Diagnostic system and learning device

Abstract

Provided are a diagnosis system and a learning device capable of estimating the life period of a motor requiring replacement. A diagnostic system for diagnosing deterioration in insulation performance of a motor includes: a receiving unit that receives an input of a detection value of a leakage current flowing through a closed circuit including a coil of the motor and a ground portion of the motor; and a diagnosis unit that uses a data set in which a detection value of an initial leakage current, which is a leakage current detected at a point in time that becomes a base point, and a detection value of the leakage current received by the reception unit, are associated with each other, and a prediction model for deducing the remaining life of the motor based on the insulation performance from the data set. And outputting information for estimating the remaining life.

Description

Technical Field

[0001] The present disclosure relates to a diagnostic system and a learning device. Background Art

[0002] Patent Document 1 discloses an insulation diagnostic device for diagnosing the insulation performance of a motor. By using this insulation diagnostic device, the insulation performance of the motor can be comprehensively evaluated based on the insulation resistance value, the leakage current, and the measured value of tan δ.

[0003] Patent Document 1: Japanese Patent Application Laid-Open No. 3-163368 Summary of the invention

[0004] Problem that the invention aims to solve

[0005] However, the insulation diagnostic device described in Patent Document 1 cannot quantitatively evaluate how the insulation performance will change in the future. Therefore, it is not possible to evaluate the life span of the motor when replacement is required from the viewpoint of the insulation performance.

[0006] The present disclosure is made to solve the above-mentioned problems. An object of the present disclosure is to provide a diagnosis system and a learning device that can estimate the time when a motor needs to be replaced.

[0007] Solutions for solving problems

[0008] The diagnostic system involved in the present disclosure is a diagnostic system for diagnosing the degradation of the insulation performance of a motor, and comprises: an accepting unit that accepts the input of a detection value of a leakage current flowing through a closed circuit including a coil of a motor and a grounding portion of the motor; and a diagnostic unit that uses a data set that includes a detection value of a leakage current detected at a time point that becomes a base point, i.e., an initial leakage current, and a detection value of the leakage current accepted by the accepting unit in correspondence with each other, and a prediction model for inferring the remaining life of the motor based on the insulation performance from the data set, to output information for estimating the remaining life.

[0009] The learning device involved in the present disclosure is a learning device for generating a prediction model for inferring the remaining life of a motor based on the insulation performance based on the electrical characteristics of a closed circuit including a motor coil and a grounding part of the motor, and comprises: a data acquisition unit, which acquires a first learning data set that includes a detection value of an initial test leakage current flowing through a test circuit that simulates a closed circuit using a model coil that simulates the coil of the motor and a detection value of a test leakage current flowing through the test circuit after the initial test leakage current is detected and after a degradation test is performed on the model coil under specified degradation conditions; and a generation unit, which uses the first learning data set to generate a learned model, i.e., a prediction model, for inferring the remaining life of the motor based on a data set that includes a detection value of an initial leakage current flowing through the closed circuit at a time point that becomes a base point and a detection value of the leakage current flowing through the closed circuit.

[0010] Effects of the Invention

[0011] According to the present disclosure, by using the prediction model, information for estimating the remaining life is output, so that the life period when the motor needs to be replaced can be estimated. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a diagram showing an outline of an elevator apparatus provided with a motor to which the diagnostic system in the first embodiment is applied.

[0013] Figure 2 This is a diagram showing a motor and a diagnostic device to which the diagnostic system in the first embodiment is applied.

[0014] Figure 3 This is a block diagram of the diagnostic system in the first embodiment.

[0015] Figure 4 This is a diagram showing an example of a life function stored in the diagnostic system in the first embodiment.

[0016] Figure 5 This is a side view of a model coil unit for acquiring data learned in the diagnostic system in the first embodiment.

[0017] Figure 6 This is a top view of a model coil unit used to acquire data learned in the diagnosis system in Embodiment 1.

[0018] Figure 7 This is a diagram showing data obtained when the water vapor pressure is changed in a degradation test performed by the diagnostic system in the first embodiment.

[0019] Figure 8 This is a diagram showing data on a thermal shock cycle in a degradation test performed on the diagnostic system in the first embodiment.

[0020] Fig. 9 This is a diagram showing data when temperature conditions are changed in a degradation test performed by the diagnostic system in the first embodiment.

[0021] Fig.10 This is a diagram showing data obtained when the core length of the dummy core is changed in the degradation test performed by the diagnostic system in the first embodiment.

[0022] Fig.11 This is a diagram showing data obtained when the contact area between the dummy core and the dummy coil is changed in the degradation test performed by the diagnostic system in the first embodiment.

[0023] Fig.12 This is a diagram showing the transition of the current flowing through the test circuit in the degradation test performed for the diagnostic system in the first embodiment.

[0024] Fig.13 This is a diagram showing the transition of the current flowing through the test circuit in the degradation test performed for the diagnostic system in the first embodiment.

[0025] Fig.14 This is a diagram showing the transition of the current flowing through the test circuit in the degradation test performed for the diagnostic system in the first embodiment.

[0026] Fig.15 This is a flowchart for explaining an outline of the diagnostic work performed by the diagnostic system in the first embodiment.

[0027] Fig.16 This is a flowchart for explaining an outline of a learning process performed by the first learning device of the diagnostic system in the first embodiment.

[0028] Fig.17 This is a flowchart for explaining an outline of the operation performed by the diagnosis executor of the diagnosis system in the first embodiment.

[0029] Fig.18 This is a block diagram of a modified example of the diagnostic system in the first embodiment.

[0030] Fig.19 This is a diagram showing an example of a learning algorithm using a neural network model used in a modified example of the diagnostic system in the first embodiment.

[0031] Fig. 20 This is a flowchart for explaining an outline of the operation performed by the diagnosis executor of the modified example of the diagnosis system in the first embodiment.

[0032] Fig.21 This is a hardware configuration diagram of the diagnosis executor of the diagnosis system in the first embodiment.

[0033] Fig. 22 This is a block diagram showing another example of the diagnostic system in the first embodiment.

[0034] Fig.23 This is a block diagram of a diagnostic system in the second embodiment.

[0035] Fig.24 This is a block diagram of a diagnostic system in the third embodiment.

[0036] Fig.25 This is a diagram showing the relationship between the operating performance and the leakage current measured in the motor of the hoisting machine in a general elevator device.

[0037] Fig.26 This is a diagram showing the relationship between the operating performance and the leakage current measured in the motor of the hoisting machine in a general elevator device.

[0038] Fig. 27 This is a diagram showing the relationship between the operating performance and the leakage current measured in the motor of the hoisting machine in a general elevator device.

[0039] Fig.28 This is a flowchart for explaining an outline of a learning process performed by the second learning device of the diagnostic system in the third embodiment.

[0040] Fig.29 This is a block diagram of a diagnostic system in a fourth embodiment. DETAILED DESCRIPTION

[0041] The method for implementing the present disclosure is described in accordance with the accompanying drawings. In addition, in each figure, the same or corresponding parts are given the same symbols. The repeated description of the part is appropriately simplified or omitted.

[0042] Implementation method 1.

[0043] Figure 1 This is a diagram showing an outline of an elevator apparatus provided with a motor to which the diagnostic system in the first embodiment is applied. Figure 2 This is a diagram showing a motor and a diagnostic device to which the diagnostic system in the first embodiment is applied.

[0044] exist Figure 1 In the figure, an elevator device 2 is shown as an example of an object to which the diagnostic system 1 is applied. The elevator device 2 is installed in a building 3. A hoistway 4 passes through each floor of the building 3. A machine room 5 is installed just above the hoistway 4. A winch 6 is installed inside the machine room 5. A motor 7 is installed in the winch 6. The motor 7 drives the sheave of the winch 6 as a rotating machine. A car 8 is suspended by a main rope 9 inside the hoistway 4. The car 8 moves up and down inside the hoistway 4 by driving the winch 6.

[0045] The control panel 10 is installed in the machine room 5. The control panel 10 can control the elevator device 2 including the motor 7 as a whole. The control panel 10 stores information such as the model information of the motor 7 and the operation information of the elevator device 2. The remote monitoring device 11 is installed in the machine room 5. The remote monitoring device 11 obtains information such as the operation information from the control panel 10 and stores it. The information center device 12 is installed in an information center 13 that exists in a place far away from the building 3. For example, the information center 13 is a building of a maintenance company of the elevator device 2. The information center device 12 obtains information from the remote monitoring device 11 via the network 14. The information center device 12 can grasp the state of the elevator device 2 based on the obtained information. In addition, the elevator device 2 may be a type in which the machine room 5 is not provided. In this case, for example, the hoist 6, the control panel 10, and the remote monitoring device 11 may be installed in a place other than the machine room 5, such as the inside of the hoistway 4.

[0046] The diagnostic system 1 may also include a permanent measuring device 15, for example. The permanent measuring device 15 is provided in the machine room 5. In particular, the permanent measuring device 15 is provided around the motor 7. The permanent measuring device 15 may also be installed on the hoist 6. The permanent measuring device 15 measures the temperature and humidity of the air as the surrounding environment of the motor 7. The permanent measuring device 15 stores the measured values ​​during the operation of the motor 7. In addition, the above is only an example, and the measuring device provided in the diagnostic system 1 is not limited to whether it is permanent.

[0047] In the elevator device 2, maintenance work is regularly performed at a predetermined cycle such as once a month. The maintenance work is performed by a worker of a maintenance company affiliated with the elevator device 2. For example, a diagnosis work on the insulation performance of the motor 7 is performed in the maintenance work.

