JP2017060333A - Abnormality diagnosis method for semiconductor power conversion device, abnormality diagnosis device, abnormality diagnosis program and semiconductor power conversion device having abnormality diagnosis function - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable diagnosis and detection of abnormality/trouble of a semiconductor power conversion device without being dependent on an output to the outside such as a voltage or current of the semiconductor power conversion device.SOLUTION: A spectrum value for each frequency component is calculated with respect to audio data acquired by picking up sounds generated from a semiconductor power conversion device to be monitored (S1, S2), the spectrum value for each frequency component is used to determine the occurrence or non-occurrence of abnormality/trouble in the semiconductor power conversion device to be monitored (S3), and when occurrence of an abnormality/trouble is determined, an abnormality notification signal is output (S4).SELECTED DRAWING: Figure 1

Description

  The present invention relates to an abnormality diagnosis method for a semiconductor power conversion device, an abnormality diagnosis device, an abnormality diagnosis program, and a semiconductor power conversion device having an abnormality diagnosis function. More specifically, the present invention relates to a technique suitable for use in diagnosis and detection of abnormality / failure in a semiconductor power conversion device.

  As a semiconductor power conversion device that performs conversion from AC power to DC power or conversion from DC power to AC power using a semiconductor switching element, for example, based on a combination of signals applied to each gate, including a plurality of gates The first mode that operates with the characteristic that the on-time voltage is relatively high but the switching speed is fast and the second mode that operates with the characteristic that the switching speed is relatively slow but the on-state voltage is low are switched. One of the turn-on and the short period immediately after and the turn-off and the short period immediately before or both in the energization period by the combination of the power semiconductor switching element and the signal sent to the gate of the power semiconductor switching element During the period, the power semiconductor switching element is driven by the characteristics of the first mode, In the period of Ri those having a control unit for driving the characteristics according to the second mode (Patent Document 1).

  Conventionally, diagnosis and detection of abnormalities / failures of semiconductor elements and the like in the semiconductor power conversion device as mentioned above as an example has hitherto been a voltage or current abnormality (in other words, a change), that is, an internal malfunction has occurred. Thus, it has been carried out by inspecting whether there is an abnormality in the output to the outside, which has been manifested as a result of the inability to maintain normal operation. Conventionally, abnormalities and failures of semiconductor elements and the like have been diagnosed and detected by regular maintenance and inspection.

Japanese Patent Laid-Open No. 6-98561

  However, the conventional method of inspecting the presence or absence of abnormalities (changes) in voltage or current for diagnosing or detecting abnormalities or failures of semiconductor elements in semiconductor power conversion devices has caused abnormalities or failures in semiconductor elements. As a result, there are many cases where an abnormality cannot be found until after the semiconductor power conversion device is stopped, and therefore, there is a problem that it is impossible to prevent the failure stop of the semiconductor power conversion device. Therefore, it is difficult to say that the conventional technique is highly useful and reliable as a technique for detecting an abnormality / failure in a semiconductor power converter.

  Furthermore, in the conventional method, diagnosis and detection are performed by maintenance inspections that are periodically performed at predetermined intervals. Therefore, even if an abnormality occurs between inspections, a failure is detected and dealt with in advance. There is a problem that can not be. Therefore, it is difficult to say that the conventional method is highly useful and reliable as a method for detecting an abnormality / failure in the semiconductor power conversion device.

  Therefore, the present invention provides a semiconductor power conversion device that diagnoses and detects an abnormality / failure of the semiconductor power conversion device without relying on an external output such as a voltage or current of the semiconductor power conversion device, and does not depend on regular maintenance and inspection. Semiconductor power conversion device abnormality diagnosis method, abnormality diagnosis device, abnormality diagnosis program, and abnormality diagnosis function capable of preventing failure stop of semiconductor power conversion device by performing diagnosis and detection of device abnormality / failure It aims at providing a semiconductor power converter provided with.

  The inventors have conducted a test to collect sound while intentionally failing a semiconductor switching element of a semiconductor power converter, and before the semiconductor power converter is stopped due to a malfunction of the semiconductor element, It was found that a sound different from the operation sound (in other words, the sound during normal operation) is generated. Note that the time from the occurrence of abnormal noise to the failure stop of the semiconductor power conversion device depending on the type or degree of failure of the semiconductor element is not limited to a specific time, but in our test, for example, the semiconductor There were cases in which abnormal noise occurred several times between about 9 seconds and 6 seconds before the power converter stopped and stopped.

  The present invention is based on the above knowledge, and the method for diagnosing abnormality of a semiconductor power conversion device according to the present invention is a spectrum for each frequency component of sound data obtained by collecting sound generated from a monitored semiconductor power conversion device. The value is calculated, and the spectrum value for each frequency component is used to determine whether or not an abnormality or failure has occurred in the monitored semiconductor power conversion device.

  Further, according to the abnormality diagnosis method for a semiconductor power conversion device of the present invention, a spectrum value for each frequency component is calculated for each piece of sound data acquired by collecting sounds generated from the monitored semiconductor power conversion device in time series. Then, the spectrum value for each time-series frequency component is used to determine whether or not an abnormality / failure has occurred in the semiconductor power conversion device to be monitored.

  Further, the abnormality diagnosis method for a semiconductor power conversion device according to the present invention provides a frequency component for each of sound data acquired by collecting sounds generated from a plurality of semiconductor power conversion devices including at least a semiconductor power conversion device to be monitored. Each spectrum value is calculated, and the spectrum value for each frequency component for each semiconductor power converter is used to determine whether or not an abnormality or failure has occurred in the monitored semiconductor power converter.

  Further, the abnormality diagnosis device for a semiconductor power conversion device according to the present invention calculates a spectrum value for each frequency component of the sound data by means for sampling the sound generated from the semiconductor power conversion device to be monitored and outputting the sound data. Means and means for determining the presence / absence of an abnormality / failure in the monitored semiconductor power conversion device using the spectrum value for each frequency component.

  Further, the abnormality diagnosis program for a semiconductor power conversion device according to the present invention includes a process for calculating a spectrum value for each frequency component for sound data obtained by collecting sound generated from a monitored semiconductor power conversion device, and the frequency The computer is caused to perform processing for determining whether or not an abnormality or failure has occurred in the semiconductor power conversion device to be monitored using the spectrum value for each component.

