis unit, Detecting the entry of the train set into storage, When the power diagnostic unit detects a failure sign and detects that the train has entered storage, an inspection air conditioning control command is issued to the train; A diagnostic method characterized by a temperature diagnostic unit that detects signs of failure in the air conditioning device of each vehicle using changes in interior temperature obtained during the period in which the test air conditioning control is being performed. [Explanation of symbols]

[0114] 1a: A diagram showing the configuration of a train set and a diagnostic device according to a first embodiment. 2. Formation 3, 3a, 3b, 3c...Diagnostic equipment 4. Graph showing information necessary for generating a normal state model according to the first embodiment 5. A graph showing a method for calculating deviation and a specific example of deviation according to the first embodiment 1b: A diagram showing the configuration of a train set and a diagnostic device according to a second embodiment. 6. A graph summarizing the estimation target of the normal model according to Example 2 and the information used for estimation. 7. An example of a data set 8 shows the configuration of a train set and a diagnostic device according to the third embodiment. 9. Graph showing an example of an air conditioning control method and a cooling performance evaluation method according to Example 3 21. Composition information transmitter 22...Air conditioner 23...Internal power output device 24. Power change factor equipment 31 Data storage unit 32. Normal state model generation part 33 Normal model 34 Power Diagnosis Unit 35. Entry detection unit 36 Inspection control command unit 37 Temperature diagnosis section

Claims

1. a power diagnosis unit that detects a sign of failure in any of the air conditioning units in the train set based on the power consumption in the train set; a data storage unit that stores information acquired by vehicles included in the formation; a normal state model generation unit that generates a normal state model when the air conditioner is normal; A diagnostic device comprising: The power diagnosis unit An estimated value of the power consumption in the formation under normal conditions is estimated by using the diagnostic power change factors and the normal condition model; determining that a failure symptom has occurred in one of the air conditioning units in the train set when a deviation obtained by subtracting the power consumption in the train set at the time of diagnosis from the estimated value of the power consumption in the train set under normal conditions exceeds a predetermined value; A diagnostic device comprising:

2. The diagnostic device according to claim 1, The normal state model generation unit Using the factors that change power during learning and the power consumption within the train during learning, determining an increase in power consumption in the learning configuration when a specific learning power change factor transitions from a first state to a second state; generating a normal-state model by calculating the sum of the products of the increase in power consumption in a specific learning-time configuration and the learning-time power change factors at a certain time for all learning-time power change factors; A diagnostic device comprising:

3. The diagnostic device according to claim 1, The normal state model generation unit Using the factors that change power during learning and the power consumption within the train during learning, determining an increase in power consumption in the learning configuration when a specific learning power change factor transitions from a first state to a second state; Furthermore, the time constant is calculated from the time transition of the power consumption in the learning configuration. generating a normal-state model by calculating the sum of the products of the increase in power consumption in a specific learning-time configuration, the learning-time power change factors at a certain time, and a term representing a unit step transient response obtained from the time constant, for all learning power change factors; A diagnostic device comprising:

4. The diagnostic device according to claim 1, The power diagnosis unit estimating an estimated value of power consumption in a train set during normal operation by using the power change factors during diagnosis, the power consumption in the train set during diagnosis, and the normal operation model; A diagnostic device comprising:

5. The diagnostic device according to claim 4, The normal state model generation unit Using the factors that change power during learning and the power consumption within the train during learning, a second time being a time a predetermined time before a first time, and a third time being a time before the second time, generating a normal-state model capable of calculating an estimated value of the power consumption in the train set under normal conditions at the first time, using the factors of power change during diagnosis from the third time to the first time and the power consumption in the train set during diagnosis from the third time to the second time; A diagnostic device comprising:

6. The diagnostic device according to claim 5, The normal state model generation unit Regarding the predetermined time interval set between the first time point and the second time point, In the normal state model, the power consumption in the diagnosis unit at the second time is set to a value that is not duplicated with the power consumption in the diagnosis unit at the first time; A diagnostic device comprising:

7. The diagnostic device according to claim 1, The diagnostic power change factors are: Including air conditioning compressor operation information, compressor operation information, and outside air temperature information; A diagnostic device comprising:

8. The diagnostic device according to claim 1, a warehousing detection unit that detects the warehousing of the train set; an inspection control command unit that issues an inspection air conditioning control command to the train set when a failure symptom is detected by the power diagnosis unit and when entering storage is detected; a temperature diagnosis unit that detects signs of failure in the air conditioning device of each vehicle using changes in the vehicle interior temperature obtained during the implementation period of the test air conditioning control; The diagnostic device further comprises:

9. 9. The diagnostic device according to claim 8, After receiving the inspection air conditioning control command, The normal interior temperature is estimated by using the diagnostic interior temperature when the air conditioning compressor is operating, which is obtained by operating the air conditioning compressor for a predetermined period of time after stopping it for a predetermined period of time, and the normal temperature model. When a temperature deviation obtained by subtracting an estimated value of the vehicle interior temperature under normal conditions from the vehicle interior temperature at diagnosis at a certain time when the air conditioning compressor is operating exceeds a predetermined value, it is determined that a malfunction symptom has occurred in the air conditioning device of the vehicle corresponding to the vehicle interior temperature at diagnosis at the certain time. A diagnostic device comprising:

10. a power diagnosis unit that detects a sign of failure in any of the air conditioning units in the train set based on the power consumption in the train set; a data storage unit that stores information acquired by vehicles included in the formation; a normal state model generation unit that generates a normal state model when the air conditioner is normal; A diagnostic method for a diagnostic device comprising: An estimated value of the power consumption in the formation under normal conditions is estimated by using the diagnostic power change factors and the normal condition model; determining that a failure symptom has occurred in one of the air conditioning units in the train set when a deviation obtained by subtracting the power consumption in the train set at the time of diagnosis from the estimated value of the power consumption in the train set under normal conditions exceeds a predetermined value; A diagnostic method characterized by:

11. The diagnostic method according to claim 10, In the normal state model generation unit, Using the factors that change power during learning and the power consumption within the train during learning, a second time being a time a predetermined time before a first time, and a third time being a time before the second time, generating a normal-state model capable of calculating an estimated value of the power consumption in the train set under normal conditions at the first time, using the factors of power change during diagnosis from the third time to the first time and the power consumption in the train set during diagnosis from the third time to the second time; A diagnostic method characterized by:

12. The diagnostic method according to claim 10, The diagnostic device comprises: The system further includes an entry detection unit, an inspection control command unit, and a temperature diagnosis unit, Detecting the entry of the train set into storage, When the power diagnostic unit detects a failure sign and detects that the train has entered storage, an inspection air conditioning control command is issued to the train; A diagnostic method characterized by a temperature diagnostic unit that detects signs of failure in the air conditioning device of each vehicle using changes in interior temperature obtained during the period in which the test air conditioning control is being performed.

Images