[0048] Figure 2 The cross section perpendicular to the rotation axis of the motor 7 diagnosed by the diagnostic operation is shown. For example, the motor 7 is a motor driven by a three-phase alternating current. In particular, the motor 7 may be a low-voltage motor that is not expected to generate partial discharge during operation.

[0049] The motor 7 is an example of a motor to which the diagnostic system 1 can be applied. Therefore, the diagnostic system 1 may not be applied to the motor of the hoist 6 of the elevator device 2, and may not have any motor as long as it generates a magnetic field by passing current through a coil and a core. Figure 2 The construction of the motor 7 is shown.

[0050] The motor 7 has a core 7a and a coil 7b as a stator. The coil 7b is wound around the core 7a with a slot as a structural unit. Although not shown in the figure, the coil 7b includes three-phase coils of U phase, V phase, and W phase. A three-phase alternating current is supplied to the motor 7 from a commercial power supply not shown in the figure. The three-phase alternating current flows through the coil 7b of the U phase, V phase, and W phase respectively, and the permanent magnet as a rotor not shown in the figure rotates. In this way, the motor 7 is driven.

[0051] The motor 7 is provided with three measuring terminals 7c, 7d, and 7e. The measuring terminal 7c is electrically connected to the coil 7b of the U phase. The measuring terminal 7d is electrically connected to the coil 7b of the V phase. The measuring terminal 7e is electrically connected to the coil 7b of the W phase.

[0052] The diagnostic system 1 further includes a diagnostic device 20. The diagnostic device 20 includes two detection terminals 21a and 21b. For example, the diagnostic device 20 has a shape that can be carried by an operator. During the diagnostic operation, the diagnostic device 20 is connected to the control panel 10 and the permanent measuring device 15 in a manner that allows communication. During the diagnostic operation, the detection terminal 21a is electrically connected to one of the three measuring terminals 7c, 7d, and 7e of the motor 7. The detection terminal 21b is electrically connected to the base 7f that supports the motor 7. The base 7f is equivalent to the grounding portion in the internal circuit of the motor 7.

[0053] In the diagnostic operation, a series closed circuit including the diagnostic device 20, the base 7f as the grounding part, and the coil 7b of any one of the UV and W phases is formed. After the closed circuit is formed, the operator operates the diagnostic device 20 in a manner of applying a diagnostic voltage. The diagnostic device 20 applies a DC voltage of 1000 [V] between the detection terminal 21a and the detection terminal 21b as a diagnostic voltage. In addition, the diagnostic voltage may be a DC voltage of 1600 [V] or less, and more preferably a DC voltage of 1000 [V] or less.

[0054] The diagnostic device 20 detects the value of the leakage current, i.e., the direct current, which is the current flowing through the closed circuit by the diagnostic voltage. The diagnostic device 20 obtains the information of the measured value stored in the permanent measuring device 15. The diagnostic device 20 obtains the operation information of the elevator device 2 including the working information of the motor 7 from the control panel 10. The diagnostic device 20 diagnoses the deterioration of the insulation performance of the motor 7, for example, using the value of the detected leakage current (hereinafter also referred to as the detected value of the leakage current), the information of the measured value obtained from the permanent measuring device 15, and the operation information obtained from the control panel 10. In addition, the information used by the diagnostic device 20 only needs to include at least the information of the value of the detected leakage current. The diagnostic device 20 calculates the remaining period until the insulation performance falls below the management level, i.e., the remaining life based on the insulation performance as the result of the diagnosis. For example, the diagnostic device 20 notifies the operator of the remaining life by displaying the calculated remaining life on a display device not shown. In addition, the diagnostic device 20 can also be used to diagnose the deterioration of the insulation performance of the motor 7 via a network, etc. Figure 2 The information center device 12 (not shown) transmits the calculated remaining life information. Figure 2 In the information center 13 (not shown), a replacement plan of the motor 7 corresponding to the remaining life can also be formulated based on the information sent from the diagnostic device 20. After that, the operator performs the same diagnostic operation on the coil 7b of the phase that has not been diagnosed among the UVW phases. When the coil 7b of all phases is diagnosed, the operator removes the two detection terminals 21a and 21b to end the diagnostic operation.

[0055] Next, use Figure 3 A diagnostic system 1 is described.

[0056] Figure 3 is a block diagram of the diagnostic system in Embodiment 1. Figure 3 The core 7a and the coil 7b are not shown in the figure.

[0057] like Figure 3 As shown, the diagnostic device 20 may also include a detector 22 and a diagnostic executor 23 .

[0058] The detector 22 can detect, for example, 100 [nA] or less, that is, 1×10 -7 [A] or less. In addition, if the detector 22 can detect 1×10 -9 [A] or less is more preferable. If the detector 22 can detect 1×10 -11 [A] or less is more preferable. The detector 22 has a current value of 1×10 -11[A] Resolution below. Two detection terminals 21a, 21b are connected to the detector 22. The detector 22 can be electrically connected to the coil 7b of the motor 7 set as the diagnosis object (in the example shown in the present embodiment, the coil 7b of any one of the UVW phases) and the base 7f as the grounding part of the motor 7 via the two detection terminals 21a, 21b. In this example, the detector 22 can apply a diagnostic voltage to a closed circuit including the coil 7b of the motor 7 and the base 7f as the grounding part of the motor 7. In addition, the grounding part of the motor 7 can also be a part other than the base 7f. The detector 22 detects the value of the leakage current of the DC flowing through the closed circuit. The detector 22 can output a signal representing the detection value of the current.

[0059] For example, the detector 22 may detect the value of the leakage current of the direct current flowing through the closed circuit when the diagnostic voltage is applied, and output a signal indicating the detected value (the detected value of the current) in real time. In addition, for example, as a detection process, the detector 22 may store the time corresponding to the value of the leakage current of the direct current flowing through the closed circuit at a predetermined timing (e.g., a certain period) during the application of the diagnostic voltage. Furthermore, the detector 22 may also store the value of the leakage current of the output leakage current value corresponding to the change of 1×10 -11 The current value when [A] is less than or equal to the detection value of the leakage current is taken as the detection value of the leakage current.

[0060] The diagnosis executor 23 can be electrically connected to the control panel 10, the permanent measuring device 15, and the detector 22. Figure 3 The processor and memory (not shown) perform calculation processing, etc. The diagnosis executor 23 has a function of diagnosing the deterioration of the insulation performance of the motor 7 through the calculation processing. The diagnosis executor 23 includes, for example, a storage unit 23a, a receiving unit 23b, a pre-processing unit 23c, an acquisition unit 23d, and a diagnosis unit 23e as a mechanism for performing the processing included in this function.

[0061] The storage unit 23a is a medium for storing information, and stores information required for diagnosis. For example, the storage unit 23a stores information of a prediction model. The prediction model is a calculation model for inferring the remaining life based on the input information. For example, the prediction model is a life function that represents the relationship between "one or more parameters including information indicating the leakage current at the time of diagnosis" and "estimated value of the accumulated elapsed time after considering the deterioration condition of the insulation performance of the motor 7". Here, as information indicating the leakage current at the time of diagnosis, it is not limited to the detected value of the leakage current, and may also include the standard leakage current value described later or other information stored in the storage unit 23a, information that can calculate the detected value of the leakage current or the standard leakage current value (for example, the amount of change in the leakage current from the time point that becomes the base point), etc. That is, by inputting one or more parameters to the life function, it is possible to derive an estimated value of the accumulated elapsed time after considering the deterioration condition of the insulation performance of the motor 7 (hereinafter also referred to as the suspected elapsed time) as a value corresponding to the parameter. The storage unit 23a may also store the estimated value of the accumulated elapsed time corresponding to the management upper limit value of the detected value of the leakage current in the life function as the life elapsed time. That is, the storage unit 23a may store the accumulated elapsed time considered to have reached the management upper limit value of the detection value of the leakage current as the life elapsed time. In addition, the storage unit 23a may store a plurality of life functions as information of the prediction model. For example, the function types of the plurality of life functions, the values ​​of the internal constants, etc. are different from each other. In this case, the storage unit 23a may store the life elapsed time corresponding to each of the plurality of life functions.

[0062] The storage unit 23a may store information on the leakage current detected by the detector 22 at a time point that becomes a base point, that is, the detection value of the initial leakage current. For example, the initial leakage current is a value detected just before the hoisting machine 6 including the motor 7 is shipped from the assembly factory as a time point that becomes a base point. Alternatively, for example, the initial leakage current may be a value detected after the motor 7 is installed in the elevator device 2 and before the official operation is started as a time point that becomes a base point. In addition, for example, the initial leakage current may be a value detected after a certain period of time has passed since the start of the operation, such as when the initial diagnostic work is performed after the motor 7 is installed in the elevator device 2 and officially starts to work as a time point that becomes a base point. In addition, the storage unit 23a may temporarily or permanently store information accepted by the acceptance unit 23b. Here, the information accepted by the acceptance unit 23b may include a signal indicating the detection value of the leakage current including the initial leakage current from the detector 22, the operation information of the elevator device 2, and the information on the structural characteristics of the motor 7.

[0063] The receiving unit 23b receives input of a signal indicating a detected value of the current from the detector 22. The receiving unit 23b may receive input of operation information of the elevator device 2 from the control panel 10. In addition, the receiving unit 23b may receive input of information on the structural characteristics of the motor 7 from the control panel 10. The receiving unit 23b may receive input of information on the measured value from the permanent measuring device 15. The receiving unit 23b may acquire operation information and information on the structural characteristics of the motor 7 from the control panel 10. In addition, the receiving unit 23b may acquire information on the measured value from the permanent measuring device 15.