  Therefore, according to the abnormality diagnosis method, abnormality diagnosis device, and abnormality diagnosis program for these semiconductor power conversion devices, it is possible to determine whether or not an abnormality or failure has occurred in the semiconductor power conversion device based on the sound generated from the semiconductor power conversion device. As a result, abnormalities in the semiconductor power conversion device are captured before changes in the output to the outside, such as the voltage and current of the semiconductor power conversion device, and more specifically, signs of abnormality in the semiconductor power conversion device are captured. Is done.

  According to these semiconductor power conversion device abnormality diagnosis method, abnormality diagnosis device, and abnormality diagnosis program, further, semiconductor power conversion without intervention in the semiconductor power conversion device as in the case of directly measuring output such as voltage and current Since the determination of the occurrence of abnormality / failure in the semiconductor power conversion device is performed by collecting the sound generated from the device, the semiconductor power conversion device is constantly monitored while performing normal operation.

  Further, the abnormality diagnosis method, abnormality diagnosis apparatus, and abnormality diagnosis program for a semiconductor power conversion device according to the present invention provide an output voltage and an output current when sound collection is started and stopped when the semiconductor power conversion device to be monitored is started. Or at least one of changing the switching frequency. In this case, an abnormal sound that does not become apparent during normal continuous / continuous and steady operation is manifested and collected by a special action.

  Further, the abnormality diagnosis method, abnormality diagnosis apparatus, and abnormality diagnosis program for a semiconductor power conversion device according to the present invention are such that the spectrum value of each frequency component is a spectrum value of a frequency that is an integral multiple of the switching frequency of the semiconductor power conversion device to be monitored Alternatively, it may be a spectrum value of a frequency other than an integral multiple of the switching frequency. In this case, an abnormal sound that is manifested as sound in a frequency band that is an integral multiple of the switching frequency or a frequency band other than an integral multiple of the switching frequency is collected.

  Moreover, the semiconductor power converter device provided with the abnormality diagnosis function of the present invention is provided with the above-described abnormality diagnosis device for the semiconductor power converter device. Therefore, according to this semiconductor power conversion device, a semiconductor power conversion device capable of exhibiting the action exhibited by the above-described abnormality diagnosis device for a semiconductor power conversion device is realized.

  According to the semiconductor power conversion device abnormality diagnosis method, abnormality diagnosis device, and semiconductor power conversion device provided with the abnormality diagnosis program and abnormality diagnosis function of the present invention, there is a change in the output to the outside such as the voltage and current of the semiconductor power conversion device. Capture abnormalities in the semiconductor power conversion device before it appears, and more specifically, because it can capture signs of abnormalities in the semiconductor power conversion device, it is possible to prevent outages due to failure of the semiconductor power conversion device. As a result, it is possible to improve the usefulness and reliability as a method for detecting an abnormality / failure in the semiconductor power conversion device.

  According to the semiconductor power conversion device abnormality diagnosis method, abnormality diagnosis device, and semiconductor power conversion device provided with the abnormality diagnosis program and abnormality diagnosis function of the present invention, the semiconductor power conversion device further performs normal monitoring while performing normal operation. Therefore, it is possible to always detect abnormalities / failures in semiconductor power converters without relying on regular maintenance and inspection, and as a method for detecting abnormalities / failures in semiconductor power converters. Usability and reliability can be improved.

  The semiconductor power conversion device abnormality diagnosis method, abnormality diagnosis device, and semiconductor power conversion device having an abnormality diagnosis program and an abnormality diagnosis function according to the present invention are activated, stopped, changed in output voltage and output current, or switched. When sound is collected when changing the frequency, abnormal sounds that do not appear during normal continuous / continuous and steady operation are made to appear by a special action. Since it can be collected, it is possible to more reliably capture abnormalities in the semiconductor power conversion device, more specifically, capture signs of abnormalities, and more reliably prevent failure stop of the semiconductor power conversion device. As a result, it becomes possible to further improve the usefulness and reliability as a method for detecting an abnormality / failure of the semiconductor power conversion device.

  In addition, the semiconductor power conversion device abnormality diagnosis method, abnormality diagnosis device, and semiconductor power conversion device having an abnormality diagnosis program and abnormality diagnosis function of the present invention may have a frequency value that is an integral multiple of the switching frequency of the device or a value other than an integral multiple. When the spectrum value of the frequency is used, it is possible to collect an abnormal sound that is manifested as a sound in a frequency band that is an integral multiple of the switching frequency or a frequency band other than the integral multiple. In addition, it is possible to more reliably capture abnormalities, more specifically, signs of abnormalities, and to more reliably prevent failure stop of the semiconductor power conversion device. As a result, it is possible to further improve the usefulness and reliability as a method for detecting abnormalities and failures.

It is a flowchart which shows an example of embodiment of the abnormality diagnosis method of the semiconductor power converter device of this invention. It is a functional block diagram which shows an example of embodiment of the abnormality diagnosis apparatus of the semiconductor power converter device of this invention. It is a functional block diagram of the abnormality diagnosis device of the semiconductor power conversion device realized by the program when the abnormality diagnosis method of the semiconductor power conversion device of the embodiment is implemented using the abnormality diagnosis program of the semiconductor power conversion device. It is a functional block diagram which shows an example of embodiment of the semiconductor power converter device provided with the abnormality diagnosis function of the semiconductor power converter device of this invention provided with the abnormality diagnosis function.

  Hereinafter, the configuration of the present invention will be described in detail based on an example of an embodiment shown in the drawings.

  In the following description, in order to clarify that it is a unit, a symbol or character as a unit may be enclosed in [].

  1 to 3 show an example of an embodiment of an abnormality diagnosis method, an abnormality diagnosis device, and an abnormality diagnosis program for a semiconductor power conversion device according to the present invention.

  The abnormality diagnosis method for a semiconductor power conversion device according to the present embodiment is a frequency component obtained by subjecting sound data acquired from a monitored semiconductor power conversion device to Fourier transform processing or wavelet transform processing. The spectrum value for each frequency component is calculated (S1, S2), and the spectrum value for each frequency component is used to determine whether or not an abnormality / failure has occurred in the monitored semiconductor power conversion device (S3) and the abnormality / failure When it is determined that the error has occurred, an abnormality notification signal is output (S4) (see FIG. 1).

  An abnormality diagnosis device 10 for a semiconductor power conversion device according to the present embodiment includes a sound detection unit 1 as a means for collecting sound generated from a monitored semiconductor power conversion device and outputting sound data, and Fourier for the sound data. Conversion unit 2 as means for calculating a spectrum value for each frequency component by performing a conversion process or a wavelet transform process, and whether or not an abnormality / failure has occurred in the semiconductor power conversion device to be monitored using the spectrum value for each frequency component And determining unit 3 as means for outputting an abnormality notification signal when it is determined that abnormality / failure has occurred, and alarm output unit 4 as means for issuing an alarm according to the abnormality notification signal (See FIG. 2).

