Adaptive monitoring method for insulated gate bipolar transistors based on multi-objective optimization

Abstract

The present invention relates to the technical field of insulated gate bipolar transistor monitoring, and an insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization, comprising: calculating a test timing matching similarity set of test fault timing data and training fault timing data, judging whether there is a target test timing similarity greater than a matching judgment threshold in the test timing matching similarity set, if not, updating the initial matching period, if so, identifying the target test fault type sequence, adjusting the original index matching weight sequence according to the judgment accuracy, obtaining the target index matching weight sequence, judging whether there is a target current timing similarity, if not, updating the initial matching period, if so, performing a fault warning according to the target current fault type. The present invention can solve the problems of poor monitoring efficiency and low fault identification accuracy in the current IGBT monitoring method.

Description

Technical Field

[0001] The present invention relates to the technical field of insulated gate bipolar transistor monitoring, and in particular to an insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization. Background Art

[0002] An insulated gate bipolar transistor (IGBT) is an electronic component that combines the advantages of a field effect transistor (MOSFET) and a bipolar junction transistor (BJT).

[0003] With the continuous development of IGBT technology and the expansion of its application fields, the requirements for IGBT performance monitoring and fault warning are becoming higher and higher. Traditional monitoring methods are often based on a single target or parameter, such as current, voltage, temperature, etc., but these monitoring methods are often difficult to fully and accurately reflect the working status and performance changes of IGBT when facing complex power electronic systems. Therefore, the current IGBT monitoring methods have the problems of poor monitoring efficiency and low fault identification accuracy. Summary of the invention

[0004] The present invention provides an insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization, the main purpose of which is to solve the problems of poor monitoring efficiency and low fault identification accuracy in the current IGBT monitoring method.

[0005] To achieve the above object, the present invention provides an insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization, comprising:

[0006] Acquire a historical fault time series data set, classify the historical fault time series data set, and obtain a training fault time series data set and a test fault time series data set;

[0007] Extracting test fault timing data in sequence from the test fault timing data set;

[0008] Calculate the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set;

[0009] Determine whether there is a target test timing similarity greater than a preset matching determination threshold in the test timing matching similarity set;

[0010] If there is no target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, then the initial matching period is updated and the step of calculating the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence is returned to the above step;

[0011] If there is a target test timing similarity greater than a matching determination threshold in the test timing matching similarity set, identifying a target test fault type corresponding to the target test timing similarity to obtain a target test fault type sequence;

[0012] Calculating the determination accuracy of the target test fault type sequence, and adjusting the original indicator matching weight sequence using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence;

[0013] Get current running data;

[0014] Determine whether the current operating data contains preset abnormal indicator data;

[0015] If the current running data does not contain abnormal indicator data, return to the above step of obtaining the current running data;

[0016] If the current operating data has abnormal indicator data, then the current time series matching similarity set is calculated according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence;

[0017] Determine whether there is a target current time series similarity greater than a matching determination threshold in the current time series matching similarity set;

[0018] If there is no target current time series similarity greater than the matching determination threshold in the current time series matching similarity set, then the initial matching period is updated and the step of calculating the current time series matching similarity set according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence is returned to the above step;

[0019] If there is a target current time sequence similarity greater than the matching determination threshold in the current time sequence matching similarity set, the target current fault type corresponding to the target current time sequence similarity is identified, and a fault warning is performed according to the target current fault type.

[0020] Optionally, obtaining a historical fault time series data set includes:

[0021] Obtain historical time series data sets and indicator normal interval threshold sets;

[0022] Sequentially extracting historical time series data from the historical time series data set, wherein the historical time series data set refers to a data set of changes in monitoring values ​​of various preset monitoring indicators over time, and the indicator normal interval threshold set refers to a threshold set for determining whether various monitoring indicators are in an abnormal state;

[0023] Determine whether the historical time series data is abnormal historical time series data according to the indicator normal interval threshold set;

[0024] If the historical time series data is not abnormal historical time series data, returning to the above step of sequentially extracting historical time series data from the historical time series data set;

[0025] If the historical time series data is abnormal historical time series data, identifying the historical fault type and abnormal data time series point of the historical time series data according to the indicator normal interval threshold set;

[0026] According to the abnormal data time series point, the historical time series data is segmented into abnormal time series points and the fault type is marked using the historical fault type to obtain the historical fault time series data;

[0027] The historical fault time series data of each abnormal historical time series data are collected to obtain a historical fault time series data set.

[0028] Optionally, the calculating of the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set includes:

[0029] Extracting training fault time series data in sequence from the training fault time series data set;

[0030] Performing time series data interception on the training fault time series data and the test fault time series data according to the initial matching period to obtain training matching time series data and test matching time series data;

[0031] The test time series matching similarity between the training matching time series data and the test matching time series data is calculated using the pre-constructed matching formula and the original indicator matching weight sequence, wherein the matching formula is as follows:

[0032]

[0033] Among them, s T represents the test time series matching similarity between the training matching time series data and the test matching time series data when the initial matching period is T, n represents the number of monitoring indicators, j represents the sequence number of the monitoring indicator, t represents the sequence number of the matching time point in the initial matching period, k jrepresents the original indicator matching weight of the jth monitoring indicator in the original indicator matching weight sequence, |*| represents the absolute value symbol, It represents the monitoring value of the jth monitoring indicator at the tth matching time point in the training matching time series data. Indicates the monitoring value of the jth monitoring indicator at the tth matching time point in the test matching time series data;

[0034] The test timing matching similarities between each training fault timing data and the test fault timing data are summarized to obtain a test timing matching similarity set.

[0035] Optionally, updating the initial matching period includes:

[0036] The initial matching period is updated using a preset matching period update formula, wherein the matching period update formula is as follows:

[0037] T′=T+Δt

[0038] Wherein, T′ represents the initial matching period after the update, T represents the initial matching period before the update, and Δt represents the update duration.

[0039] Optionally, the calculating the determination accuracy of the target test fault type sequence includes:

[0040] Identify the labeled test fault type corresponding to the target test timing similarity to obtain a labeled test fault type sequence;

[0041] Serializing and grouping the labeled test fault type sequence and the target test fault type sequence to obtain a test fault type group set, wherein the test fault type groups in the test fault type group set include labeled test fault types and target test fault types with the same serial number;

[0042] Extract test fault type groups in the test fault type group set in sequence, and determine whether the marked test fault type and the target test fault type in the test fault type group are the same;

[0043] If the marked test fault type and the target test fault type in the test fault type group are the same, the test fault type group is used as the correct judgment type group;

[0044] If the marked test fault type and the target test fault type in the test fault type group are different, the test fault type group is used as the error determination type group;

[0045] Collect all correctly determined type groups and incorrectly determined type groups to obtain correctly determined type group sets and incorrectly determined type group sets;

[0046] The determination accuracy of the target test fault type sequence is calculated using the following formula:

[0047]

[0048] Among them, δ represents the judgment accuracy of the target test fault type sequence, z T represents the number of correctly determined type groups in the correctly determined type group set, z F Indicates the number of error judgment type groups in the error judgment type group set.