[0064] The preprocessing unit 23c calculates the value of the water vapor pressure of the surrounding environment of the motor 7 during the period in which the measured value is measured based on the information of the measured value from the permanent measuring device 15. The value of the water vapor pressure is the value of the partial pressure of water vapor in the air calculated based on the temperature and humidity of the air in the surrounding environment. For example, the value of the water vapor pressure may also be the average value of the partial pressure of water vapor during the period in which the measured value is measured.

[0065] The preprocessing unit 23c normalizes the received leakage current detection value based on the information of the structural characteristics of the motor 7. In this example, for example, the information of the structural characteristics of the motor 7 may include information such as the core length, which is the length of the core 7a in the direction of the rotating shaft, the contact circumference of the core 7a and the coil 7b per slot, the number of slots corresponding to the coils 7b of the UVW phases in the motor 7, and the thickness of the insulating layer of the coil 7b. For example, the preprocessing unit 23c calculates the total contact area between the coil 7b and the core 7a in the measured phase using the core length of the core 7a, the contact circumference of the core 7a and the coil 7b per slot, and the number of slots. The preprocessing unit 23c divides the leakage current detection value by the total contact area between the coil 7b and the core 7a to calculate the leakage current per unit area, and sets it as a standard leakage current value, that is, a normalized leakage current detection value. In addition, the preprocessing unit 23c may normalize the initial leakage current detection value stored in the storage unit 23a in the same way as the leakage current detection value.

[0066] The acquisition unit 23d acquires information such as information stored in the storage unit 23a, information accepted by the acceptance unit 23b, and information acquired by the preprocessing unit 23c through calculation, and creates a data set for input into the prediction model. In the data set, information corresponding to the input parameters accepted by the prediction model is included in correspondence with each other. For example, in the data set, the detection value of the initial leakage current is included in correspondence with the detection value of the leakage current. In addition, in addition to the detection value of the initial leakage current, the data set may also include a standardized detection value of the initial leakage current, that is, a standard initial leakage current value, or include a standardized detection value of the initial leakage current, that is, a standard initial leakage current value instead of the detection value of the initial leakage current. In addition, in addition to the detection value of the leakage current, the data set may also include a standardized detection value of the leakage current, that is, a standard leakage current value, or include a standardized detection value of the leakage current, that is, a standard leakage current value instead of the detection value of the leakage current. In addition, in the data set, in addition to the initial leakage current detection value and the leakage current detection value, the difference value (i.e., the change amount) between the initial leakage current detection value and the leakage current detection value may be included, or the difference value (i.e., the change amount) between the initial leakage current detection value and the leakage current detection value may be included instead of the initial leakage current detection value and the leakage current detection value. In addition, in the data set, in addition to the standard initial leakage current value and the standard leakage current value, the difference value (i.e., the change amount) between the standard initial leakage current value and the standard leakage current value may be included, or the difference value (i.e., the change amount) between the standard initial leakage current value and the standard leakage current value may be included instead of the standard initial leakage current value and the standard leakage current value. In the data set, the value of the water vapor pressure calculated by the preprocessing unit 23c may be further included in correspondence with the leakage current detection value. In the data set, at least one of the average value of the voltage applied to the motor 7 during the operation of the motor 7 and the number of times the voltage is applied may be further included in correspondence with the leakage current detection value. In the data set, at least one of the average temperature of the motor 7 in operation, the number of times the thermal shock is applied to the motor 7, and the average temperature of the thermal shock when the thermal shock is applied to the motor 7 may be further included in correspondence with the detection value of the leakage current. In the data set, at least one of the transformation start temperature Tg of the resin, the degree of hydrolysis of the resin, the thermal decomposition start temperature Td of the resin, the average molecular weight of each molecule constituting the resin, and the dielectric constant of the resin may be further included as physical properties of the resin constituting the insulating layer of the coil 7b in correspondence with the detection value of the leakage current. In the data set, at least one of the core length of the core 7a, the space factor of the coil 7b in the motor 7, and the wire diameter of the coil 7b may be further included in correspondence with the detection value of the leakage current.

[0067] The diagnosis unit 23e outputs information for estimating the remaining life using the prediction model and the data set. In the example shown in this embodiment, the diagnosis unit 23e outputs the remaining life itself as information for estimating the remaining life.

[0068] The following is an example in which the diagnosis unit 23e outputs the remaining life after inferring the pseudo elapsed time indicating the degree of progress of the deterioration of the insulation performance of the motor 7 using the prediction model. In this example, the diagnosis unit 23e includes an inference unit 23f and a calculation unit 23g.

[0069] The inference unit 23f infers the suspected elapsed time by inputting each input parameter generated based on the information included in the data set into the prediction model. The prediction model here may be, for example, a life function that represents the relationship between "one or more parameters including information representing the leakage current at the time of diagnosis" and "an estimated value of the accumulated elapsed time after considering the deterioration state of the insulation performance of the motor 7". The information representing the leakage current at the time of diagnosis may be, for example, a detected value of the initial leakage current, a detected value of the leakage current, etc. In addition, the information representing the leakage current at the time of diagnosis may also be a standard initial leakage current value, a standard leakage current value, a difference value between the detected value of the initial leakage current and the detected value of the leakage current, a difference value between the standard initial leakage current value and the standard leakage current value, etc. The suspected elapsed time is an estimated value of the accumulated elapsed time corresponding to the input parameter in the life function. The suspected elapsed time is often different from the accumulated elapsed time of the actual operation of the motor 7. The suspected elapsed time is an index value representing the degree of progress of the deterioration of the insulation performance of the motor 7. In this example, when the prediction model includes a plurality of life functions, the inference unit 23f may select any life function based on the input parameters included in the data set. Specifically, the inference unit 23f may also use the water vapor pressure value included in the data set to select a life function used to infer the suspected elapsed time. For example, the water vapor pressure value may be pre-set to be divided into a plurality of value ranges, and the diagnostic system 1 may store a plurality of life functions corresponding to the plurality of value ranges.

[0070] The calculation unit 23g calculates and outputs the remaining life by subtracting the life elapsed time stored in the storage unit 23a based on the pseudo elapsed time inferred by the inference unit 23f, for example.

[0071] In addition, the calculation unit 23g may also calculate the ratio of the actual accumulated elapsed time to the suspected elapsed time, and consider the value after taking into account the influence of the ratio in the difference between the life elapsed time and the suspected elapsed time as the remaining life. In this case, the calculation unit 23g calculates the increase or decrease rate of the ratio of the actual accumulated elapsed time to the suspected elapsed time, that is, the increase or decrease rate of the degradation speed relative to the actual elapsed time. The calculation unit 23g calculates the first remaining time by subtracting the suspected elapsed time from the life elapsed time. The calculation unit 23g multiplies the first remaining time by the increase or decrease rate to calculate the second remaining time, and outputs the second remaining time as the remaining life. Specifically, for example, when the suspected elapsed time is 1.1 times the length of the actual elapsed time, the calculation unit 23g regards the increase or decrease rate as 1.1 and outputs 1.1 times the first remaining time as the remaining life. In addition, the calculation unit 23g may also output the shorter time of the first remaining time and the second remaining time as the remaining life.

[0072] In this way, the diagnosis section 23 e of the diagnosis executor 23 outputs the remaining life itself as information for estimating the remaining life using the prediction model.

[0073] The diagnostic system 1 may further include a first learning device 30. For example, the first learning device 30 is provided as one of the structures of the information center device 12. The first learning device 30 may also generate a prediction model by performing a machine learning method, such as supervised learning. The first learning device 30 includes a first model storage unit 30a, a first data acquisition unit 30b, and a first generation unit 30c.

[0074] The first model storage unit 30a stores information of the learned prediction model. For example, until the diagnosis operation is performed, the information of the prediction model stored in the first model storage unit 30a is pre-stored in the storage unit 23a of the diagnosis device 20. That is, the prediction model stored in the storage unit 23a is the same as the prediction model stored in the first model storage unit 30a.

[0075] The first data acquisition unit 30 b acquires a first learning data set for generating a prediction model from a first learning database (not shown). For example, the first learning database is stored in a storage medium included in the information center device 12 .

[0076] In the first learning data set, the values ​​measured in the degradation test implemented in advance, the values ​​set in the degradation test, etc. are included in correspondence with each other. In the degradation test, by placing the model coil unit simulating a part of the core 7a and the coil 7b under the prescribed degradation conditions, the degradation of the insulation performance of the model coil unit is accelerated. In the degradation test, the test leakage current in the test circuit including the model coil unit is measured. The test circuit is a circuit that simulates a closed circuit formed during the diagnostic operation. By applying a test voltage to both ends of the model coil unit in the test circuit, a test leakage current simulating the leakage current in the closed circuit flows through the test circuit. In the degradation test, the initial test leakage current before the model coil unit is placed in the degradation condition, the test leakage current measured during the period of being placed in the degradation condition, and the test leakage current after being placed in the degradation condition are measured more than once. In addition, with respect to the degradation condition, multiple degradation conditions in which the set value of the item is changed can be included for one item. Specifically, for example, when water vapor partial pressure is an item of degradation condition, the degradation test is performed under a plurality of water vapor partial pressures, and a plurality of test leakage currents corresponding to the plurality of water vapor partial pressures are detected.