  Then, as an implementation of the abnormality diagnosis method for the semiconductor power converter, first, sound generated from the semiconductor power converter is collected and sound data is acquired (S1).

  Specifically, a sound detection unit 1 having a sound collection function is installed for a semiconductor power conversion device to be monitored, and a sound generated from the semiconductor power conversion device in operation is generated by the sound collection function of the sound detection unit 1. Collected (in other words, the sound pressure signal of the sound is collected or the sound pressure level of the sound is measured).

  The specific mechanism that constitutes the sound sampling function of the sound detection unit 1 is not limited to a specific device or apparatus, but may be an operation sound generated from the semiconductor power converter (however, it may be caused by the generation of an operation sound). Appropriate devices and apparatuses are selected as appropriate for collecting (in other words, collecting and collecting sounds) including various physical quantities related to each other. For example, a sound sensor (microphone) or a vibration sensor (vibration displacement detection type sensor, speed detection type sensor, or acceleration detection type sensor) can be used as a specific mechanism that constitutes the sound sampling function of the sound detection unit 1.

  The mode of installation of the sound detection unit 1 is not limited to a specific mode, and an appropriate location or sound collection is possible so that operation sound generated from the semiconductor power conversion device to be monitored can be collected (sound collection, sound collection). The method of installation and fixing is appropriately selected. The sound detection unit 1 may be installed outside the semiconductor power conversion device or in contact with the housing, or may be installed inside the semiconductor power conversion device. good.

  The sound detection unit 1 includes an amplification function as necessary in addition to a sound collection function, and also includes an A / D conversion function.

  Then, the sound generated from the semiconductor power conversion device to be monitored is collected as sound (in other words, sound pressure or sound pressure level) by the sound detection unit 1, amplified as necessary, and converted into a digital signal. Output after being converted. Here, the digital signal output from the sound detector 1 is referred to as sound data.

  It is not an essential requirement that the sound detection unit 1 includes a sound sampling function, an amplification function as necessary, an A / D conversion function, and a signal output function as an integrated device / device. The sound detection unit 1 may be configured as a collection / combination of different devices / devices having functions.

  Further, the abnormality diagnosis device 10 of one semiconductor power conversion device may include a plurality of sound detection units 1. In this case, the following processing after S2 is performed on each of the sound data output from the plurality of sound detectors 1.

  Next, a Fourier transform process or a wavelet transform process is performed on the sound data acquired by the process of S1, and the frequency intensity is calculated (S2).

  Specifically, the sound data output from the sound detection unit 1 in the process of S1 is input to the conversion unit 2, and the conversion unit 2 performs a Fourier transform process or a wavelet transform process on the sound data, so that a frequency component is obtained. The frequency intensity P (f) at the frequency f [Hz] is calculated as the spectrum value for each. When the Fourier transform process or the wavelet transform process is performed, the frequency intensity includes time information (T) and can be expressed as “P (f, T)”. Both are expressed as “P (f)” as being subjected to the conversion process and the wavelet conversion process.

  Here, the method of calculating the spectrum value for each frequency component of the sound data is not limited to the Fourier transform process or the wavelet transform process, and the frequency intensity P (() at the frequency f [Hz] is used as the spectrum value for each frequency component. A suitable method by which f) can be calculated is appropriately selected.

  Then, the conversion unit 2 outputs the calculated frequency intensity P (f) associated with the frequency f [Hz] (in other words, as combined data of the frequency f [Hz] and the frequency intensity P (f)). Is done.

  Here, the frequency intensity P (f) at the frequency f [Hz] (in other words, the frequency f [Hz] and the frequency intensity P obtained from the sound data obtained by sampling the sound generated from the semiconductor power converter. The combination data with (f) is called determination data.

  Next, it is determined whether or not an abnormality / failure has occurred in the semiconductor power conversion device using the value of the frequency intensity calculated by the process of S2 (S3).

  Specifically, the frequency intensity P (f) output from the conversion unit 2 in the process of S2 is input to the determination unit 3, and the determination unit 3 causes an abnormality or failure in the monitored semiconductor power conversion device. It is determined whether or not there is.

The method of determining whether or not an abnormality / failure has occurred in the semiconductor power conversion device by the determination unit 3 is roughly divided into the following four types. Hereinafter, each of the following (1) to (4) will be described.
(1) Pay attention to the frequency intensity of a predetermined frequency related to the semiconductor power conversion device to be monitored.
(2) Attention is paid to the frequency intensity at two points related to the semiconductor power conversion device to be monitored.
(3) Pay attention to the time-series frequency intensity related to the semiconductor power conversion device to be monitored.
(4) Focus on the frequency intensity related to a plurality of semiconductor power converters.

(1) Method Focusing on Frequency Intensity of Predetermined Frequency Related to Monitored Semiconductor Power Converter Device In this method, the frequency intensity of a predetermined frequency in the determination data related to the monitored semiconductor power converter device is a predetermined threshold value. Is exceeded (in other words, when a sound of a predetermined frequency is collected), it is determined that an abnormality / failure has occurred in the semiconductor power conversion device.

  Here, the predetermined frequency of interest in the method (1) among the frequency intensities P (f) for each frequency f as the determination data is not limited to a specific frequency, and the semiconductor power conversion device to be monitored Is appropriately set to an appropriate frequency corresponding to the frequency of the sound that is generated when an abnormality or failure occurs in, based on the results of prior analysis and examination. Note that the predetermined frequency of interest in the method (1) may be a specific frequency f [Hz] or a frequency band centered on a specific frequency f [Hz].

  Further, the threshold value relating to the frequency intensity used in the method (1) is not limited to a specific value, and the volume of sound generated when an abnormality or failure occurs in the semiconductor power conversion device to be monitored ( In other words, it is appropriately set to an appropriate value corresponding to the sound pressure and the spectrum value based on the results of prior analysis and examination.

  Specifically, a sound is collected when a semiconductor switching element of the same type and type as that used in the semiconductor power conversion device to be monitored is intentionally failed and the semiconductor switching element fails. Data is acquired, and the spectrum value for each frequency component obtained for the sound data is analyzed, whereby the frequency or frequency band of the sound generated at the time of failure and the frequency intensity are specified. Then, based on the frequency / frequency band and frequency intensity specified for the sound at the time of failure, a predetermined frequency and a threshold relating to the frequency intensity are set.