[0049] Optionally, the original indicator matching weight sequence is adjusted using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence, including:

[0050] The original indicator matching weight sequence is randomly initialized according to a preset population number to obtain multiple groups of initial indicator matching weight sequences, wherein the number of groups of the initial indicator matching weight sequence is equal to the number of populations;

[0051] Identify an initial test fault type sequence of each group of initial indicator matching weight sequences in the multiple groups of initial indicator matching weight sequences according to the training fault timing data set and the test fault timing data set;

[0052] Calculating the initial determination accuracy of the initial test fault type sequence to obtain an initial determination accuracy set;

[0053] Identify a maximum initial accuracy rate in the initial determination accuracy rate set, and identify a maximum indicator matching weight sequence corresponding to the maximum initial accuracy rate;

[0054] Taking the maximum index matching weight sequence as the global optimal weight sequence;

[0055] Randomly initialize the weight update speed sequence of each group of initial indicator matching weight sequence to obtain multiple groups of weight update speed sequences, identify the historical optimal weight sequence of each group of initial indicator matching weight sequence, and obtain multiple groups of historical optimal weight sequences;

[0056] Calculate the speed update sequence of each group of initial indicator matching weight sequence according to the preset inertia weight, acceleration coefficient pair, random number pair, multiple groups of weight update speed sequences, multiple groups of historical optimal weight sequences and the global optimal weight sequence to obtain multiple groups of speed update sequences;

[0057] Calculate the update index matching weight sequence of each group of initial index matching weight sequence according to the multiple groups of speed update sequences and the multiple groups of initial index matching weight sequences to obtain multiple groups of update index matching weight sequences;

[0058] Calculating an update test fault type sequence for each set of update index matching weight sequences in the plurality of sets of update index matching weight sequences;

[0059] Calculating the update determination accuracy of the update test fault type sequence to obtain an update determination accuracy set;

[0060] identifying a maximum update accuracy rate in the update determination accuracy rate set;

[0061] Determining whether the maximum update accuracy is greater than a preset update accuracy threshold;

[0062] If the maximum update accuracy is not greater than the update accuracy threshold, then obtaining the current update number, and determining whether the current update number is greater than a preset update number threshold;

[0063] If the current update number is not greater than the update number threshold, the multiple groups of initial indicator matching weight sequences are updated using the multiple groups of updated indicator matching weight sequences, and the step of returning to the above step of identifying the initial test fault type sequence of each group of initial indicator matching weight sequences in the multiple groups of initial indicator matching weight sequences according to the training fault timing data set and the test fault timing data set is returned;

[0064] If the current update number is greater than the update number threshold, the global optimal weight sequence is used as the target indicator matching weight sequence;

[0065] If the maximum update accuracy is greater than the update accuracy threshold, then a target indicator matching weight sequence corresponding to the maximum update accuracy is identified.

[0066] Optionally, the identifying of the historical optimal weight sequence of each group of initial indicator matching weight sequences to obtain multiple groups of historical optimal weight sequences includes:

[0067] Identify multiple groups of historical weight sequences for each group of initial indicator matching weight sequences, wherein the number of groups of the historical weight sequences is equal to the current update number + 1;

[0068] Identify the historical judgment accuracy sets of multiple groups of historical weight sequences;

[0069] Identifying a maximum historical accuracy rate in the historical determination accuracy rate set;

[0070] The historical optimal weight sequence corresponding to the maximum historical accuracy is identified, and the historical optimal weight sequence corresponding to each group of initial indicator matching weight sequences is collected to obtain multiple groups of historical optimal weight sequences.

[0071] Optionally, the speed update sequence of each group of initial indicator matching weight sequence is calculated according to the preset inertia weight, acceleration coefficient pair, random number pair, multiple groups of weight update speed sequences, multiple groups of historical optimal weight sequences and the global optimal weight sequence to obtain multiple groups of speed update sequences, including:

[0072] Extracting weight update speed sequences in sequence from the multiple groups of weight update speed sequences, and extracting historical optimal weight sequences in sequence from the multiple groups of historical optimal weight sequences;

[0073] The speed update sequence of each group of initial indicator matching weight sequence is calculated using the pre-built speed update formula according to the inertia weight, acceleration coefficient pair, random number pair, weight update speed sequence, historical optimal weight sequence and the global optimal weight sequence, wherein the speed update formula is as follows:

[0074]

[0075] Among them, V1 represents the first update speed in the speed update sequence, ω represents the inertia weight, v1 represents the first weight update speed in the weight update speed sequence, c1 represents the first acceleration coefficient in the acceleration coefficient pair, and r1 represents the first random number in the random number pair. represents the first historical optimal weight in the historical optimal weight sequence, q1 represents the first initial indicator matching weight in the initial indicator matching weight sequence, c2 represents the second acceleration coefficient in the acceleration coefficient pair, and r2 represents the second random number in the random number pair. represents the first global optimal weight in the global optimal weight sequence, V n represents the nth update speed in the speed update sequence, v n represents the nth weight update speed in the weight update speed sequence, represents the nth historical optimal weight in the historical optimal weight sequence, q n represents the nth initial indicator matching weight in the initial indicator matching weight sequence, represents the nth global optimal weight in the global optimal weight sequence;

[0076] The speed update sequences corresponding to the initial indicator matching weight sequences are collected to obtain multiple groups of speed update sequences.

[0077] Optionally, the step of calculating an updated index matching weight sequence of each group of initial index matching weight sequences according to the multiple groups of speed update sequences and the multiple groups of initial index matching weight sequences to obtain multiple groups of updated index matching weight sequences includes:

[0078] Extracting speed update sequences in sequence from the multiple groups of speed update sequences, and extracting initial indicator matching weight sequences in sequence from the multiple groups of initial indicator matching weight sequences;

[0079] The pre-constructed weight update formula is used to calculate the updated index matching weight sequence according to the speed update sequence and the initial index matching weight sequence, wherein the weight update formula is as follows:

[0080]

[0081] Among them, q′1 represents the first update index matching weight in the update index matching weight sequence, q′ n represents the nth update indicator matching weight in the update indicator matching weight sequence;

[0082] The updated indicator matching weight sequences of each group of initial indicator matching weight sequences are collected to obtain multiple groups of updated indicator matching weight sequences.

[0083] Optionally, the calculating of the current time series matching similarity set according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence includes:

[0084] Identify the current abnormal data timing point of the current operation data, and segment the current operation data according to the current abnormal data timing point to obtain current fault timing data;

[0085] The current timing matching similarity between the current fault timing data and each historical fault timing data in the historical fault timing data set is calculated according to the initial matching period and the target indicator matching weight sequence to obtain the current timing matching similarity.

[0086] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0087] A memory storing at least one instruction; and a processor executing the instruction stored in the memory to implement the above-mentioned insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization.

[0088] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization.