[0077] For example, in the first learning data set, the detection value of the initial test leakage current is included in correspondence with the detection value of the test leakage current. In addition, in the first learning data set, the detection value of the standardized initial test leakage current may be included in place of the detection value of the initial test leakage current. In the first learning data set, the detection value of the standardized standard test leakage current may be included in place of the detection value of the test leakage current. In the first learning data set, the value of the water vapor pressure in the surrounding environment of the model coil unit may be further included as a degradation condition in correspondence with the detection value of the test leakage current. In the first learning data set, at least one of the average value of the test voltage applied to the model coil unit and the number of times the test voltage is applied may be further included as a degradation condition in correspondence with the detection value of the test leakage current. In the first learning data set, at least one of the average temperature of the model coil unit in the degradation test, the number of times the thermal shock is applied to the model coil unit, and the average temperature of the thermal shock when the thermal shock is applied to the model coil unit may be further included as a degradation condition in correspondence with the detection value of the test leakage current. In the first learning data set, at least one of the transformation start temperature Tg of the resin, the degree of hydrolysis of the resin, the thermal decomposition start temperature Td of the resin, the average molecular weight of each molecule constituting the resin, and the dielectric constant of the resin may be further included as physical properties of the resin constituting the insulating layer of the model coil included in the model coil unit, corresponding to the detection value of the test leakage current. In the first learning data set, at least one of the core length of the model core included in the model coil, the suspected space factor of the model coil, and the wire diameter of the model coil may be further included as conditions simulating the physical properties of the motor 7, corresponding to the detection value of the test leakage current.

[0078] For example, Table 1 below shows an example of a portion of the first learning data set. In Table 1, ignored conditions are indicated by blank columns. In Table 1, some numerical values ​​are expressed using exponential notation. Specifically, for example, 5.00E-12 means 5.00×10 -12 .

[0079] [Table 1]

[0080]

[0081] In the first learning data set in Table 1, the numerical values ​​in the same row correspond to each other as one combination.

[0082] The first generation unit 30c learns a prediction model based on the first learning data set acquired by the first data acquisition unit 30b. That is, based on the first learning data set, one or more life functions are generated as prediction models. For example, the first generation unit 30c generates a life function of a curve graph representing the shortest distance between the relationship with each combination included in the first learning data set. In addition, such a learning algorithm executed by the first generation unit 30c may be a known algorithm. The first generation unit 30c stores the information of the generated prediction model in the first model storage unit 30a.

[0083] Next, use Figure 4 An example of a life span function represented by a prediction model will be described.

[0084] Figure 4 This is a diagram showing an example of a life function stored in the diagnostic system in the first embodiment.

[0085] Figure 4 A graph G1 is a two-dimensional representation of the life function in terms of leakage current and the estimated value of the accumulated elapsed time. The vertical axis is the leakage current value. The horizontal axis is the estimated value of the accumulated elapsed time. Alternatively, the life function may be a function that can be represented in an n-dimensional space corresponding to the input parameters. Figure 4 A graph expressing the two-dimensional space thereof is shown in FIG.

[0086] The dotted line L1 is the management upper limit value of the leakage current. The estimated value of the accumulated elapsed time corresponding to the management upper limit value of the leakage current on the graph G1 is the life elapsed time. For example, when the detection value of the leakage current is I1, the estimated value of the accumulated elapsed time corresponding to I1 on the graph G is the suspected elapsed time. Figure 4 As shown, the remaining life is the period from the estimated value of the pseudo elapsed time to the life operation.

[0087] Next, use Figure 5 and Figure 6 Description of the model coil unit.

[0088] Figure 5 This is a side view of a model coil unit for acquiring data learned in the diagnostic system in the first embodiment. Figure 6 This is a top view of a model coil unit used to acquire data learned in the diagnosis system in Embodiment 1.

[0089] like Figure 5 and Figure 6 As shown, the model coil unit 40 simulates, for example Figure 2The core 7a and coil 7b shown are coils of the same type as the core 7a and coil 7b. For example, the same type of coil refers to a coil that has a similar shape relative to the core 7a and coil 7b. In addition, the same type of coil may also be a coil that has a space factor that is similar to the space factor of the core 7a and coil 7b.

[0090] The dummy coil unit 40 includes a dummy core 41, a dummy coil 42, insulating paper 43, and a terminal 44. The dummy coil unit 40 simulates two slots of the coil 7b and includes two dummy coils 42 with the same structure.

[0091] The model core 41 is formed of a raw material simulating the core 7a. The model core 41 has a core length Lc in the long side direction in the same direction as the core 7a. The model coil 42 is composed of a raw material simulating the coil 7b. The model coil 42 is annular. Two sides of the model coil 42 are covered by the model core 41. In this way, in the model coil unit 40, a state in which the coil 7b is wound around the core 7a is simulated. Two model coil units 40 are arranged in an array. Insulating paper 43 is sandwiched between the portions of the two model coils 42 that are not covered by the model core 41. The two model coils 42 are insulated by the insulating paper 43. The terminal 44 is provided at a position simulating the state in which the model coil 7b is wound around the core 7a. Figure 5 and Figure 6 The positions of the measurement terminals 7c, 7d, and 7e are not shown. The terminal 44 is electrically connected to the model coil 42.

[0092] The test circuit is formed by simulating a closed circuit. Specifically, for example, the test detector is electrically connected in such a way that a test voltage can be applied to the terminal 44 and a part of the model core 41. The test detector may also have a Figure 5 and Figure 6 For example, the current measured by the test detector may be changed by 1×10 -11 The current value below [A] is regarded as the detection value of the test leakage current.

[0093] Next, use Figures 7 to 11 A part of the results of the degradation test will be described.

[0094] Figure 7 This is a diagram showing data obtained when the water vapor pressure is changed in a degradation test performed by the diagnostic system in the first embodiment. Figure 8 This is a diagram showing data on a thermal shock cycle in a degradation test performed on the diagnostic system in the first embodiment. Fig. 9 This is a diagram showing data when temperature conditions are changed in a degradation test performed by the diagnostic system in the first embodiment. Fig.10This is a diagram showing data obtained when the core length of the dummy core is changed in the degradation test performed by the diagnostic system in the first embodiment. Fig.11 This is a diagram showing data obtained when the contact area between the dummy core and the dummy coil is changed in the degradation test performed by the diagnostic system in the first embodiment.

[0095] Figures 7 to 11 A graph showing the relationship between the change in time and the test leakage current based on each degradation condition is shown. In each graph, the vertical axis represents the test leakage current [A].

[0096] exist Figure 7 In the graph, the horizontal axis is the number of days of degradation [days] that have passed under a given degradation condition. Point groups 1, 2, 3, 4, 5, and 6 of the graph represent data with different water vapor pressures as degradation conditions. Figure 7 It is known that under a water vapor pressure of a certain level or more, the test leakage current increases in a relatively short period of time such as three days. The increase in the test leakage current indicates that the insulation performance of the dummy coil 42 has deteriorated.

[0097] exist Figure 8 In the graph, the horizontal axis is the number of cycles of thermal shock, which indicates the number of times the thermal shock is applied to the model coil unit 40. The conditions of thermal shock are the same for point groups 1, 2, 3, and 4 in the curve. The initial test leakage currents of point groups 1, 2, 3, and 4 in the curve are different. Figure 8 It can be seen that, in any model coil unit 40 showing any initial test leakage current, the test leakage current increases when the thermal shock is applied.

[0098] exist Fig. 9 In the graph, the horizontal axis is the number of days of degradation [days] that have passed under a given degradation condition. Point groups 1, 2, and 3 of the graph represent data with different temperatures of the atmosphere as the degradation condition. Fig. 9 It can be seen that the test leakage current increases at a temperature above a certain level.

[0099] from Figures 7 to 9 It can be seen that the insulation performance of the model coil unit 40 deteriorates under any test condition. Therefore, it can be seen that by inputting the value of water vapor pressure, the number of thermal shocks, and the temperature into the prediction model, the accuracy of the degradation level estimated by the prediction model is improved. In particular, when the condition of water vapor pressure is added, the insulation performance of the model coil unit 40 deteriorates significantly compared to the condition of, for example, adding a temperature change. Therefore, in order to improve the accuracy of the prediction model, the prediction model preferably uses the value of water vapor pressure as an input parameter.

[0100] exist Fig.10In FIG. 1 , the horizontal axis is the core length [mm] of the model core 41. The point group of the graph represents data with the same or different core lengths as test conditions. Fig.10 It can be seen that the longer the core length is, the more the test leakage current increases.

[0101] exist Fig.11 The unit of the vertical axis is [nA]. The horizontal axis is the contact area [m 2 The point groups in the curve graph represent data with the same or different contact areas as test conditions. Fig.11 It can be seen that the larger the contact area, the greater the test leakage current.

[0102] from Fig.10 and Fig.11 It can be seen that it is desirable to utilize the leakage current for the remaining life diagnosis after being normalized according to the structural characteristics of the core 7a and the coil 7b.

[0103] Next, use Figure 12 to Figure 14 The conditions for determining the detection value adopted as the leakage current will be described.

[0104] Figure 12 to Figure 14 This is a diagram showing the transition of the current flowing through the test circuit in the degradation test performed for the diagnostic system in the first embodiment.

[0105] The determination condition of the detection value adopted as the leakage current during the diagnosis operation can be determined based on the temporal transition of the test leakage current in the degradation test. Figure 12 to Figure 14 This is a graph showing the change in the current value flowing through the test circuit. The horizontal axis is the elapsed time [minutes] and the vertical axis is the test leakage current [A].

[0106] exist Figure 12 to Figure 14 In the test circuit, a test voltage is applied to the test circuit at time point P. Immediately after the test voltage is applied, an excessive current flows through the test circuit. After that, the current value gradually decreases toward the equilibrium value, which is the ideal value of the test leakage current. In the past, the current value 10 minutes after the voltage was applied was adopted as the detection value of the leakage current as a value roughly close to the equilibrium value.