  In addition, the state of the semiconductor power converter at the time when the determination data is acquired may be a normal operation (in other words, a continuous / continuous and steady operation), or a specific operation may be performed. You may be in the state.

  Specific examples of the specific operation of the semiconductor power conversion device at the time when the determination data is acquired include the following.

A) At startup In this case, at the time of determining whether or not an abnormality / failure has occurred, an operation sound is collected when the semiconductor power conversion device to be monitored is started up.

B) At the time of stop In this case, the operation sound at the time when the semiconductor power conversion device to be monitored is stopped is collected at the time of judging whether or not an abnormality or failure has occurred.

C) At the time of output change In this case, at the time of judging whether or not an abnormality / failure has occurred, an operation sound is collected when the monitored semiconductor power conversion device changes the output voltage or output current.

D) When switching frequency is changed In this case, at the time of determination as to whether or not an abnormality or failure has occurred, an operation sound is collected when the semiconductor power conversion device to be monitored changes the switching frequency.

  Then, the frequency intensity of the predetermined frequency of interest in the determination data acquired at the time of normal operation of the semiconductor power conversion device or at least one of the above-mentioned items a) to d) is determined in advance. It is compared with a threshold for frequency intensity.

  As a result, it is determined that an abnormality / failure has occurred in the semiconductor power conversion device when the frequency intensity of the predetermined frequency of interest exceeds a predetermined threshold relating to the frequency intensity.

  Note that the predetermined frequency of interest may be one or plural. When paying attention to a plurality of frequencies (including the frequency band as described above), for example, when the first frequency fa is less than the threshold and the second frequency fb exceeds the threshold, the semiconductor power If it is determined that an abnormality / failure has occurred in the converter, and if both the first frequency fa and the second frequency fb are below the threshold or exceed the threshold, the semiconductor power converter has an abnormality / failure. It may be determined that the occurrence of abnormality / failure of the semiconductor power conversion device is performed based on a combination of comparison results with threshold values for a plurality of frequencies. As a precaution, the example given above focuses on the first and second frequencies, but it is also possible to focus on three or more frequencies.

(2) Method Focusing on Two Time Frequency Intensities Related to Monitored Semiconductor Power Conversion Device In this method, reference data and determination data related to a monitored semiconductor power conversion device are compared, and the frequency strength and determination of these reference data are compared. When there is a difference between the frequency intensity of data, it is determined that an abnormality / failure has occurred in the semiconductor power converter.

  The reference data refers to the sound generated from the semiconductor power conversion device and collected for the sound before starting the process for determining whether or not an abnormality or failure has occurred (that is, the above-described processes of S1 and S2). Frequency intensity Po (f) at a frequency f [Hz] (in other words, combination data of the frequency f [Hz] and the frequency intensity Po (f)).

  The acquisition of the reference data is performed, for example, during a test operation of the semiconductor power conversion device to be monitored or at an initial stage of full-scale operation as an actual machine. In this case, it is expected that the semiconductor power converter is in a healthy and stable state, that is, it is expected that the semiconductor power converter is in a normal state in which no abnormality or failure has occurred. By comparing with such reference data, the device status compared with the normal status is diagnosed.

  The acquisition of the reference data may be performed after the semiconductor power conversion device is selected as a monitoring target. In this case, since the semiconductor power conversion device is operating although it cannot be said to be in the initial state, that is, the semiconductor power conversion device is operating without any problem as a full-scale operation of the actual machine, such a standard. By comparing with the data, the device state compared with the normal operation state is diagnosed.

  The reference data may be set individually for each semiconductor power conversion device to be monitored, or may be set for each type and model of the semiconductor power conversion device.

  Note that the combination data of the frequency f [Hz] and the frequency intensity Po (f) as the reference data can be referred to by the determination unit 3, in other words, for example, the determination unit 3 can read the combination data. An appropriate storage circuit is provided in the unit 3 and stored in the storage circuit.

  Here, the state of the semiconductor power conversion device at two time points in the method (2), that is, the time point when the reference data is acquired and the time point when the determination data is acquired is normal operation (in other words, continuous / continuous). A normal and steady operation) or a specific operation.

  Specific examples of the specific operation of the semiconductor power conversion device at the time when the reference data and the determination data are acquired include the following.

A) At startup In this case, when the semiconductor power converter is started up during the test operation, at the initial stage of full-scale operation or after selection as a monitoring target, and at the time of determination of whether an abnormality or failure has occurred. The operation sound is collected.

B) When stopped In this case, when the semiconductor power converter stops at the time of test operation, the initial stage of full-scale operation or the time after selection as a monitoring target, and the determination of whether or not an abnormality or failure has occurred The operation sound is collected.

C) When the output is changed In this case, the semiconductor power conversion device outputs the output voltage at the test operation, at the initial stage of the full-scale operation or after the selection as the monitoring target, and at the time of judging whether or not an abnormality or failure has occurred. And the operation sound when changing the output current.

D) When switching frequency is changed In this case, the semiconductor power converter switches at the test operation, at the initial stage of full-scale operation or after selection as a monitoring target, and at the time of judging whether or not an abnormality or failure has occurred. The operation sound when the frequency is changed is collected.

  Further, the frequency of interest in the frequency intensity P (f) for each frequency f as reference data or determination data may be an arbitrary frequency or a specific frequency.

  An arbitrary frequency as a frequency of interest in the frequency intensity P (f) of the reference data or the determination data is specifically described as a method of setting a predetermined frequency of interest in the method (1) described above, for example. It is conceivable that the setting is made by the above method.

  Specific examples of the specific frequency include the following.

i) A frequency that is an integral multiple of the switching frequency The frequency intensity Po (f) of the reference data and the frequency intensity P (f) of the determination data that are a frequency that is an integral multiple of the switching frequency of the semiconductor power conversion device to be monitored are used. In this case, the frequency may be one integer multiple of the integral multiples of the switching frequency, or may be a plurality of integer multiples.

ii) Frequency other than an integral multiple of the switching frequency The frequency intensity Po (f) of the reference data and the frequency intensity P (f) of the determination data of a frequency other than the integral multiple of the switching frequency of the semiconductor power conversion device to be monitored are used. It is done. In this case, only one certain frequency among frequencies other than an integral multiple of the switching frequency may be used, or a plurality of frequencies may be used.