[0089] The present invention solves the problem described in the background technology, and mainly predicts the fault type of the current operation data by adjusting and updating the original indicator matching weight sequence. First, before updating the original indicator matching weight sequence, it is necessary to obtain a historical fault time series data set. In order to perform similarity matching, it is necessary to classify the historical fault time series data set to obtain a training fault time series data set and a test fault time series data set. At this time, the test fault time series data can be extracted in sequence from the test fault time series data set. The embodiment of the present invention performs similarity matching on the test fault time series data and the training fault time series data set by setting an initial matching period and an original indicator matching weight sequence, as follows: according to the prediction The initial matching period and the original indicator matching weight sequence are assumed to calculate the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set to obtain a test timing matching similarity set, and then determine whether there is a target test timing similarity greater than a preset matching judgment threshold in the test timing matching similarity set. If there is no target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, it may be because the initial matching period is too short. Therefore, the initial matching period can be updated and the above-mentioned calculation method based on the preset initial matching period and the original indicator matching weight sequence to calculate the test fault timing data and each training fault timing data in the training fault timing data set is returned. A step of training the test timing matching similarity of fault timing data. If there is a target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, the target test fault type corresponding to the target test timing similarity can be directly identified. Since there are multiple test fault timing data, a target test fault type sequence can be obtained. In order to optimize the original indicator matching weight sequence, the judgment accuracy of the target test fault type sequence can be calculated at this time. According to the judgment accuracy, the original indicator matching weight sequence is adjusted using a pre-constructed particle swarm optimization algorithm to obtain a target indicator matching weight sequence. After the target indicator matching weight sequence is obtained, the current The previous operation data is obtained, and then it is determined whether the current operation data has preset abnormal indicator data. If the current operation data does not have abnormal indicator data, the step of obtaining the current operation data is returned to above. If the current operation data has abnormal indicator data, the current timing matching similarity set can be calculated according to the initial matching period, the current operation data, the historical fault timing data set and the target indicator matching weight sequence. The initial matching period may still be too short, so it is necessary to determine whether there is a target current timing similarity greater than the matching judgment threshold in the current timing matching similarity set. If there is no target current timing similarity greater than the matching judgment threshold in the current timing matching similarity set,Then the initial matching period is updated and the step of calculating the current timing matching similarity set according to the initial matching period, current operation data, historical fault timing data set and the target indicator matching weight sequence is returned. If the current timing matching similarity set contains a target current timing similarity greater than the matching judgment threshold, the target current fault type corresponding to the target current timing similarity can be directly identified, and then a fault warning can be performed according to the target current fault type. Therefore, the present invention can solve the problems of poor monitoring efficiency and low fault identification accuracy in the current IGBT monitoring method. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 A schematic flow chart of an insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization provided by an embodiment of the present invention;

[0091] Figure 2 A schematic structural diagram of an electronic device for implementing the insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization provided by an embodiment of the present invention.

[0092] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0093] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0094] The embodiment of the present application provides an adaptive monitoring method for an insulated gate bipolar transistor based on multi-objective optimization. The execution subject of the adaptive monitoring method for an insulated gate bipolar transistor based on multi-objective optimization includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the adaptive monitoring method for an insulated gate bipolar transistor based on multi-objective optimization can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0095] Embodiment 1:

[0096] Reference Figure 1 FIG. 1 is a flow chart of an insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization provided by an embodiment of the present invention. In this embodiment, the insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization includes:

[0097] S1. Obtain a historical fault time series data set, classify the historical fault time series data set, and obtain a training fault time series data set and a test fault time series data set.

[0098] It can be understood that the historical fault timing data set refers to a historical fault data set of an insulated gate transistor (IGBT), and the historical fault data set refers to a set of monitoring values ​​of various monitoring indicators that change over time when abnormal indicator data occurs. The abnormal indicator data refers to monitoring indicator data that is outside the normal value range, for example: the normal value range of turn-on loss can be (50μJ, 200μJ), and the monitoring indicator refers to an indicator preset by the user to monitor whether an insulated gate transistor has a fault, such as: IGBT junction temperature, gate voltage, collector current, turn-on time, turn-off time, turn-on loss, turn-off loss, etc.

[0099] Furthermore, the training fault timing data set refers to a fault timing data set used for similarity matching of each test fault timing data in the test fault timing data set, and the test fault timing data set refers to a fault timing data set used for similarity matching test, as detailed in the following embodiments. The classification ratio of the training fault timing data set and the test fault timing data set can be 7:3.

[0100] In an embodiment of the present invention, the step of obtaining a historical fault time series data set includes:

[0101] Obtain historical time series data sets and indicator normal interval threshold sets;

[0102] Sequentially extracting historical time series data from the historical time series data set, wherein the historical time series data set refers to a data set of changes in monitoring values ​​of various preset monitoring indicators over time, and the indicator normal interval threshold set refers to a threshold set for determining whether various monitoring indicators are in an abnormal state;

[0103] Determine whether the historical time series data is abnormal historical time series data according to the indicator normal interval threshold set;

[0104] If the historical time series data is not abnormal historical time series data, returning to the above step of sequentially extracting historical time series data from the historical time series data set;

[0105] If the historical time series data is abnormal historical time series data, identifying the historical fault type and abnormal data time series point of the historical time series data according to the indicator normal interval threshold set;

[0106] According to the abnormal data time series point, the historical time series data is segmented into abnormal time series points and the fault type is marked using the historical fault type to obtain the historical fault time series data;

[0107] The historical fault time series data of each abnormal historical time series data are collected to obtain a historical fault time series data set.

[0108] It is understandable that the historical time series data set refers to the historical operation data set of the insulated gate transistor. The normal interval threshold set of indicators refers to the threshold set for judging that each monitoring indicator is in an abnormal state, for example: the normal value range of the turn-on loss (50μJ, 200μJ) corresponds to the normal interval threshold of the indicator of 50μJ and 200μJ. The historical fault type refers to the fault type of the historical time series data, for example: thermal runaway, aluminum metal reconstruction, silicon decomposition, device failure, electrical overstress, open circuit, short circuit, latch effect, gate breakdown, bond wire breakage, poor line contact, manufacturing defects, etc. The abnormal data timing point refers to the time point when the abnormal indicator data appears in the historical time series data. The abnormal timing point segmentation refers to segmenting the data before the abnormal data timing point in the historical time series data and retaining the data after the abnormal data timing point.

[0109] S2. Extracting test fault timing data in sequence from the test fault timing data set.

[0110] S3. Calculate the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set.

[0111] It can be understood that the initial matching period refers to the initial set time segment for calculating the similarity between the test fault timing data and the training fault timing data. For example: when the initial matching period is 0min-10min, the similarity calculation can be performed on the test fault timing data and the data belonging to the first 10min in the training fault timing data. The original indicator matching weight sequence refers to the similarity calculation weights of the various monitoring indicators set by the user in the early stage, for example: the original indicator matching weight sequence corresponding to IGBT junction temperature, gate voltage, collector current, turn-on time, turn-off time, turn-on loss, and turn-off loss can be: 0.1, 0.2, 0.13, 0.27, 0.1, 0.1, 0.1, that is, the original indicator matching weight of IGBT junction temperature is 0.1, the original indicator matching weight of gate voltage is 0.2, and so on. The test timing matching similarity refers to the similarity between the test fault timing data and the training fault timing data. The test timing matching similarity set refers to the test timing matching similarity set of each training fault timing data and the test fault timing data.

[0112] In an embodiment of the present invention, the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set is calculated according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set, including:

[0113] Extracting training fault time series data in sequence from the training fault time series data set;

[0114] Performing time series data interception on the training fault time series data and the test fault time series data according to the initial matching period to obtain training matching time series data and test matching time series data;

[0115] The test time series matching similarity between the training matching time series data and the test matching time series data is calculated using the pre-constructed matching formula and the original indicator matching weight sequence, wherein the matching formula is as follows:

[0116]

[0117] Among them, s T represents the test time series matching similarity between the training matching time series data and the test matching time series data when the initial matching period is T, n represents the number of monitoring indicators, j represents the sequence number of the monitoring indicator, t represents the sequence number of the matching time point in the initial matching period, k j represents the original indicator matching weight of the jth monitoring indicator in the original indicator matching weight sequence, |*| represents the absolute value symbol, It represents the monitoring value of the jth monitoring indicator at the tth matching time point in the training matching time series data. Indicates the monitoring value of the jth monitoring indicator at the tth matching time point in the test matching time series data;

[0118] The test timing matching similarities between each training fault timing data and the test fault timing data are summarized to obtain a test timing matching similarity set.