[0107] On the other hand, Figure 12 to Figure 14 As shown in the area inside the dotted line Q in the figure and in Table 2 below, at a time point of about 10 minutes after the test voltage was applied, the change in the current value during one minute is approximately 1×10 -11[A] or less. Although not shown in the figure, range 1 and range 2 in Table 2 are time ranges of about 10 minutes from the time the test voltage is applied. That is, from a quantitative point of view, it is preferable to use the change in the current value during 1 minute as the determining condition for the detection value of the test leakage current, rather than the time from the time the test voltage is applied. In addition, it can be seen that before 10 minutes have passed since the test voltage was applied, the change in the current value during 1 minute can be 1×10 -11 Therefore, by setting the change in the current value during one minute to 1×10 -11 [A] The following conditions can be used to determine the degradation test efficiently. In addition, by setting the change in the current value during one minute to 7×10 -12 [A] The following determination conditions can be used to determine the detection value of the test leakage current with higher accuracy. In addition, by setting the change in the current value during one minute to 5×10 -12 [A] The following determination conditions can determine the detection value of the test leakage current with higher accuracy.

[0108] [Table 2]

[0109]

[0110] The determination condition of the test leakage current can be applied to the determination condition of the leakage current detection value in the diagnostic operation. Therefore, in the diagnostic operation in the first embodiment, the change in the current value flowing through the closed circuit during one minute is set to 1×10 -11 [A] The following is set as the determination condition of the leakage current detection value. In addition, the change amount of the current value during one minute can be set to 7×10 -12 [A] The following is set as the conditions for determining the detection value of the leakage current. Alternatively, the change in the current value during one minute may be set to 5×10 -12 [A] The following are set as the conditions for determining the detection value of the leakage current.

[0111] Next, use Fig.15 An example of a diagnostic operation will be described.

[0112] Fig.15 This is a flowchart for explaining an outline of the diagnostic work performed by the diagnostic system in the first embodiment.

[0113] Fig.15 The diagnostic work shown is started by an operator every time the elevator apparatus 2 is regularly inspected, for example.

[0114] In step S001 , the operator forms a closed circuit. That is, the two detection terminals 21 a and 21 b are connected to the coil 7 b and the grounding portion of the motor 7 , respectively. The diagnostic device 20 is also connected to the control panel 10 and the permanent measuring device 15 .

[0115] Thereafter, in step S002, the diagnostic device 20 applies a diagnostic voltage to the motor 7. The diagnostic device 20 acquires a detection value of the leakage current.

[0116] Thereafter, in step S003 , the diagnostic device 20 obtains information from the control panel 10 and the permanent measuring device 15 .

[0117] Then, in step S004, the diagnostic device 20 calculates the remaining life and outputs it. For example, the diagnostic device 20 displays the remaining life on a screen as an output.

[0118] Thereafter, in step S005, the worker collects each wiring.

[0119] After that, the diagnostic work ends.

[0120] Next, use Fig.16 The learning process in which the first learning device 30 generates a prediction model will be described.

[0121] Fig.16 This is a flowchart for explaining an outline of a learning process performed by the first learning device of the diagnostic system in the first embodiment.

[0122] Fig.16 The learning process is performed by a maintenance person of the information center 13, for example.

[0123] In step S101 , the first data acquisition unit 30 b acquires a first learning data set.

[0124] Thereafter, in step S102 , the first generation unit 30 c generates a prediction model using the first learning data set.

[0125] Thereafter, in step S103 , the first generation unit 30 c causes the first model storage unit 30 a to store the information of the prediction model generated in step S102 .

[0126] After that, the first learning device 30 ends the learning process.

[0127] Next, use Fig.17 The operation of the diagnosis executor 23 to calculate the remaining life will be described.

[0128] Fig.17 This is a flowchart for explaining an outline of the operation performed by the diagnosis executor of the diagnosis system in the first embodiment.

[0129] Fig.17The processing shown is similar to Fig.15 The processing corresponds to step S004 in the flowchart of . That is, Fig.17 The treatment shown is in Fig.15 The process starts after step S003 in the flowchart.

[0130] In step S201 , the preprocessing unit 23 c normalizes the detection value of the leakage current based on the acquired information.

[0131] Thereafter, in step S202, the acquisition unit 23d creates a data set.

[0132] Thereafter, in step S203 , the inference unit 23 f of the diagnosis unit 23 e infers the suspected elapsed time based on the data set and the prediction model.

[0133] Thereafter, in step 204, the calculation unit 23g of the diagnosis unit 23e calculates the remaining life based on the suspected elapsed time and the life elapsed time.

[0134] Thereafter, the diagnosis executor 23 ends the processing.

[0135] According to the first embodiment described above, the diagnostic system 1 includes a receiving unit 23b and a diagnostic unit 23e. The diagnostic unit 23e outputs information for estimating the remaining life of the motor 7 based on the insulation performance based on the prediction model. For example, the diagnostic unit 23e outputs the remaining life itself as information for estimating the remaining life. The prediction model is a model for inferring the remaining life based on a data set that includes the initial leakage current and the detection value of the leakage current in correspondence. By using the prediction model, the remaining life is output. Therefore, it is possible to estimate the life period when the motor 7 needs to be replaced. In addition, the accuracy of predicting the life of the motor 7 based on the insulation performance can be improved. In addition, the initial leakage current and the leakage current can be measured even in a low-voltage standard motor in which the generation of partial discharge is not expected in the design. That is, there is no need to disassemble such a standard motor, and the remaining life can be estimated based on the inspection items of the non-destructive inspection. Therefore, it will not affect the performance of the motor 7 during operation, and the degradation of the insulation performance can be accurately diagnosed. However, the diagnostic system 1 can also be applied to a motor in which the generation of partial discharge is expected as a design.

[0136] The diagnostic system 1 of the present disclosure can function as long as it includes the above-mentioned configurations of the receiving unit 23b and the diagnostic unit 23e. The configurations described below are additional configurations and are not essential to the diagnostic system 1 of the present disclosure.

[0137] The diagnostic system 1 may also include an inference unit 23f and a calculation unit 23g. The prediction model is a learned model representing the life function. The calculation unit 23g calculates the remaining life based on the suspected elapsed time output by the inference unit 23f. Therefore, the existing life function can be used and the accuracy of predicting the remaining life can be improved.

[0138] Furthermore, the diagnostic unit 23e may output a life function based on the prediction model as information for estimating the remaining life. In this case, for example, the operator may calculate the remaining life by applying input parameters such as the management upper limit value of the detection value of the leakage current to the output life function. In addition, the diagnostic unit 23e may output information such as the recommended replacement period calculated based on the remaining life as information for estimating the remaining life. In these cases, it is also possible to estimate the life period when the motor 7 needs to be replaced.

[0139] In the diagnostic system 1, the receiving unit 23b may also receive input of a measured value of water vapor pressure for calculating the surrounding environment of the motor 7. The diagnostic unit 23e may also output information for estimating the remaining life using a data set including the value of water vapor pressure and the detected value of leakage current in correspondence. Therefore, the remaining life prediction accuracy can be improved.

[0140] In addition, the diagnostic system 1 may further include a detector 22. The detector 22 can detect 1×10 -7 Therefore, the prediction accuracy of the information for estimating the remaining life can be improved.

[0141] Alternatively, a value whose change per minute is 1×10 -11 [A] or less. In the past, the current flowing through the closed circuit was considered to have reached equilibrium based on the time from when the voltage was applied. In this embodiment, the value at which the current is considered to have reached equilibrium can be determined more accurately than in the past. In addition, the current can be considered to have reached equilibrium in a shorter time than in the past, which can improve the efficiency of the diagnostic work.

[0142] In addition, the diagnostic system 1 may also include a preprocessing unit 23c. The preprocessing unit 23c standardizes the detection value of the leakage current. In the diagnostic unit 23e, information for estimating the remaining life is output based on the standard leakage current value. Therefore, for multiple motors of different models, the remaining life is predicted in a manner that minimizes the influence of differences caused by the structural characteristics of each motor. As a result, the prediction accuracy of the remaining life can be improved. In addition, the prediction model can be applied to motors of various structures.

[0143] In addition, the diagnostic system 1 may also include a first learning device 30 as a learning device. The first learning device 30 generates a prediction model using data obtained in a degradation test of the motor 7 that simulates an actual machine. Therefore, the amount of data when making a prediction model can be increased. The data obtained by performing degradation tests under various conditions can be reflected in the prediction model. By using the prediction model, information for estimating the life period when the motor 7 needs to be replaced can be output. As a result, the prediction accuracy based on the prediction model can be improved.

[0144] In addition, the first learning data set may also include the value of water vapor pressure as a degradation condition. The influence of water vapor pressure on the degradation of insulation performance is greater than the influence of other indicators. Therefore, the prediction accuracy based on the prediction model can be further improved.

[0145] In addition, the first learning data set may include the value of the test voltage applied to the model coil 42. Therefore, the prediction accuracy based on the prediction model can be further improved.

[0146] The first learning data set may also include at least one of items caused by material properties of the insulating layer, items suspected to be caused by structural properties of the motor 7, and items related to degradation conditions. Therefore, the diagnostic system 1 can be applied to motors 7 having different materials, structures, etc.

[0147] In addition, the value of the leakage current detection value may be determined by the receiving unit 23b of the diagnosis execution unit 23 instead of the detector 22 based on the amount of change in the leakage current value. That is, in this case, the detector 22 may send a signal indicating the detected current value to the diagnosis execution unit 23 at a predetermined period. The receiving unit 23b may also set the amount of change in the leakage current value per minute to 1×10 -11 The current value below [A] is adopted as the detection value of the leakage current.