  The frequency intensity in the above i) and ii) may be a value of frequency intensity at a specific frequency f [Hz] (that is, Po (f), P (f)), or It may be a feature amount such as an average value or a variance value of frequency intensity in a predetermined frequency band centered on a specific frequency f [Hz]. In this case, the width of the frequency band as the predetermined frequency band (that is, the range expressed as the frequency f1 [Hz] to f2 [Hz]) is not limited to a specific size. Appropriate size is set in consideration of the results of prior analysis and examination. Hereinafter, Po (f) and P (f) include both the value of the frequency intensity at a specific frequency f and the feature quantity of the frequency intensity in a predetermined frequency band centered on the frequency f ( In other words, it represents either one).

  The reference data and the determination data acquired at the time of normal operation of the semiconductor power conversion device or at least one of the above-mentioned items a) to d) are set to any frequency or the above-described i) and ii). The frequency intensities Po (f) and P (f) of at least one of the frequencies are compared.

  As a result, when there is a difference between these two frequency intensities Po (f) and P (f), it is determined that an abnormality / failure has occurred in the semiconductor power converter.

  As an index for determining whether or not there is a difference between the two frequency intensities Po (f) and P (f), for example, an absolute error (ie, | P (f) −Po (f) | ), Relative error (ie, | P (f) −Po (f) | / Po (f)), or ratio (ie, P (f) / Po (f)) or the like.

  When the index values as exemplified above are not more than a predetermined threshold corresponding to each index, there is no difference between the two frequency intensities Po (f) and P (f). On the other hand, if it is greater than the threshold, it is determined that there is a difference between the two frequency intensities Po (f) and P (f).

  The threshold value for each index is not limited to a specific value, and is appropriately set to an appropriate value in consideration of, for example, a prior analysis / consideration result.

  Further, only one type of index value is used, and it is determined whether there is a difference between the two frequency intensities Po (f) and P (f) by comparison with a threshold value related to the one type of index value. Alternatively, a difference between the two frequency intensities Po (f) and P (f) may be obtained by using a plurality of types of index values and comparing the threshold values with respect to each of the plurality of types of index values. It may be determined whether or not there is.

  In addition, machine learning (also referred to as pattern learning) may be used in determining whether or not an abnormality / failure has occurred in the semiconductor power conversion device by comparing the reference data with the determination data.

  In this case, the frequency intensity Po (f) of the reference data and the frequency intensity P (f) of the determination data are used, or the feature amount obtained from the frequency intensity Po (f) and the frequency intensity P (f ) Is used, and the frequency intensity P (f) or its feature quantity is different from the frequency intensity Po (f) or its feature quantity (corresponding to teacher data in machine learning). At some time, it is determined that an abnormality or failure has occurred in the semiconductor power conversion device.

  Further, instead of the reference data, a sound when an abnormality / failure occurs in the semiconductor power converter is collected and the frequency intensity Pw (f) (“failure data”) at the frequency f [Hz] obtained for the sound is collected. May be used.

  In this case, the frequency intensity Pw (f) of the failure data and the frequency intensity P (f) of the determination data are used, or the feature amount obtained from the frequency intensity Pw (f) and the frequency intensity P (f ) Is used, and the frequency intensity P (f) or a feature quantity thereof is the same as or similar to the frequency intensity Pw (f) or a feature quantity thereof (corresponding to teacher data in machine learning). It is determined that an abnormality or failure has occurred in the semiconductor power conversion device when the pattern is

(3) Method Focusing on Time-Series Frequency Intensity for Monitored Semiconductor Power Converter Device In this method, the monitored semiconductor power converter device operates normally (in other words, continuous, continuous and steady operation). When the determination data in the state of performing the monitoring is monitored in time series and a specific change occurs in the frequency intensity in the time series, it is determined that an abnormality / failure has occurred in the semiconductor power conversion device.

  Of the frequency intensity P (f) for each frequency f as the determination data, the frequency of interest may be an arbitrary frequency or a specific frequency.

  Specific examples of the specific frequency as the frequency of interest in the frequency intensity P (f) of the determination data include i) and ii) of (2) above.

  Then, the frequency intensity P (f) of an arbitrary frequency or at least one of the above-mentioned (2) i) and ii) of the determination data acquired during the normal operation of the semiconductor power conversion device. ) Are compared in chronological order.

  As a result, when a specific change occurs in the time series frequency intensity P (f) (in other words, the time series transition of the frequency intensity), it is determined that an abnormality or failure has occurred in the semiconductor power converter. Is done.

  How to determine whether or not there is a specific change in time-series frequency intensity P (f, t) (where t represents an identifier given at the time when sound data relating to the frequency intensity is acquired) For example, the absolute error (that is, | P (f, t2) between the frequency intensity P (f, t2) at a certain time t2 and the frequency intensity P (f, t1) at the time t1 before the time t2 ) −P (f, t1) |), relative error (ie, | P (f, t2) −P (f, t1) | / P (f, t1)), or ratio (ie, P (f, t2) ) / P (f, t1)) or the like.

  Alternatively, the frequency intensity P (f, t2) at the time t2 and the frequency intensity P (f, t1) at the time t1 are used, or the feature amount obtained from the frequency intensity P (f, t2) and the frequency intensity. A feature amount obtained from P (f, t1) may be used, and machine learning (pattern learning) may be used. In this case, when the frequency intensity P (f, t2) or the feature amount thereof is a pattern different from the frequency intensity P (f, t1) or the feature amount thereof, the semiconductor power conversion device has an abnormality / failure. It is determined that it has occurred.

  In the above case, the time t2 in the above description becomes the time t1 in the next process, and the time next to the time t2 in the above description becomes a new time t2 in the next process. And t2 are similarly updated.

  As a method of determining whether or not a peculiar change has occurred in the frequency intensity P (f, t) of the time series, the frequency intensity P (f, t2) at a certain time t2 and before the time t2 can be used. Average value Pav (f, tb) of frequency intensity in a predetermined period before time t1 (that is, a past period before time t2) (where tb is a period during which sound data relating to the frequency intensity is acquired) Absolute error (ie, | P (f, t 2) −Pav (f, tb) |), relative error (ie, | P (f, t 2) −Pav (f, tb)) ) | / Pav (f, tb)), or ratio (ie, P (f, t2) / Pav (f, tb)) or the like. Note that the time length as the predetermined period in this case is not limited to a specific length, and is appropriately set to an appropriate length in consideration of, for example, a prior analysis / consideration result.