[0119] It is understandable that the training matching timing data refers to the timing data in the training fault timing data that belongs to the initial matching period, and the test matching timing data refers to the timing data in the test fault timing data that belongs to the initial matching period. The test timing matching similarity refers to the degree of similarity between the training matching timing data and the test matching timing data. The matching time point sequence can be 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, and the initial matching period is 0min-10min. At this time, matching time point sequence 1 corresponds to the 1st minute in the initial matching period, matching time point sequence 2 corresponds to the 2nd minute in the initial matching period, and so on.

[0120] S4. Determine whether there is a target test sequence similarity greater than a preset matching determination threshold in the test sequence matching similarity set.

[0121] It can be understood that the match determination threshold refers to the threshold used to determine the fault type of the test fault timing data, see the following embodiment for details. The target test timing similarity refers to the maximum test timing matching similarity in the test timing matching similarity set that is greater than the match determination threshold. When there is only one test timing matching similarity in the test timing matching similarity set that is greater than the match determination threshold, the target test timing similarity is the test timing matching similarity.

[0122] If the target test sequence matching similarity set does not contain a target test sequence similarity greater than the matching determination threshold, then S5 is executed to update the initial matching period.

[0123] It is understandable that when there is no target test sequence similarity greater than the match determination threshold in the test sequence matching similarity set, it may be caused by the initial matching period being too short, so the initial matching period may be updated.

[0124] In the embodiment of the present invention, the updating of the initial matching period includes:

[0125] The initial matching period is updated using a preset matching period update formula, wherein the matching period update formula is as follows:

[0126] T′=T+Δt

[0127] Wherein, T′ represents the initial matching period after the update, T represents the initial matching period before the update, and Δt represents the update duration.

[0128] Furthermore, the update duration refers to the duration added during the update of the initial matching period, which may be 5 minutes.

[0129] Return to the above step of calculating the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence.

[0130] If there is a target test sequence similarity greater than the matching determination threshold in the test sequence matching similarity set, S6 is executed to identify the target test fault type corresponding to the target test sequence similarity to obtain a target test fault type sequence.

[0131] It is understandable that the target test fault type refers to the labeled fault type of the training fault sequence data corresponding to the target test sequence similarity, for example, latch effect. The target test fault type sequence refers to the set of target test fault types corresponding to each test fault sequence data in the test fault sequence data set.

[0132] S7. Calculate the determination accuracy of the target test fault type sequence, and adjust the original indicator matching weight sequence using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence.

[0133] It should be understood that the judgment accuracy refers to the proportion of fault types judged correctly in the target test fault type sequence. The particle swarm optimization algorithm (PSO for short) is a prior art, see the following embodiment for details. The target index matching weight sequence refers to the index matching weight sequence corresponding to the highest judgment accuracy.

[0134] In the embodiment of the present invention, the step of calculating the determination accuracy of the target test fault type sequence includes:

[0135] Identify the labeled test fault type corresponding to the target test timing similarity to obtain a labeled test fault type sequence;

[0136] Serializing and grouping the labeled test fault type sequence and the target test fault type sequence to obtain a test fault type group set, wherein the test fault type groups in the test fault type group set include labeled test fault types and target test fault types with the same serial number;

[0137] Extract test fault type groups in the test fault type group set in sequence, and determine whether the marked test fault type and the target test fault type in the test fault type group are the same;

[0138] If the marked test fault type and the target test fault type in the test fault type group are the same, the test fault type group is used as the correct judgment type group;

[0139] If the marked test fault type and the target test fault type in the test fault type group are different, the test fault type group is used as the error determination type group;

[0140] Collect all correctly determined type groups and incorrectly determined type groups to obtain correctly determined type group sets and incorrectly determined type group sets;

[0141] The determination accuracy of the target test fault type sequence is calculated using the following formula:

[0142]

[0143] Among them, δ represents the judgment accuracy of the target test fault type sequence, z T represents the number of correctly determined type groups in the correctly determined type group set, z F Indicates the number of error judgment type groups in the error judgment type group set.

[0144] Furthermore, the serialized grouping refers to pairing the labeled test fault type and the target test fault type with the same serial number in the labeled test fault type sequence and the target test fault type sequence.

[0145] In an embodiment of the present invention, the method of adjusting the original indicator matching weight sequence using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence includes:

[0146] The original indicator matching weight sequence is randomly initialized according to a preset population number to obtain multiple groups of initial indicator matching weight sequences, wherein the number of groups of the initial indicator matching weight sequence is equal to the number of populations;

[0147] Identify an initial test fault type sequence of each group of initial indicator matching weight sequences in the multiple groups of initial indicator matching weight sequences according to the training fault timing data set and the test fault timing data set;

[0148] Calculating the initial determination accuracy of the initial test fault type sequence to obtain an initial determination accuracy set;

[0149] Identify a maximum initial accuracy rate in the initial determination accuracy rate set, and identify a maximum indicator matching weight sequence corresponding to the maximum initial accuracy rate;

[0150] Taking the maximum index matching weight sequence as the global optimal weight sequence;

[0151] Randomly initialize the weight update speed sequence of each group of initial indicator matching weight sequence to obtain multiple groups of weight update speed sequences, identify the historical optimal weight sequence of each group of initial indicator matching weight sequence, and obtain multiple groups of historical optimal weight sequences;

[0152] Calculate the speed update sequence of each group of initial indicator matching weight sequence according to the preset inertia weight, acceleration coefficient pair, random number pair, multiple groups of weight update speed sequences, multiple groups of historical optimal weight sequences and the global optimal weight sequence to obtain multiple groups of speed update sequences;

[0153] Calculate the update index matching weight sequence of each group of initial index matching weight sequence according to the multiple groups of speed update sequences and the multiple groups of initial index matching weight sequences to obtain multiple groups of update index matching weight sequences;

[0154] Calculating an update test fault type sequence for each set of update index matching weight sequences in the plurality of sets of update index matching weight sequences;

[0155] Calculating the update determination accuracy of the update test fault type sequence to obtain an update determination accuracy set;

[0156] identifying a maximum update accuracy rate in the update determination accuracy rate set;

[0157] Determining whether the maximum update accuracy is greater than a preset update accuracy threshold;

[0158] If the maximum update accuracy is not greater than the update accuracy threshold, then obtaining the current update number, and determining whether the current update number is greater than a preset update number threshold;

[0159] If the current update number is not greater than the update number threshold, the multiple groups of initial indicator matching weight sequences are updated using the multiple groups of updated indicator matching weight sequences, and the step of returning to the above step of identifying the initial test fault type sequence of each group of initial indicator matching weight sequences in the multiple groups of initial indicator matching weight sequences according to the training fault timing data set and the test fault timing data set is returned;

[0160] If the current update number is greater than the update number threshold, the global optimal weight sequence is used as the target indicator matching weight sequence;

[0161] If the maximum update accuracy is greater than the update accuracy threshold, then a target indicator matching weight sequence corresponding to the maximum update accuracy is identified.