[0148] In addition, the detection value of the leakage current detected by the detector 22 may be directly input to the diagnosis executor 23 via an input interface provided in the diagnosis device 20 .

[0149] Alternatively, the leakage current detection value may be corrected according to the environment in which the detection value is detected. The environment in which the detection value is detected includes the temperature and humidity around the detector 22 during the diagnostic operation.

[0150] The storage unit 23a may store the management upper limit of the leakage current detection value corresponding to the life function instead of the life elapsed time. In this case, the calculation unit 23g may calculate the life elapsed time using the life function and the management upper limit of the leakage current detection value when calculating the remaining life.

[0151] Alternatively, as the standard leakage current value, a detection value of the leakage current per unit thickness of the insulating layer may be used.

[0152] In addition, the prediction model is not limited to the example described above, but can also be a learned model as follows: the model is a learned model obtained through machine learning, and the model represents the relationship between "one or more parameters containing information indicating the leakage current during diagnosis" and "the estimated value of the accumulated elapsed time after taking into account the degradation condition of the insulation performance of motor 7", or the relationship between "the one or more parameters" and "the remaining life when viewed from the insulation performance of the motor as the detection source of the leakage current".

[0153] In addition, the information set as being included in the data set can be used not only as an input data set for the prediction model, but also as information for selecting a prediction model to be used from a plurality of prediction models, and can also be used as information for correcting an input data set including a detection value of a leakage current. In this case, the expression "a data set input to the prediction model" in the description of the present disclosure can also be appropriately replaced with "a data set input to the diagnosis unit 23e". Here, the information set as being included in the data set refers to the detection value of the initial leakage current, the standard initial leakage current value, the detection value of the leakage current, the standard leakage current value, the value of the water vapor pressure, the average value of the applied voltage to the working motor 7, the number of times the applied voltage is applied, the average temperature of the atmosphere, the number of thermal shocks, the average temperature of the thermal shock, the transformation start temperature Tg of the resin forming the insulation layer of the coil, the degree of hydrolysis of the resin, the thermal decomposition start temperature Td of the resin, the average molecular weight of each structural molecule of the resin, the dielectric constant of the resin, the core length of the motor coil, the occupancy rate, the wire diameter and other information.

[0154] Next, a modification of the diagnostic system 1 will be described.

[0155] Fig.18 This is a block diagram of a modified example of the diagnostic system in the first embodiment.

[0156] In the modified example, the inference unit 23f infers the remaining life from the data set and outputs it as information for estimating the remaining life. Fig.18 As shown, in a modified example, the calculation unit 23g may not be provided.

[0157] Specifically, in the modified example, the prediction model is a learned model that outputs the corresponding remaining life according to the data set. In addition, the data set may also include the same information as in the first embodiment. Therefore, the inference unit 23f inputs each input parameter included in the data set into the prediction model to infer the remaining life shown by the prediction model. As information for estimating the remaining life, the diagnosis unit 23e outputs the remaining life itself.

[0158] In the modified example, the first learning data set created in the first learning device 30 further includes the remaining life in correspondence with the detection value of the leakage current. In the first learning database, the remaining life calculated according to each numerical condition is stored in correspondence with the numerical condition. The remaining life can also be calculated based on the same method as that performed by the calculation unit 23g in the first embodiment.

[0159] That is, in the modified example, a part of the first learning data set is created as shown in the following Table 3. In Table 3, the unit of the remaining life is [year].

[0160] [Table 3]

[0161]

[0162] As an example, the first learning device 30 generates a prediction model using the first learning data set through so-called supervised learning. Fig.16 The learning process shown is the same.

[0163] Next, use Fig.19 An example of a learning algorithm used by the first learning device 30 when generating a prediction model will be described.

[0164] Fig.19 This is a diagram showing an example of a learning algorithm using a neural network model used in a modified example of the diagnostic system in the first embodiment.

[0165] like Fig.19 As shown, for example, a neural network model can also be used in the learning algorithm. That is, the first generation unit 30c learns the remaining life by so-called supervised learning according to the neural network model. Here, supervised learning refers to a method of learning the characteristics of the first learning data set as a set of input and result data by providing the first learning device 30, and inferring the result based on the input.

[0166] A neural network includes an input layer consisting of multiple neurons, an intermediate layer consisting of multiple neurons, and an output layer consisting of multiple neurons. The intermediate layer is also called a hidden layer, which can be one or more than two layers.

[0167] For example, if it is Fig.19 In the three-layer neural network shown in FIG. 1 , when multiple inputs are input to the input layer (X1-X3), the values ​​are multiplied by weights W1 (w11-w16) and input to the middle layer (Y1-Y2). The output from the middle layer (Y1-Y2) as a result of the input is multiplied by weights W2 (w21-w26) and output from the output layer (Z1-Z3). The final output result varies according to the values ​​of weights W1 and W2.

[0168] In this modification, the neural network learns the prediction model by so-called supervised learning according to the first learning data set produced based on the combination of each value acquired by the first data acquisition unit 30b and the remaining life. That is, the neural network learns by adjusting the weights W1 and W2 so that the result output by inputting each value to the input layer is close to the remaining life.

[0169] Next, use Fig. 20 An example of the operation of the diagnosis executor 23 in the modification example to calculate the remaining life will be described.

[0170] Fig. 20 This is a flowchart for explaining an outline of the operation performed by the diagnosis executor of the modified example of the diagnosis system in the first embodiment.

[0171] Fig. 20 The processing shown is similar to Fig.15 The processing corresponds to step S004 in the flowchart of . That is, Fig. 20 The treatment shown is in Fig.15 The process starts after step S003 in the flowchart.

[0172] In step S301 , the preprocessing unit 23 c normalizes the detection value of the leakage current based on the acquired information.

[0173] Thereafter, in step S302, the acquisition unit 23d creates a data set.

[0174] Then, in step S303, the inference unit 23f of the diagnosis unit 23e infers the remaining life based on the data set.

[0175] Thereafter, the diagnosis executor 23 ends the processing.

[0176] According to the modification of the first embodiment described above, the diagnostic system 1 has an inference unit 23f included in the diagnostic unit 23e. The prediction model outputs the remaining life itself as information for estimating the remaining life. When outputting the remaining life, there is no need to calculate a life curve, etc., compared with the first embodiment. Therefore, the diagnostic system 1 can predict the remaining life with high accuracy at a low calculation cost.

[0177] In addition, in a modified example, the information used to output the remaining life may also be other information other than the remaining life. In this case, the other information is included in the first learning data set as data to be output instead of the remaining life. Supervised learning may also be performed using the other information as data to be output instead of the remaining life.

[0178] Next, use Fig.21 An example of hardware constituting the diagnosis executor 23 will be described.

[0179] Fig.21 This is a hardware configuration diagram of the diagnosis executor of the diagnosis system in the first embodiment.

[0180] Each function of the diagnosis executor 23 can be realized by a processing circuit. For example, the processing circuit includes at least one processor 100a and at least one memory 100b. For example, the processing circuit includes at least one dedicated hardware 200.

[0181] In the case where the processing circuit has at least one processor 100a and at least one memory 100b, the functions of the diagnostic executor 23 are implemented by software, firmware, or a combination of software and firmware. At least one of the software and firmware is described as a program. At least one of the software and firmware is stored in at least one memory 100b. At least one processor 100a implements the functions of the diagnostic executor 23 by reading out and executing the program stored in at least one memory 100b. At least one processor 100a is also called a central processing unit, a processing unit, an arithmetic unit, a microprocessor, a microcomputer, and a DSP. For example, at least one memory 100b is a non-volatile or volatile semiconductor memory such as a RAM, a ROM, a flash memory, an EPROM, an EEPROM, a magnetic disk, a floppy disk, an optical disk, a high-density disk, a mini disk, a DVD, etc.

[0182] In the case where the processing circuit has at least one dedicated hardware 200, the processing circuit is implemented, for example, by a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination thereof. For example, each function of the diagnosis executor 23 is implemented by the processing circuit separately. For example, each function of the diagnosis executor 23 is uniformly implemented by the processing circuit.

[0183] The functions of the diagnosis executor 23 may be partially implemented by dedicated hardware 200, and the other part may be implemented by software or firmware. For example, the functions performed by the preprocessing unit 23c may be implemented by a processing circuit as dedicated hardware 200, and the functions other than the functions performed by the preprocessing unit 23c may be implemented by at least one processor 100a reading and executing a program stored in at least one memory 100b.

[0184] In this way, the processing circuit implements the functions of the diagnosis executor 23 through hardware 200, software, firmware, or a combination thereof. In addition, the operation of the processing circuit can be implemented not only by the processing circuit installed in one place, but also by the processing circuits installed in multiple places being integrated through a network, etc. For example, it can also be configured to implement the same operation as the processing circuit on the so-called cloud server implemented in this way.

[0185] For example, the functions of the storage unit 23a, the preprocessing unit 23c, and the acquisition unit 23d may be realized in the information center device 12 connected to the diagnosis executor 23 via a network or the like. That is, the information center device 12 may include the storage unit 23a, the preprocessing unit 23c, and the acquisition unit 23d.

[0186] Although not shown in the figure, each function of the control panel 10, each function of the information center device 12, each function of the detector 22 and each function of the first learning device 30 are also implemented by processing circuits equivalent to the processing circuits that implement the functions of the diagnosis executor 23.