  In the above case, machine learning (pattern learning) is applied to the frequency intensity P (f, t2) at the time t2 and the average value Pav (f, tb) of the frequency intensity in a predetermined period, and both are different. When it is a pattern, it may be determined that an abnormality or failure has occurred in the semiconductor power conversion device.

  In the above case, the time t2 in the above description becomes the time t1 in the next process, the time next to the time t2 in the above description becomes a new time t2 in the next process, and the average value Pav ( f, tb) is updated, and in the subsequent processing, the times t1 and t2 and the average value Pav (f, tb) of the frequency intensity are similarly updated.

  When the index value as exemplified above is equal to or less than a predetermined threshold corresponding to each index, the time series frequency intensity P (f, t) (in other words, the frequency intensity time series). On the other hand, when it is larger than the threshold, it is determined that there is a peculiar change in the frequency intensity P (f, t) in the time series.

  The threshold value for each index is not limited to a specific value, and is appropriately set to an appropriate value in consideration of, for example, a prior analysis / consideration result.

  Further, only one type of index value is used, and it is determined whether or not there is a specific change in the time series frequency intensity P (f, t) by comparison with a threshold value related to the one type of index value. Alternatively, whether or not there are peculiar changes in the frequency intensity P (f, t) in time series by comparing a plurality of types of index values with a threshold value for each of these types of index values. May be determined.

(4) Method Focusing on Frequency Intensity for Multiple Semiconductor Power Converters In this method, multiple semiconductor power converters including at least the monitored semiconductor power converters (that is, a plurality of all the monitored power converters) The determination data for each of the power conversion devices may be compared, and when the frequency intensities of the plurality of determination data are compared side by side, the frequency intensity of one determination data and the frequency intensity of other determination data Is determined to be abnormal or faulty in the semiconductor power conversion device from which the certain determination data has been acquired.

  Specifically, the plurality of units related to the semiconductor power conversion device in this method is three or more units.

  In the case of this method, the process of S1 is performed in each of the plurality of semiconductor power converters, and the process of S2 is performed for each of the sound data for each semiconductor power converter, corresponding to each semiconductor power converter. Thus, a plurality of frequency intensities P (N, f) output from the converter 2 in the process of S2 (where N is an identifier assigned to each semiconductor power conversion device in order to distinguish between the plurality of semiconductor power conversion devices) And represents the identifier assigned to the semiconductor power converter from which the sound data related to the frequency intensity has been acquired). In addition, it is preferable that the timing which the several frequency intensity | strength P (N, f) corresponding to each semiconductor power converter device is input into the determination part 3 is synchronized.

  Here, the state of the semiconductor power conversion device at the time when the determination data is acquired may be a normal operation (in other words, a continuous, continuous and steady operation) or a specific operation. You may be in a state of doing.

  Specific examples of the specific operation of the semiconductor power conversion device at the time when the determination data is acquired include (1) a) to d) described above.

  Further, the frequency of interest in the frequency intensity P (f) for each frequency f as determination data may be an arbitrary frequency or a specific frequency.

  Specific examples of the specific frequency as the frequency of interest in the frequency intensity P (f) of the determination data include i) and ii) of (2) above.

  The determination data of the plurality of semiconductor power conversion devices acquired at the time of normal operation of the semiconductor power conversion device or at least one of the above-mentioned (1) a) to d) Frequency intensities P (N, f) of the frequency or at least one of the above-mentioned (2) i) and ii) are compared side by side.

  As a result, when there is a difference between the frequency intensity of a certain determination data and the frequency intensity of other determination data, the semiconductor power conversion device from which the certain determination data is acquired has an abnormality / failure. It is determined that it has occurred.

  In addition, it is determined that an abnormality / failure has occurred in the semiconductor power conversion device from which the determination data having a pattern different from other patterns is obtained by applying machine learning (pattern learning) to a plurality of determination data. You may do it.

  The above is the description of (1) to (4) regarding how the determination unit 3 determines whether or not an abnormality / failure has occurred in the semiconductor power conversion device.

  If it is not determined that an abnormality / failure has occurred in the semiconductor power conversion device to be monitored (S3: No), the process returns to S1 and the processes from S1 to S3 are repeated. .

  Note that the processing of S1 to S3 is repeated according to the output pitch of the sound data from the sound detection unit 1 (in other words, triggered by the output of the sound data from the sound detection unit 1), or in advance. Repeated by a predetermined processing pitch or processing trigger.

  On the other hand, if it is determined that an abnormality / failure has occurred in the monitored semiconductor power conversion device (S3: Yes), the determination unit 3 causes the abnormality / failure in the monitored semiconductor power conversion device. A predetermined signal (referred to as an “abnormality notification signal”) for notifying that the error has occurred is output.

  And when an abnormality notification signal is output by the process of S3, a semiconductor power converter stops an operation | movement or a warning is issued (S4).

  Specifically, for example, the abnormality notification signal is output from the determination unit 3 of the abnormality diagnosis device 10 of the semiconductor power conversion device to the control device / control unit that controls the operation of the semiconductor power conversion device to be monitored. The target semiconductor power conversion device stops operating.

  In addition, an abnormality notification signal is output from the determination unit 3 of the abnormality diagnosis device 10 of the semiconductor power conversion device to the alarm output unit 4, thereby causing the alarm output unit 4 to externally (specifically, for example, to the semiconductor power conversion device). Warnings are issued to workers and managers involved.

  In accordance with the abnormality notification signal from the determination unit 3, the alarm output unit 4 emits a sound by a speaker, a buzzer, or the like, lights a light by a warning light or a rotating light, or warns a display or the like. An alarm is issued by displaying a message or vibrating with a vibration function.

  It should be noted that only one of the stoppage of the operation of the semiconductor power conversion device to be monitored and the issuing of an alarm may be performed, or both may be performed.

  Here, the abnormality diagnosis device 10 for the semiconductor power conversion device described above may be configured such that all the components are integrated and installed in or near the monitored semiconductor power conversion device. May be distributed and installed in a plurality of locations.

  Specifically, for example, the sound detection unit 1 is installed in or near the monitored semiconductor power conversion device, and the conversion unit 2, the determination unit 3, and the alarm output unit 4 are located away from the semiconductor power conversion device. The sound detection unit 1, the conversion unit 2, and the determination unit 3 are installed in or near the semiconductor power conversion device to be monitored and the alarm output unit 4 is separated from the semiconductor power conversion device. The sound detection unit 1 is installed in or near the monitored semiconductor power conversion device, and the conversion unit 2 and the determination unit 3 are installed with the semiconductor power conversion device. The alarm output unit 4 may be installed in a location (for example, a management control room or a monitoring center) away from the semiconductor power conversion device after being installed in a compartment (for example, indoors). Be

  Moreover, each part which needs to be connected among each part which comprises the abnormality diagnosis apparatus 10 of a semiconductor power converter device can transmit / receive (namely, input / output) signals, such as data and a control command. Is electrically connected.