[0162] It can be understood that the population number refers to the multiple of the original indicator matching weight sequence. For example, when the population number is 3 and the original indicator matching weight sequence is 0.1, 0.2, 0.13, 0.27, 0.1, 0.1, 0.1, 3 groups of initial indicator matching weight sequences will be obtained after random initialization, which can be 0.1, 0.1, 0.1, 0.2, 0.3, 0.1, 0.1; 0.1, 0.2, 0.1, 0.2, 0.1, 0.1, 0.2; 0.1, 0.2, 0.3, 0.1, 0.1, 0.1, 0.1.

[0163] In detail, the process of identifying the initial test fault type sequence of each group of initial indicator matching weight sequences according to the training fault timing data set and the test fault timing data set is the same as the process of identifying the target test fault type sequence of the original indicator matching weight sequence, which will not be repeated here. The initial test fault type sequence refers to the set of fault types corresponding to each test fault timing data calculated according to the initial indicator matching weight sequence.

[0164] Furthermore, the initial judgment accuracy refers to the judgment accuracy of the initial test fault type sequence, which is calculated in the same way as the judgment accuracy of the target test fault type sequence, and will not be repeated here. The maximum initial accuracy refers to the maximum initial judgment accuracy in the initial judgment accuracy set. The maximum indicator matching weight sequence refers to the initial indicator matching weight sequence corresponding to the maximum initial accuracy. The global optimal weight sequence refers to the optimal initial indicator matching weight sequence among all initial indicator matching weight sequences (the corresponding initial judgment accuracy is the largest).

[0165] In detail, the weight update speed sequence refers to a sequence composed of the initialization update speed corresponding to each initial indicator matching weight in each group of initial indicator matching weight sequences. The historical optimal weight sequence refers to the best update indicator matching weight sequence in the update process of each group of initial indicator matching weight sequences (its corresponding update judgment accuracy is the largest). The update indicator matching weight sequence refers to the updated initial indicator matching weight sequence, and the update judgment accuracy refers to the judgment accuracy corresponding to the update indicator matching weight sequence (the calculation method is the same as the judgment accuracy of the target test fault type sequence). The inertia weight generally takes a number in the [0,1] interval, which can be 0.5. The acceleration coefficient pair includes a first acceleration coefficient and a second acceleration coefficient, which are usually 2.0. The random number pair includes a first random number and a second random number, which are usually random numbers in the [0,1] interval. The speed update sequence refers to the sequence after the weight update speed sequence is updated. The update indicator matching weight sequence refers to the indicator matching weight sequence obtained after the initial indicator matching weight sequence completes one update. The update test fault type sequence refers to the sequence of fault types after the update indicator matching weight sequence uses the training fault time series data set to predict the fault type of the test fault time series data set. The update judgment accuracy refers to the judgment accuracy of the update test fault type sequence (the calculation method is the same as the judgment accuracy of the target test fault type sequence). The maximum update accuracy refers to the maximum update judgment accuracy in the update judgment accuracy set. The update accuracy threshold refers to the accuracy threshold preset by the user, which can be 98%. The current number of updates refers to the number of times the weight sequence is currently updated according to the initial indicator matching, that is, the number of times multiple sets of update indicator matching weight sequences are obtained. The update number threshold can be set by the user, for example: 100 times.

[0166] In the embodiment of the present invention, the identifying of the historical optimal weight sequence of each group of initial indicator matching weight sequences to obtain multiple groups of historical optimal weight sequences includes:

[0167] Identify multiple groups of historical weight sequences for each group of initial indicator matching weight sequences, wherein the number of groups of the historical weight sequences is equal to the current update number + 1;

[0168] Identify the historical judgment accuracy sets of multiple groups of historical weight sequences;

[0169] Identifying a maximum historical accuracy rate in the historical determination accuracy rate set;

[0170] The historical optimal weight sequence corresponding to the maximum historical accuracy is identified, and the historical optimal weight sequence corresponding to each group of initial indicator matching weight sequences is collected to obtain multiple groups of historical optimal weight sequences.

[0171] It is understandable that, since the historical weight sequence includes the initial indicator matching weight sequence and the updated indicator matching weight sequence, the number of historical weight sequence groups is equal to the current update number + 1. The historical judgment accuracy set refers to the set of judgment accuracy corresponding to each group of historical weight sequences. The maximum historical accuracy refers to the maximum historical judgment accuracy in the historical judgment accuracy set.

[0172] In the embodiment of the present invention, the speed update sequence of each group of initial indicator matching weight sequence is calculated according to the preset inertia weight, acceleration coefficient pair, random number pair, multiple groups of weight update speed sequences, multiple groups of historical optimal weight sequences and the global optimal weight sequence to obtain multiple groups of speed update sequences, including:

[0173] Extracting weight update speed sequences in sequence from the multiple groups of weight update speed sequences, and extracting historical optimal weight sequences in sequence from the multiple groups of historical optimal weight sequences;

[0174] The speed update sequence of each group of initial indicator matching weight sequence is calculated using the pre-built speed update formula according to the inertia weight, acceleration coefficient pair, random number pair, weight update speed sequence, historical optimal weight sequence and the global optimal weight sequence, wherein the speed update formula is as follows:

[0175]

[0176] Among them, V1 represents the first update speed in the speed update sequence, ω represents the inertia weight, v1 represents the first weight update speed in the weight update speed sequence, c1 represents the first acceleration coefficient in the acceleration coefficient pair, and r1 represents the first random number in the random number pair. represents the first historical optimal weight in the historical optimal weight sequence, q1 represents the first initial indicator matching weight in the initial indicator matching weight sequence, c2 represents the second acceleration coefficient in the acceleration coefficient pair, and r2 represents the second random number in the random number pair. represents the first global optimal weight in the global optimal weight sequence, V n represents the nth update speed in the speed update sequence, v n represents the nth weight update speed in the weight update speed sequence, represents the nth historical optimal weight in the historical optimal weight sequence, q n represents the nth initial indicator matching weight in the initial indicator matching weight sequence, represents the nth global optimal weight in the global optimal weight sequence;

[0177] The speed update sequences corresponding to the initial indicator matching weight sequences are collected to obtain multiple groups of speed update sequences.

[0178] For example, when ω=0.5, c1=c2=2, r1=r2=0.3, the weight update speed sequence is (0.05, 0.02, 0.13, 0.17, 0.03, 0.02, 0.11), the historical optimal weight sequence is (0.1, 0.2, 0.3, 0.1, 0.1, 0.1, 0.1), the global optimal weight sequence is (0.1, 0.1, 0.1, 0.2, 0.3, 0.1, 0.1), and the initial indicator matching weight sequence is (0.1, 0.2, 0.1, 0.2, 0.1, 0.1, 0.2), the calculation formula of the speed update sequence is as follows:

[0179]

[0180] In the embodiment of the present invention, the step of calculating the update index matching weight sequence of each group of initial index matching weight sequences according to the multiple groups of speed update sequences and the multiple groups of initial index matching weight sequences to obtain multiple groups of update index matching weight sequences includes:

[0181] Extracting speed update sequences in sequence from the multiple groups of speed update sequences, and extracting initial indicator matching weight sequences in sequence from the multiple groups of initial indicator matching weight sequences;

[0182] The pre-constructed weight update formula is used to calculate the updated index matching weight sequence according to the speed update sequence and the initial index matching weight sequence, wherein the weight update formula is as follows:

[0183]

[0184] Among them, q′1 represents the first update index matching weight in the update index matching weight sequence, q′ n represents the nth update indicator matching weight in the update indicator matching weight sequence;

[0185] The updated indicator matching weight sequences of each group of initial indicator matching weight sequences are collected to obtain multiple groups of updated indicator matching weight sequences.