[0187] In the first embodiment including the modified example, the diagnostic system 1 may not include the first learning device 30 . Fig. 22 FIG. 2 is a block diagram showing another example of the diagnostic system in Embodiment 1. Fig. 22 As shown, in the diagnosis system 1 , as long as the remaining life is inferred using the prediction model, the diagnosis system 1 does not need to include the first learning device 30 .

[0188] Implementation method 2.

[0189] Fig.23 This is a block diagram of a diagnostic system in Embodiment 2. In addition, the same reference numerals are attached to the same or corresponding parts as those in Embodiment 1, and the description of such parts is omitted.

[0190] like Fig.23 As shown, in the second embodiment, the diagnosis executor 23 is provided in the information center 13. For example, the diagnosis executor 23 is a part of the function of the information center device 12.

[0191] The diagnostic device 20 is provided with a detector 22. The operator detects the leakage current during the diagnostic work using the detector 22. For example, the remote monitoring device 11 transmits the detected leakage current and its time-lapse information, the operation information stored in the control panel 10, and the measurement result information stored in the permanent measuring device 15 to the information center device 12 based on the instruction from the diagnostic device 20.

[0192] Thereafter, the diagnosis executor 23 of the information center device 12 calculates the remaining life in the same manner as in Embodiment 1. For example, the calculation result of the remaining life may be notified to a maintenance worker of the information center 13 .

[0193] Furthermore, the information of the measurement result of the detector 22 may be directly transmitted to the information center device 12 by the operator after returning to the information center 13 , not via the remote monitoring device 11 .

[0194] As shown in the second embodiment described above, the diagnosis executor 23 may not be provided in the diagnosis device 20. In this case, the diagnosis system 1 also includes the receiving unit 23b and the diagnosis unit 23e. Therefore, in the diagnosis system 1, the remaining life prediction accuracy can be improved regardless of the position where the diagnosis executor 23 is provided.

[0195] Although not shown in the drawings, the prediction model and the configuration of the diagnosis unit 23 e in the first embodiment may be applied to the diagnosis system 1 in the second embodiment.

[0196] Implementation method 3.

[0197] Fig.24 This is a block diagram of a diagnostic system in Embodiment 3. In addition, the same reference numerals are attached to the same or corresponding parts as those in Embodiment 1 or Embodiment 2, and the description of such parts is omitted.

[0198] like Fig.24 As shown, in the third embodiment, the diagnostic system 1 further includes a second learning device 50. For example, the second learning device 50 is provided inside the diagnostic device 20.

[0199] The second learning device 50 updates the prediction model and generates an updated prediction model using information on the leakage current detection value, information measured by the permanent measuring device 15, and operation information of the elevator device 2. The second learning device 50 includes a second model storage unit 50a, a second data acquisition unit 50b, and a second generation unit 50c.

[0200] The second model storage unit 50a stores information of the prediction model. The prediction model stored in the second model storage unit 50a may be a prediction model before updating or a prediction model after updating.

[0201] The second data acquisition unit 50b acquires information on the detected value of the leakage current, information measured by the permanent measuring device 15, and operation information of the elevator device 2 from the diagnosis executor 23. At this time, the second data acquisition unit 50b may also acquire information of the data set. The second data acquisition unit 50b creates a second learning data set based on the acquired information.

[0202] In the second learning data set, at least one of the leakage current detection value used by the diagnosis executor 23 in the most recent diagnosis, the water vapor pressure value based on the measurement result of the permanent measuring device 15, and the operation information is included in correspondence with the leakage current detection value at the time of diagnosis. Specifically, in the second learning data set, at least one of the number of starts of the motor 7, the start time of the motor 7, the accumulated operation time of the motor 7, and the accumulated travel distance of the car 8 may be included in correspondence with the leakage current detection value at the time of diagnosis. In the second learning data set, information of the data set used in the most recent diagnosis may be further included in correspondence with the leakage current detection value at the time of diagnosis.

[0203] The second generation unit 50c learns the prediction model based on the second learning data set acquired by the second data acquisition unit 50b, thereby updating the prediction model. That is, based on the second learning data set, one or more life functions are regenerated as the prediction model. For example, the function parameters of the life function are adjusted. In addition, the learning algorithm executed by the second generation unit 50c can be a well-known algorithm. The second generation unit 50c stores the information of the updated prediction model as the updated prediction model in the second model storage unit 50a.

[0204] When the prediction model is updated, the diagnosis executor 23 obtains information of the updated prediction model from the second learning device 50, and rewrites the prediction model stored in the storage unit 23a to the updated prediction model. In the next diagnostic operation, the diagnosis unit 23e of the diagnosis executor 23 uses the updated prediction model generated by the second generation unit 50c to output information for estimating the remaining life. For example, the diagnosis unit 23e outputs the remaining life itself using the updated prediction model.

[0205] Next, use Figure 25 to Figure 27 The reason why the operation information is included in the second learning data set is explained.

[0206] Figure 25 to Figure 27 This is a diagram showing the relationship between the operating performance and the leakage current measured in the motor of the hoisting machine in a general elevator device.

[0207] Fig.25 This is a graph showing the relationship between the measurement time and the leakage current per unit area measured in the motors of multiple hoisting machines in a general elevator device. The vertical axis is the leakage current per unit contact area between the motor core and the coil, i.e., the leakage current per unit area. The unit of the leakage current per unit area is [nA] / [m 2 The detection value of the leakage current is measured by the same method as in Embodiment 1. The horizontal axis is the travel time of the car traveling by the hoisting machine. The unit of the travel time is [hour].

[0208] The point group in the graph is the initial leakage current detection value and the leakage current detection value of the motor at different running times. Fig.25 It is understood that there is a correlation that the value of the leakage current per unit area increases as the running time increases.

[0209] Fig.26 It means Fig.25 The graph of the relationship between the number of years of use and the leakage current per unit area in the same motor is shown in FIG. The vertical axis is the leakage current per unit area. The horizontal axis is the number of years that have passed since the motor was installed. Fig.26 , no clear correlation was seen between the number of years and the value of the leakage current per unit area.

[0210] Fig. 27 It means Fig.25 A graph showing the relationship between the travel distance and the leakage current per unit area measured in the same motor of the plurality of motors shown in FIG. The vertical axis is the leakage current per unit area. The horizontal axis is the cumulative travel distance of the car driven by the motor. The unit of the travel distance is [kilometer]. Fig. 27 It is understood that there is a correlation that the value of the leakage current per unit area increases as the travel distance increases.

[0211] In addition, from Figure 25 to Figure 27 It can be seen that the travel distance has the strongest correlation with the value of the leakage current per unit area. Therefore, the second learning data set includes operation information including the accumulated travel distance of the car 8.

[0212] In addition, the correlation between the leakage current per unit area and the number of years is particularly weak. Based on this, it is believed that the motor speed has a great influence on the increase in the leakage current per unit area, that is, the degradation of the insulation performance. It is believed that the greater the average value of the motor speed, the faster the degradation of the insulation performance progresses. The output power [W] affects the motor speed. Here, in general, the operating voltage of the elevator is constant, so the output power is proportional to the output current. In implementation mode 1, etc., sometimes at least one of the average value of the voltage applied to the model coil unit and the number of applications is included in the first learning data set as a degradation condition. The degradation condition is set to simulate the influence of the degradation of the insulation performance caused by the output current by applying a voltage to the model coil 42 to circulate the current.

[0213] Next, use Fig.28 The learning process in which the second learning device 50 generates an updated prediction model will be described.

[0214] Fig.28 This is a flowchart for explaining an outline of a learning process performed by the second learning device of the diagnostic system in the third embodiment.

[0215] Fig.28 The learning process is executed by, for example, an operator who has performed a diagnostic task. After performing the diagnostic task, the operator inputs a command for starting the learning process.

[0216] In step S401 , the second data acquisition unit 50 b acquires a second learning data set.

[0217] Thereafter, in step S402 , the second generation unit 50 c generates a prediction model using the second learning data set.

[0218] Thereafter, in step S403 , the second generation unit 50 c causes the second model storage unit 50 a to store the information of the prediction model generated in step S402 .

[0219] After that, the second learning device 50 ends the learning process.

[0220] As shown in the third embodiment described above, the diagnostic system 1 may further include a second learning device 50. The second learning device 50 updates the prediction model using a second learning data set that corresponds the operation information of the elevator device 2 to the leakage current. The diagnostic unit 23e outputs information for estimating the remaining life based on the updated prediction model. Therefore, the remaining life can be predicted based on the actual degree of progress of the deterioration of the motor 7. As a result, the prediction accuracy of the remaining life can be improved.

[0221] Furthermore, the second learning device 50 may generate an updated prediction model different from the updated prediction model corresponding to the motor 7 for a motor provided in an elevator device different from the motor 7 .

[0222] In addition, the prediction model in the diagnostic system 1 of the third embodiment may also output information for estimating the remaining life in the same manner as the prediction model in the modified example of the first embodiment. In this case, as an example, the second data acquisition unit 50b acquires the information of the remaining life corresponding to the leakage current from the diagnosis executor 23, and creates a second learning data set including the remaining life and the detection value of the leakage current in correspondence. The second generation unit 50c generates an updated prediction model for outputting the remaining life.

[0223] Implementation method 4.

[0224] Fig.29 This is a block diagram of a diagnostic system in Embodiment 4. In addition, the same reference numerals are attached to the same or corresponding parts as those in Embodiment 1, Embodiment 2, and Embodiment 3. The description of such parts will be omitted.

[0225] In the fourth embodiment, the first learning device 30 has the function of the second learning device 50 in the third embodiment. For example, after the operator diagnoses the motor 7 and returns to the information center 13, the operator causes the first learning device 30 to perform the learning process. At this time, the operator may also connect the diagnostic device 20 to the information center device 12.