  Specifically, when the units constituting the abnormality diagnosis device 10 of the semiconductor power converter are configured as an integrated unit, the units that need to be connected are appropriately connected, for example, by a signal line such as a bus.

  Moreover, when each part which comprises the abnormality diagnosis apparatus 10 of a semiconductor power converter device is comprised as a different thing (further when each part is disperse | distributed and installed in several places), each part which needs a connection As appropriate, for example, it is connected by a wired signal transmission / reception mechanism in which cables laid and connected to each other are used, or by a wireless signal transmission / reception mechanism in which a wireless signal transceiver connected to each is used Or these signal transmission / reception mechanisms are combined and connected.

<When an abnormality diagnosis program for a semiconductor power converter is executed on a computer>
The above-described conversion unit 2, determination unit 3, and alarm output unit 4 may be realized by the computer by executing an abnormality diagnosis program of the semiconductor power conversion device on the computer.

  FIG. 3 shows the overall configuration of the computer 20 for executing the abnormality diagnosis program 17 for the semiconductor power converter.

  The computer 20 includes a control unit 11, a storage unit 12, an input unit 13, a display unit 14, and a memory 15, which are connected to each other by a signal line such as a bus.

  The control unit 11 performs control related to overall control of the computer 20 and diagnosis / detection of abnormality / failure of the semiconductor power conversion device by the abnormality diagnosis program 17 of the semiconductor power conversion device stored in the storage unit 12, For example, a CPU (Central Processing Unit).

  The storage unit 12 is a device that can store at least data and programs, and is, for example, a hard disk.

  The input unit 13 is an interface (that is, an information input mechanism) for giving at least an operator's command and various information to the control unit 11, and is, for example, a keyboard, a mouse, or a touch panel. For example, a plurality of types of interfaces such as a keyboard and a mouse may be provided as the input unit 13.

  The display unit 14 performs drawing / display of characters, figures, images, and the like under the control of the control unit 11 and is, for example, a display.

  The memory 15 serves as a memory space that is a work area when the control unit 11 executes various controls and operations, and is, for example, a RAM (abbreviation of Random Access Memory).

  Further, in the computer 20, specifically, for example, a wired signal transmission / reception mechanism as described above so that the sound detection unit 1 can transmit and receive (ie, input / output) signals such as data and control commands. Or a wireless signal transmission / reception mechanism or a combination of these signal transmission / reception mechanisms.

  The control unit 11 of the computer 20 executes the abnormality diagnosis program 17 of the semiconductor power conversion device, and thereby the sound output from the sound detection unit 1 that collects sound generated from the monitored semiconductor power conversion device. A data receiving unit 11a that performs processing for receiving data, a conversion unit 11b that performs processing for calculating a spectral value for each frequency component of sound data, and a semiconductor power conversion device to be monitored using the spectral value for each frequency component A determination unit 11c that performs a process of outputting an abnormality notification signal when determining whether an abnormality / failure has occurred and determining that an abnormality / failure has occurred, and a process of issuing an alarm according to the abnormality notification signal An alarm output unit 11d to perform is configured.

  The sound detection unit 1 performs the above-described processing of S1 and outputs sound data. The sound data is received by the data receiving unit 11a of the computer 20, stored in the memory 15, and stored in the memory 15. The sound data is used, the conversion unit 11b performs the same process as the conversion unit 2 described above as the process of S2, and the determination unit 11c performs the same process as the determination unit 3 described above as the process of S3. When it is determined that an abnormality / failure has occurred in the semiconductor power conversion device, an abnormality notification signal is output, and the alarm output unit 11d performs the same process as the alarm output unit 4 described above as the process of S4. .

  Note that the computer 20 may not include the alarm output unit 11d. In this case, the computer 20 outputs an abnormality notification signal to an alarm output unit provided separately from the computer 20, or controls the operation of the monitored semiconductor power conversion device. An abnormality notification signal may be output to the apparatus / control unit.

<When the semiconductor power conversion device itself includes an abnormality diagnosis device for the semiconductor power conversion device>
The above-described semiconductor power conversion device abnormality diagnosis device 10 may be provided in the semiconductor power conversion device itself to be monitored. The overall configuration in this case is shown in FIG.

  In this case, from the determination unit 3 of the semiconductor power converter abnormality diagnosis device 10 to the control device / control unit that controls the operation of the semiconductor power conversion device provided with the semiconductor power conversion device abnormality diagnosis device 10. On the other hand, it is conceivable that an abnormality notification signal is output, whereby the monitored semiconductor power conversion device is configured to stop operating.

  In addition, an abnormality notification signal is output from the determination unit 3 of the abnormality diagnosis device 10 of the semiconductor power conversion device to the alarm output unit 4, thereby causing the alarm output unit 4 to externally (specifically, for example, to the semiconductor power conversion device). For example, an alarm may be issued to a worker or manager involved) by displaying sound, light, or a message.

  According to the abnormality diagnosis method, abnormality diagnosis apparatus, abnormality diagnosis program, and semiconductor power conversion apparatus including the abnormality diagnosis apparatus having the above-described configuration, the semiconductor power is based on the sound generated from the semiconductor power conversion apparatus. Since the determination of the occurrence of abnormality / failure in the converter is performed, the abnormality in the semiconductor power converter is captured before changes in the output to the outside such as the voltage and current of the semiconductor power converter are observed, Furthermore, it is possible to capture signs of abnormality in the semiconductor power converter. For this reason, it is possible to prevent a semiconductor power conversion device from being stopped due to a failure, and it is possible to improve the usefulness and reliability as a method for detecting an abnormality or failure of the semiconductor power conversion device. Become.

  According to the abnormality diagnosis method, abnormality diagnosis device, abnormality diagnosis program, and semiconductor power conversion device including the abnormality diagnosis device having the above-described configuration, the output of voltage or current is directly measured. In this way, it is possible to determine whether or not an abnormality or failure has occurred in the semiconductor power conversion device by collecting sound generated from the semiconductor power conversion device without intervening in the semiconductor power conversion device. It is possible to constantly monitor the conversion device while performing normal operation. For this reason, it becomes possible to always detect abnormality / failure in the semiconductor power converter without relying on regular maintenance and inspection, and as a result, usefulness as a method for detecting abnormality / failure in the semiconductor power converter. Reliability can be improved.