[0186] S8. Obtain current operation data.

[0187] In detail, the current operating data refers to the operating data of various current monitoring indicators of the insulated gate bipolar transistor.

[0188] S9. Determine whether the current operating data contains preset abnormal indicator data.

[0189] If the current running data does not contain abnormal indicator data, then return to the above step of obtaining the current running data.

[0190] If the current operating data has abnormal indicator data, S10 is executed to calculate the current time series matching similarity set according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence.

[0191] It can be understood that the current time series matching similarity set refers to a matching similarity set between the current operating data and each historical fault time series data in the historical fault time series data set.

[0192] In an embodiment of the present invention, the current time series matching similarity set is calculated according to the initial matching period, the current operation data, the historical fault time series data set and the target indicator matching weight sequence, including:

[0193] Identify the current abnormal data timing point of the current operation data, and segment the current operation data according to the current abnormal data timing point to obtain current fault timing data;

[0194] The current timing matching similarity between the current fault timing data and each historical fault timing data in the historical fault timing data set is calculated according to the initial matching period and the target indicator matching weight sequence to obtain a current timing matching similarity set.

[0195] It is understandable that the current abnormal data time series point refers to the time point when the monitoring value of each monitoring indicator in the current operating data is not within the normal interval threshold of the indicator. The current fault time series data refers to the operating data of the current operating data starting from the current abnormal data time series point. The current time series matching similarity refers to the matching similarity between the current fault time series data and each historical fault time series data calculated based on the initial matching period and the target indicator matching weight sequence.

[0196] S11 , judging whether there is a target current time series similarity greater than a matching determination threshold in the current time series matching similarity set.

[0197] If the current time sequence matching similarity set does not contain a target current time sequence similarity greater than the matching determination threshold, then S12 is executed to update the initial matching period.

[0198] In detail, the updating method of the initial matching period may refer to the matching period updating formula.

[0199] Return to the above step of calculating the current time series matching similarity set based on the initial matching period, current operating data, historical fault time series data set and the target indicator matching weight sequence.

[0200] If there is a target current time sequence similarity greater than the matching determination threshold in the current time sequence matching similarity set, S13 is executed to identify the target current fault type corresponding to the target current time sequence similarity, and a fault warning is performed according to the target current fault type.

[0201] It can be understood that the target current time series similarity refers to the current time series matching similarity greater than the matching judgment threshold in the current time series matching similarity set. When there is a target current time series similarity greater than the matching judgment threshold in the current time series matching similarity set, it indicates that a fault warning can be performed based on the target current fault type corresponding to the target current time series similarity. The target current fault type refers to the fault type prediction result made for the current operating data with abnormal indicator data based on the target indicator matching weight sequence and the historical fault time series data set.

[0202] The present invention solves the problem described in the background technology, and mainly predicts the fault type of the current operation data by adjusting and updating the original indicator matching weight sequence. First, before updating the original indicator matching weight sequence, it is necessary to obtain a historical fault time series data set. In order to perform similarity matching, it is necessary to classify the historical fault time series data set to obtain a training fault time series data set and a test fault time series data set. At this time, the test fault time series data can be extracted in sequence from the test fault time series data set. The embodiment of the present invention performs similarity matching on the test fault time series data and the training fault time series data set by setting an initial matching period and an original indicator matching weight sequence, as follows: according to the prediction The initial matching period and the original indicator matching weight sequence are assumed to calculate the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set to obtain a test timing matching similarity set, and then determine whether there is a target test timing similarity greater than a preset matching judgment threshold in the test timing matching similarity set. If there is no target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, it may be because the initial matching period is too short. Therefore, the initial matching period can be updated and the above-mentioned calculation method based on the preset initial matching period and the original indicator matching weight sequence to calculate the test fault timing data and each training fault timing data in the training fault timing data set is returned. A step of training the test timing matching similarity of fault timing data. If there is a target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, the target test fault type corresponding to the target test timing similarity can be directly identified. Since there are multiple test fault timing data, a target test fault type sequence can be obtained. In order to optimize the original indicator matching weight sequence, the judgment accuracy of the target test fault type sequence can be calculated at this time. According to the judgment accuracy, the original indicator matching weight sequence is adjusted using a pre-constructed particle swarm optimization algorithm to obtain a target indicator matching weight sequence. After the target indicator matching weight sequence is obtained, the current The previous operation data is obtained, and then it is determined whether the current operation data has preset abnormal indicator data. If the current operation data does not have abnormal indicator data, the step of obtaining the current operation data is returned to above. If the current operation data has abnormal indicator data, the current timing matching similarity set can be calculated according to the initial matching period, the current operation data, the historical fault timing data set and the target indicator matching weight sequence. The initial matching period may still be too short, so it is necessary to determine whether there is a target current timing similarity greater than the matching judgment threshold in the current timing matching similarity set. If there is no target current timing similarity greater than the matching judgment threshold in the current timing matching similarity set,Then the initial matching period is updated and the step of calculating the current timing matching similarity set according to the initial matching period, current operation data, historical fault timing data set and the target indicator matching weight sequence is returned. If the current timing matching similarity set contains a target current timing similarity greater than the matching judgment threshold, the target current fault type corresponding to the target current timing similarity can be directly identified, and then a fault warning can be performed according to the target current fault type. Therefore, the present invention can solve the problems of poor monitoring efficiency and low fault identification accuracy in the current IGBT monitoring method.

[0203] Embodiment 2:

[0204] like Figure 2 , is a schematic diagram of the structure of an electronic device for implementing an insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization provided by an embodiment of the present invention.

[0205] The electronic device 1 may include a processor 10, a memory 11 and a bus 12, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as an insulated gate bipolar transistor adaptive monitoring method program based on multi-objective optimization.

[0206] Among them, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 11 also includes an internal storage unit of the electronic device 1 and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the program of the insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization, but also can be used to temporarily store data that has been output or is to be output.

[0207] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips, etc. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules stored in the memory 11 (such as an insulated gate bipolar transistor adaptive monitoring method program based on multi-objective optimization, etc.), and calls data stored in the memory 11 to execute various functions of the electronic device 1 and process data.