[0226] That is, the first model storage unit 30a stores the prediction model before or after the update. The first data acquisition unit 30b creates the same second learning data set as in Embodiment 3. At this time, the first data acquisition unit 30b obtains the information used by the diagnosis executor 23 in the diagnosis by directly connecting to the diagnosis device 20 or via the remote monitoring device 11.

[0227] The first generation unit 30c generates a prediction model by the same method as the second generation unit 50c in Embodiment 3. Then, the first generation unit 30c causes the first model storage unit 30a to store the updated prediction model.

[0228] As shown in the fourth embodiment described above, the first learning device 30 performs the same learning process as the second learning device 50 in the third embodiment. That is, the operator does not need to additionally carry the second learning device 50. As a result, the workability of the diagnostic work can be improved.

[0229] Industrial Applicability

[0230] As described above, the diagnostic system according to the present disclosure can be utilized for diagnostic work of an elevator device.

[0231] (Explanation of Reference Numerals)

[0232] 1: diagnostic system; 2: elevator device; 3: building; 4: lifting path; 5: machine room; 6: hoist; 7: motor; 7a: core; 7b: coil; 7c: measuring terminal; 7d: measuring terminal; 7e: measuring terminal; 7f: base; 8: car; 9: main rope; 10: control panel; 11: remote monitoring device; 12: information center device; 13: information center; 14: network; 15: permanent measuring device; 20: diagnostic device; 21a, 21b: detection terminal; 22: detector; 23: diagnostic actuator; 23a: storage unit; 23b: receiving unit; 23c: preprocessing unit; 23d: acquisition unit; 23e: diagnosis unit; 23f: reasoning unit; 23g: operation unit; 30: first learning device; 30a: first model storage unit; 30b: first data acquisition unit; 30c: first generation unit; 40: model coil unit; 41: model core; 42: model coil; 43: insulation paper; 44: terminal; 50: second learning device; 50a: second model storage unit; 50b: second data acquisition unit; 50c: second generation unit; 100a: processor; 100b: memory; 200: hardware; Lc: core length

Claims

1. A diagnostic system for diagnosing deterioration of insulation performance of a motor, the diagnostic system comprising: a receiving unit that receives an input of a detection value of a leakage current flowing through a closed circuit including a coil of the motor and a ground portion of the motor; and The diagnostic unit uses a data set that includes a detection value of the leakage current detected at a time point that becomes a base point, i.e., an initial leakage current, and a detection value of the leakage current received by the receiving unit in correspondence with each other, and a prediction model for inferring the remaining life of the motor based on insulation performance from the data set, and outputs information for estimating the remaining life.

2. The diagnostic system according to claim 1, wherein: The prediction model is a learned model of a life function representing a relationship between a detected value of the leakage current and an estimated value of a cumulative elapsed time of the motor, The diagnostic unit has: an inference unit that infers a suspected elapsed time corresponding to the detection value of the leakage current included in the data set by inputting the data set into the prediction model; as well as A calculation unit calculates the life elapsed time of the motor corresponding to the management upper limit value of the leakage current in the prediction model, calculates a period from the suspected elapsed time to the life elapsed time as the remaining life, and outputs it as information for estimating the remaining life.

3. The diagnostic system according to claim 1, wherein: The prediction model is a learned model for inferring the remaining life according to the data set, The diagnosis unit includes an inference unit that infers the remaining life by inputting the data set into the prediction model, and outputs the information for estimating the remaining life.

4. The diagnostic system according to any one of claims 1 to 3, wherein: The receiving unit receives an input of a measured value for calculating a water vapor pressure of an environment surrounding the motor while the motor is operating. The diagnostic unit outputs information for estimating the remaining lifetime using the data set including the value of the water vapor pressure obtained based on the measurement value accepted by the accepting unit and the detection value of the leakage current in association with each other.

5. The diagnostic system according to any one of claims 1 to 4, wherein: A detector is further provided, the detector being electrically connectable to the coil and the grounding portion of the motor and being capable of detecting 1×10 -7 [A] or below, The receiving unit receives a detection value of the leakage current from the detector.

6. The diagnostic system according to claim 5, wherein: As the detection value of the leakage current, a change in the leakage current detected by the detector per minute of 1×10 -11 Current value when [A] is below.

7. The diagnostic system according to any one of claims 1 to 6, wherein: further comprising a preprocessing unit that uses information on the structural characteristics of the motor and the detection value of the leakage current received by the receiving unit to calculate a standard leakage current value obtained by standardizing the detection value of the leakage current according to the structural characteristics, The diagnosis section outputs information for estimating the remaining lifetime using the data set including the standard leakage current value instead of the detection value of the leakage current and the prediction model.

8. The diagnostic system according to any one of claims 1 to 7, wherein: The method further comprises a first learning device, wherein the first learning device generates the prediction model. The first learning device comprises: a first data acquisition unit that acquires a first learning data set including a detected value of a current obtained by applying a voltage to a test circuit simulating the closed circuit, wherein the test circuit is a circuit including a model coil simulating the coil; and a first generating unit, using the first learning data set, generating the prediction model which is a learned model for inferring the remaining life from the data set; The first learning data set correspondingly includes: a detection value of an initial test leakage current flowing through the test circuit; and a detection value of a test leakage current flowing through the test circuit after the initial test leakage current is detected and the model coil is subjected to a degradation test under specified degradation conditions.

9. The diagnostic system according to claim 8, wherein: The first learning data set includes, in correspondence with each other, a value of water vapor pressure of the surrounding environment of the model coil set as the degradation condition in the degradation test and the test leakage current. The first generation unit generates the prediction model for inferring the remaining lifetime based on the data set including the detected value of the water vapor pressure of the surrounding environment of the motor and the detected value of the leakage current in association with each other.

10. The diagnostic system according to claim 8 or 9, wherein: The first learning data set includes the value of the test voltage applied to the model coil as the degradation condition in the degradation test and the test leakage current in correspondence with each other.

11. The diagnostic system according to any one of claims 8 to 10, wherein: In the first learning data set, corresponding to the test leakage current, it includes: the transformation start temperature of the resin constituting the insulation layer of the model coil, the hydrolysis degree of the resin, the thermal decomposition start temperature of the resin, the length of the part corresponding to the core length in the model coil, the suspected duty cycle of the model coil, the wire diameter of the model coil, the temperature of the surrounding environment of the model coil that changes as the degradation condition, and at least one of the number of thermal shocks applied to the model coil as the degradation condition.

12. The diagnostic system according to any one of claims 8 to 11, wherein: The motor is installed in a hoisting machine that moves the elevator car up and down. The first data acquisition unit acquires a second learning data set including at least one of the number of startups of the motor, the startup time of the motor, the cumulative operating time of the motor, and the cumulative travel distance of the car in correspondence with the detection value of the leakage current, The first generating unit uses the second learning data set to generate an updated prediction model for inferring the remaining life based on the data set that further includes at least one of the number of times the motor is started, the start time of the motor, the cumulative operating time of the motor, and the cumulative driving distance of the car corresponding to the detection value of the leakage current.

13. The diagnostic system according to claim 12, wherein: The receiving unit receives information including at least one of the number of times the motor is started, the start time of the motor, the accumulated operation time of the motor, and the accumulated travel distance of the car, The diagnostic unit uses the data set that includes at least one of the number of times the motor is started, the start time of the motor, the cumulative operating time of the motor, and the cumulative travel distance of the car in correspondence with the detection value of the leakage current and the updated prediction model generated by the first generation unit to output information for estimating the remaining life.

14. The diagnostic system according to any one of claims 1 to 11, wherein: The method further comprises a second learning device for generating an updated prediction model obtained by updating the prediction model. The motor is installed in a hoisting machine that moves the elevator car up and down. The second learning device: a second data acquisition unit that acquires a second learning data set that includes at least one of the number of starts of the motor, the start time of the motor, the accumulated operation time of the motor, and the accumulated travel distance of the car in correspondence with the detection value of the leakage current; as well as The second generating unit uses the second learning data set to generate an updated prediction model for inferring the remaining life based on the data set that further includes at least one of the number of times the motor is started, the starting time of the motor, the cumulative operating time of the motor, and the cumulative driving distance of the car corresponding to the detection value of the leakage current.

15. The diagnostic system according to claim 14, wherein: The receiving unit receives information including at least one of the number of times the motor is started, the start time of the motor, the accumulated operation time of the motor, and the accumulated travel distance of the car, The diagnostic unit uses the data set that includes at least one of the number of times the motor is started, the start time of the motor, the cumulative operating time of the motor, and the cumulative travel distance of the car in correspondence with the detection value of the leakage current and the updated prediction model generated by the second generation unit to output information for estimating the remaining life.

16. A learning device for generating a prediction model for inferring the remaining life of a motor based on insulation performance based on electrical characteristics of a closed circuit including a coil of a motor and a grounded portion of the motor, the learning device comprising: A data acquisition unit acquires a first learning data set, wherein the first learning data set correspondingly includes: a detection value of an initial test leakage current flowing through a test circuit that simulates the closed circuit using a model coil that simulates the coil of the motor; and a detection value of a test leakage current flowing through the test circuit after the initial test leakage current is detected and after a degradation test is performed on the model coil under a prescribed degradation condition; and A generating unit uses the first learning data set to generate a learned model, i.e., the prediction model, for inferring the remaining life of the motor based on a data set that includes a detection value of an initial leakage current flowing through the closed circuit at a time point that becomes a base point and a detection value of a leakage current flowing through the closed circuit in correspondence with each other.

Images