  Although the above-described embodiment is an example of a preferred embodiment for carrying out the present invention, the embodiment of the present invention is not limited to the above-described embodiment, and the present invention is not deviated from the gist of the present invention. Various modifications can be made.

  For example, in the above-described embodiment, as an abnormality diagnosis method for the semiconductor power converter, the semiconductor power converter stops operating or an alarm is issued according to the abnormality notification signal (S4). The alarm output unit 4 may be provided as 10 or the alarm output unit 11d may be configured in the control unit 11 of the computer 20 by executing the abnormality diagnosis program 17 of the semiconductor power converter. It is not an essential configuration in the present invention to provide an operation stop of the power conversion device, an alarm issuance, or an alarm output unit 4 or 11d. There are various ways of using the device, and various devices and devices may be used in combination. .

  In the above-described embodiment, although the type of abnormality / failure detected in the semiconductor power conversion device to be monitored is not identified, the abnormality in addition to the determination processing for the occurrence of abnormality / failure occurs. A failure type identification process may be performed. When the type of abnormality / failure is identified, specifically, for example, the frequency intensity at the frequency f [Hz] as attribute data indicating what frequency component the spectrum value becomes large for each type of abnormality / failure. (Referred to as “abnormal sound data”) is prepared in advance, and the frequency intensity P (f) obtained in the process of S2 is compared with the abnormal sound data, and the frequency intensity P (f) is equal to or greater than a predetermined value. The type of abnormality / failure is identified based on the frequency f [Hz] (in other words, based on the degree of similarity between the frequency intensity P (f) and the abnormal sound data). The types of abnormalities / failures are specifically, for example, abnormalities / failures of semiconductor switching elements, abnormalities / failures of circuit boards, or abnormalities / failures of capacitor elements. Further, the method for evaluating the degree of similarity between the frequency intensity P (f) and abnormal sound data is not limited to a specific method, and the degree of similarity / correlation of features of a plurality of data groups can be determined. An appropriate method capable of weighing is appropriately selected.

  The abnormality diagnosis technology according to the present invention can accurately diagnose and detect an abnormality / fault in a semiconductor power conversion device. For example, AC / DC forward conversion and reverse conversion, frequency conversion, and voltage adjustment are performed by semiconductor power. The utility value is high in fields such as power generation, power transmission and distribution performed by the converter.

DESCRIPTION OF SYMBOLS 1 Sound detection part 2 Conversion part 3 Determination part 4 Alarm output part 10 Abnormality diagnosis apparatus 11 of semiconductor power converter 11 Control part 11a Data receiving part 11b Conversion part 11c Determination part 11d Alarm output part 12 Storage part 13 Input part 14 Display part 15 Memory 17 Abnormality diagnosis program 20 for semiconductor power conversion device Computer

Claims (12)

  A spectrum value for each frequency component is calculated for sound data acquired by collecting sound generated from the monitoring target semiconductor power conversion device, and the monitoring target semiconductor power conversion device is used using the spectrum value for each frequency component. A method for diagnosing abnormality in a semiconductor power conversion device, wherein the presence / absence of occurrence of an abnormality / failure is determined.   Spectral values for each frequency component are calculated for each of the sound data obtained by collecting the sound generated from the monitored semiconductor power conversion device in time series, and the spectral values for each frequency component in the time series are used. An abnormality diagnosis method for a semiconductor power conversion device, characterized in that it is determined whether or not an abnormality / failure has occurred in the monitored semiconductor power conversion device.   Spectral values for each frequency component are calculated for each of the sound data obtained by collecting sounds obtained from a plurality of semiconductor power conversion devices including at least the semiconductor power conversion device to be monitored. A method for diagnosing an abnormality in a semiconductor power conversion device, wherein a spectrum value for each frequency component is used to determine whether or not an abnormality / failure has occurred in the monitored semiconductor power conversion device.   The sound sampling is performed at least one of when the monitored semiconductor power conversion device starts, stops, changes the output voltage or output current, or changes the switching frequency. The abnormality diagnosis method for a semiconductor power conversion device according to claim 1 or 3, wherein:   The spectrum value for each frequency component is a spectrum value of a frequency that is an integer multiple of a switching frequency of the semiconductor power conversion device to be monitored or a spectrum value of a frequency other than an integer multiple of the switching frequency. The abnormality diagnosis method for a semiconductor power conversion device according to any one of 1 to 3.   Means for collecting sound generated from the semiconductor power conversion device to be monitored and outputting sound data; means for calculating a spectrum value for each frequency component of the sound data; and using the spectrum value for each frequency component. An abnormality diagnosis device for a semiconductor power conversion device, comprising: means for determining whether or not an abnormality / failure has occurred in a semiconductor power conversion device to be monitored.   The sound sampling is performed at least one of when the monitored semiconductor power conversion device starts, stops, changes the output voltage or output current, or changes the switching frequency. The abnormality diagnosis device for a semiconductor power conversion device according to claim 6.   The spectrum value for each frequency component is a spectrum value of a frequency that is an integer multiple of a switching frequency of the semiconductor power conversion device to be monitored or a spectrum value of a frequency other than an integer multiple of the switching frequency. 6. The abnormality diagnosis device for a semiconductor power conversion device according to 6.   Processing for calculating a spectral value for each frequency component of sound data acquired by collecting sound generated from the semiconductor power conversion device to be monitored, and the semiconductor power conversion for the monitoring target using the spectral value for each frequency component An abnormality diagnosis program for a semiconductor power conversion device, which causes a computer to perform processing for determining whether or not an abnormality or failure has occurred in the device.   The sound sampling is performed at least one of when the monitored semiconductor power conversion device starts, stops, changes the output voltage or output current, or changes the switching frequency. The abnormality diagnosis program for a semiconductor power conversion device according to claim 9, wherein   The spectrum value for each frequency component is a spectrum value of a frequency that is an integer multiple of a switching frequency of the semiconductor power conversion device to be monitored or a spectrum value of a frequency other than an integer multiple of the switching frequency. 9. An abnormality diagnosis program for a semiconductor power conversion device according to 9.   A semiconductor power conversion device having an abnormality diagnosis function, comprising the abnormality diagnosis device for a semiconductor power conversion device according to any one of claims 6 to 8.

Images