[0208] The bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus 12 may be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0209] Figure 2 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 2 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0210] For example, although not shown, the electronic device 1 may also include a power source (such as a battery) for supplying power to each component. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include any components such as one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, etc. The electronic device 1 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0211] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0212] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0213] The insulated gate bipolar transistor adaptive monitoring method program based on multi-objective optimization stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0214] Acquire a historical fault time series data set, classify the historical fault time series data set, and obtain a training fault time series data set and a test fault time series data set;

[0215] Extracting test fault timing data in sequence from the test fault timing data set;

[0216] Calculate the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set;

[0217] Determine whether there is a target test timing similarity greater than a preset matching determination threshold in the test timing matching similarity set;

[0218] If there is no target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, then the initial matching period is updated and the step of calculating the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence is returned to the above step;

[0219] If there is a target test timing similarity greater than a matching determination threshold in the test timing matching similarity set, identifying a target test fault type corresponding to the target test timing similarity to obtain a target test fault type sequence;

[0220] Calculating the determination accuracy of the target test fault type sequence, and adjusting the original indicator matching weight sequence using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence;

[0221] Get current running data;

[0222] Determine whether the current operating data contains preset abnormal indicator data;

[0223] If the current running data does not contain abnormal indicator data, return to the above step of obtaining the current running data;

[0224] If the current operating data has abnormal indicator data, then the current time series matching similarity set is calculated according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence;

[0225] Determine whether there is a target current time series similarity greater than a matching determination threshold in the current time series matching similarity set;

[0226] If there is no target current time series similarity greater than the matching determination threshold in the current time series matching similarity set, then the initial matching period is updated and the step of calculating the current time series matching similarity set according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence is returned to the above step;

[0227] If there is a target current time sequence similarity greater than the matching determination threshold in the current time sequence matching similarity set, the target current fault type corresponding to the target current time sequence similarity is identified, and a fault warning is performed according to the target current fault type.

[0228] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figure 1 to Figure 2 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0229] Furthermore, if the module / unit integrated in the electronic device 1 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0230] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor of an electronic device, the computer program can implement:

[0231] Acquire a historical fault time series data set, classify the historical fault time series data set, and obtain a training fault time series data set and a test fault time series data set;

[0232] Extracting test fault timing data in sequence from the test fault timing data set;

[0233] Calculate the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set;

[0234] Determine whether there is a target test timing similarity greater than a preset matching determination threshold in the test timing matching similarity set;

[0235] If there is no target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, then the initial matching period is updated and the step of calculating the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence is returned to the above step;

[0236] If there is a target test timing similarity greater than a matching determination threshold in the test timing matching similarity set, identifying a target test fault type corresponding to the target test timing similarity to obtain a target test fault type sequence;

[0237] Calculating the determination accuracy of the target test fault type sequence, and adjusting the original indicator matching weight sequence using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence;

[0238] Get current running data;

[0239] Determine whether the current operating data contains preset abnormal indicator data;

[0240] If the current running data does not contain abnormal indicator data, return to the above step of obtaining the current running data;

[0241] If the current operating data has abnormal indicator data, then the current time series matching similarity set is calculated according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence;

[0242] Determine whether there is a target current time series similarity greater than a matching determination threshold in the current time series matching similarity set;

[0243] If there is no target current time series similarity greater than the matching determination threshold in the current time series matching similarity set, then the initial matching period is updated and the step of calculating the current time series matching similarity set according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence is returned to the above step;

[0244] If there is a target current time sequence similarity greater than the matching determination threshold in the current time sequence matching similarity set, the target current fault type corresponding to the target current time sequence similarity is identified, and a fault warning is performed according to the target current fault type.

[0245] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, and actual implementation may have other division methods.

[0246] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0247] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0248] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0249] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. An adaptive monitoring method for insulated gate bipolar transistors based on multi-objective optimization, characterized in that: The method comprises: Acquire a historical fault time series data set, classify the historical fault time series data set, and obtain a training fault time series data set and a test fault time series data set; Extracting test fault timing data in sequence from the test fault timing data set; Calculate the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set; Determine whether there is a target test timing similarity greater than a preset matching determination threshold in the test timing matching similarity set; If there is no target test timing similarity greater than the matching judgment threshold in the test timing matching similarity set, then the initial matching period is updated and the step of calculating the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence is returned to the above step; If there is a target test timing similarity greater than a matching determination threshold in the test timing matching similarity set, identifying a target test fault type corresponding to the target test timing similarity to obtain a target test fault type sequence; Calculating the determination accuracy of the target test fault type sequence, and adjusting the original indicator matching weight sequence using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence; Get current running data; Determine whether the current operating data contains preset abnormal indicator data; If the current running data does not contain abnormal indicator data, return to the above step of obtaining the current running data; If the current operating data has abnormal indicator data, then the current time series matching similarity set is calculated according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence; Determine whether there is a target current time series similarity greater than a matching determination threshold in the current time series matching similarity set; If there is no target current time series similarity greater than the matching determination threshold in the current time series matching similarity set, then the initial matching period is updated and the step of calculating the current time series matching similarity set according to the initial matching period, the current operating data, the historical fault time series data set and the target indicator matching weight sequence is returned to the above step; If there is a target current time sequence similarity greater than the matching determination threshold in the current time sequence matching similarity set, the target current fault type corresponding to the target current time sequence similarity is identified, and a fault warning is performed according to the target current fault type.

2. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 1, characterized in that: The step of obtaining a historical fault time series data set includes: Obtain historical time series data sets and indicator normal interval threshold sets; Sequentially extracting historical time series data from the historical time series data set, wherein the historical time series data set refers to a data set of changes in monitoring values ​​of various preset monitoring indicators over time, and the indicator normal interval threshold set refers to a threshold set for determining whether various monitoring indicators are in an abnormal state; Determine whether the historical time series data is abnormal historical time series data according to the indicator normal interval threshold set; If the historical time series data is not abnormal historical time series data, returning to the above step of sequentially extracting historical time series data from the historical time series data set; If the historical time series data is abnormal historical time series data, identifying the historical fault type and abnormal data time series point of the historical time series data according to the indicator normal interval threshold set; According to the abnormal data time series point, the historical time series data is segmented into abnormal time series points and the fault type is marked using the historical fault type to obtain the historical fault time series data; The historical fault time series data of each abnormal historical time series data are collected to obtain a historical fault time series data set.

3. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 2, characterized in that: The calculating of the test timing matching similarity between the test fault timing data and each training fault timing data in the training fault timing data set according to the preset initial matching period and the original indicator matching weight sequence to obtain a test timing matching similarity set includes: Extracting training fault time series data in sequence from the training fault time series data set; Performing time series data interception on the training fault time series data and the test fault time series data according to the initial matching period to obtain training matching time series data and test matching time series data; The test time series matching similarity between the training matching time series data and the test matching time series data is calculated using the pre-constructed matching formula and the original indicator matching weight sequence, wherein the matching formula is as follows: ； in, Indicates that the initial matching period is , the test time series matching similarity between the training matching time series data and the test matching time series data, Indicates the number of monitoring indicators. Indicates the serial number of the monitoring indicator. Indicates the sequence number of the matching time point within the initial matching period. Indicates the first The original indicator matching weight of the monitoring indicator, represents the absolute value symbol, Indicates the training matching time series data The first monitoring indicator The monitoring values ​​at matching time points, Indicates the test matches the timing data. The first monitoring indicator Monitoring values ​​at matching time points; The test timing matching similarities between each training fault timing data and the test fault timing data are summarized to obtain a test timing matching similarity set.

4. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 3, characterized in that: The updating of the initial matching period includes: The initial matching period is updated using a preset matching period update formula, wherein the matching period update formula is as follows: ； in, represents the updated initial matching period, represents the initial matching period before the update, Indicates the update duration.

5. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 4, characterized in that: The calculating the determination accuracy of the target test fault type sequence includes: Identify the labeled test fault type corresponding to the target test timing similarity to obtain a labeled test fault type sequence; Serializing and grouping the labeled test fault type sequence and the target test fault type sequence to obtain a test fault type group set, wherein the test fault type groups in the test fault type group set include labeled test fault types and target test fault types with the same serial number; Extract test fault type groups in the test fault type group set in sequence, and determine whether the marked test fault type and the target test fault type in the test fault type group are the same; If the marked test fault type and the target test fault type in the test fault type group are the same, the test fault type group is used as the correct judgment type group; If the marked test fault type and the target test fault type in the test fault type group are different, the test fault type group is used as the error determination type group; Collect all correctly determined type groups and incorrectly determined type groups to obtain correctly determined type group sets and incorrectly determined type group sets; The determination accuracy of the target test fault type sequence is calculated using the following formula: ； in, represents the judgment accuracy of the target test fault type sequence, It represents the number of correctly determined type groups in the correctly determined type group set, Indicates the number of error judgment type groups in the error judgment type group set.

6. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 5, characterized in that: The method of adjusting the original indicator matching weight sequence by using a pre-built particle swarm optimization algorithm according to the determination accuracy to obtain a target indicator matching weight sequence includes: The original indicator matching weight sequence is randomly initialized according to a preset population number to obtain multiple groups of initial indicator matching weight sequences, wherein the number of groups of the initial indicator matching weight sequence is equal to the number of populations; Identify an initial test fault type sequence of each group of initial indicator matching weight sequences in the multiple groups of initial indicator matching weight sequences according to the training fault timing data set and the test fault timing data set; Calculating the initial determination accuracy of the initial test fault type sequence to obtain an initial determination accuracy set; Identify a maximum initial accuracy rate in the initial determination accuracy rate set, and identify a maximum indicator matching weight sequence corresponding to the maximum initial accuracy rate; Taking the maximum index matching weight sequence as the global optimal weight sequence; Randomly initialize the weight update speed sequence of each group of initial indicator matching weight sequence to obtain multiple groups of weight update speed sequences, identify the historical optimal weight sequence of each group of initial indicator matching weight sequence, and obtain multiple groups of historical optimal weight sequences; Calculate the speed update sequence of each group of initial indicator matching weight sequence according to the preset inertia weight, acceleration coefficient pair, random number pair, multiple groups of weight update speed sequences, multiple groups of historical optimal weight sequences and the global optimal weight sequence to obtain multiple groups of speed update sequences; Calculate the update index matching weight sequence of each group of initial index matching weight sequence according to the multiple groups of speed update sequences and the multiple groups of initial index matching weight sequences to obtain multiple groups of update index matching weight sequences; Calculating an update test fault type sequence for each set of update index matching weight sequences in the plurality of sets of update index matching weight sequences; Calculating the update determination accuracy of the update test fault type sequence to obtain an update determination accuracy set; identifying a maximum update accuracy rate in the update determination accuracy rate set; Determining whether the maximum update accuracy is greater than a preset update accuracy threshold; If the maximum update accuracy is not greater than the update accuracy threshold, then obtaining the current update number, and determining whether the current update number is greater than a preset update number threshold; If the current update number is not greater than the update number threshold, the multiple groups of initial indicator matching weight sequences are updated using the multiple groups of updated indicator matching weight sequences, and the step of returning to the above step of identifying the initial test fault type sequence of each group of initial indicator matching weight sequences in the multiple groups of initial indicator matching weight sequences according to the training fault timing data set and the test fault timing data set is returned; If the current update number is greater than the update number threshold, the global optimal weight sequence is used as the target indicator matching weight sequence; If the maximum update accuracy is greater than the update accuracy threshold, then a target indicator matching weight sequence corresponding to the maximum update accuracy is identified.

7. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 6, characterized in that: The identifying of the historical optimal weight sequence of each set of initial indicator matching weight sequence, and obtaining multiple sets of historical optimal weight sequences, includes: Identify multiple groups of historical weight sequences for each group of initial indicator matching weight sequences, wherein the number of groups of the historical weight sequences is equal to the current update number + 1; Identify the historical judgment accuracy sets of multiple groups of historical weight sequences; Identifying a maximum historical accuracy rate in the historical determination accuracy rate set; The historical optimal weight sequence corresponding to the maximum historical accuracy is identified, and the historical optimal weight sequence corresponding to each group of initial indicator matching weight sequences is collected to obtain multiple groups of historical optimal weight sequences.

8. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 7, characterized in that: The method calculates the speed update sequence of each group of initial indicator matching weight sequence according to the preset inertia weight, acceleration coefficient pair, random number pair, multiple groups of weight update speed sequences, multiple groups of historical optimal weight sequences and the global optimal weight sequence to obtain multiple groups of speed update sequences, including: Extracting weight update speed sequences in sequence from the multiple groups of weight update speed sequences, and extracting historical optimal weight sequences in sequence from the multiple groups of historical optimal weight sequences; The speed update sequence of each group of initial indicator matching weight sequence is calculated using the pre-built speed update formula according to the inertia weight, acceleration coefficient pair, random number pair, weight update speed sequence, historical optimal weight sequence and the global optimal weight sequence, wherein the speed update formula is as follows: ； in, Indicates the first updated speed in the speed update sequence. represents the inertia weight, represents the first weight update speed in the weight update speed sequence, represents the first acceleration factor in the acceleration factor pair, Represents the first random number in a random number pair, represents the first historical optimal weight in the historical optimal weight sequence, represents the first initial indicator matching weight in the initial indicator matching weight sequence, represents the second acceleration factor in the acceleration factor pair, Represents the second random number in a random number pair, represents the first global optimal weight in the global optimal weight sequence, Indicates the nth update speed in the speed update sequence, represents the nth weight update speed in the weight update speed sequence, represents the nth historical optimal weight in the historical optimal weight sequence, represents the nth initial indicator matching weight in the initial indicator matching weight sequence, represents the nth global optimal weight in the global optimal weight sequence; The speed update sequences corresponding to the groups of initial indicator matching weight sequences are collected to obtain multiple groups of speed update sequences.

9. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 8, characterized in that: The step of calculating the updated index matching weight sequence of each group of initial index matching weight sequences according to the multiple groups of speed update sequences and the multiple groups of initial index matching weight sequences to obtain multiple groups of updated index matching weight sequences includes: Extracting speed update sequences in sequence from the multiple groups of speed update sequences, and extracting initial indicator matching weight sequences in sequence from the multiple groups of initial indicator matching weight sequences; The pre-constructed weight update formula is used to calculate the updated index matching weight sequence according to the speed update sequence and the initial index matching weight sequence, wherein the weight update formula is as follows: ； in, Indicates the first update index matching weight in the update index matching weight sequence, represents the nth update indicator matching weight in the update indicator matching weight sequence; The updated indicator matching weight sequences of each group of initial indicator matching weight sequences are collected to obtain multiple groups of updated indicator matching weight sequences.

10. The insulated gate bipolar transistor adaptive monitoring method based on multi-objective optimization according to claim 9, characterized in that: The calculating of the current time series matching similarity set according to the initial matching period, the current operation data, the historical fault time series data set and the target indicator matching weight sequence includes: Identify the current abnormal data timing point of the current operation data, and segment the current operation data according to the current abnormal data timing point to obtain current fault timing data; The current timing matching similarity between the current fault timing data and each historical fault timing data in the historical fault timing data set is calculated according to the initial matching period and the target indicator matching weight sequence to obtain a current timing matching similarity set.

Images