A Fault Diagnosis Method and System for Flexible DC Capacitors

Abstract

The present application relates to the technical field of flexible DC capacitor fault diagnosis, and specifically to a flexible DC capacitor fault diagnosis method and system, the method comprising: collecting working data of the flexible DC capacitor in the current working process; obtaining the fault categories and corresponding working data of the same model of flexible DC capacitor in several historical working processes; clustering the current and all historical collected working data, and correcting the membership of each working data to each cluster cluster by analyzing the characteristics of working data of different fault categories; using the fault category of the cluster cluster with the largest membership in the current working data to diagnose the fault category of the flexible DC capacitor. The present application aims to avoid the problem that the diagnosis result of the flexible DC capacitor fault diagnosis based only on instantaneous parameters may be unreliable, and improve the reliability and credibility of the flexible DC capacitor fault diagnosis.

Description

Technical Field

[0001] The present application relates to the technical field of flexible DC capacitor fault diagnosis, and in particular to a flexible DC capacitor fault diagnosis method and system. Background Art

[0002] The flexible DC support capacitor is one of the key components of the frequency converter, the core device of low-frequency AC power transmission technology. It provides support for DC voltage and absorbs ripple current to achieve stable voltage and current, thereby providing safety protection for the stable operation of semiconductor converters in the frequency converter. Power flexible DC capacitors are commonly used energy storage and filtering components in power systems, and their stable operation plays an important role in ensuring the normal operation of power systems. As a key component in the flexible DC transmission system, the performance and reliability of the flexible DC capacitor directly affect the stability of the entire system.

[0003] Failure of power flexible capacitors may cause power outages in the power system, resulting in economic losses. Through fault diagnosis technology, potential safety hazards can be discovered and dealt with in a timely manner, reducing economic losses and social impacts caused by equipment failures. Most existing flexible capacitor fault diagnosis methods only diagnose the fault type based on relevant data at the time of fault diagnosis, but do not analyze the change trend of relevant parameters before the fault occurs. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a flexible capacitor fault diagnosis method and system, and the technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for diagnosing a fault of a flexible capacitor, the method comprising the following steps:

[0006] Step 1: Collect working data of the flexible DC capacitor in the current working process; the working data includes a parameter sequence of corresponding item parameters composed of the same parameter data; obtain the fault types and corresponding working data of the same model of flexible DC capacitor in several historical working processes;

[0007] Step 2: Cluster the current and all historical collected working data, analyze the characteristics of working data of different fault categories, and modify the membership of each working data to each cluster cluster; specifically:

[0008] S1, taking the time interval between each data and the last data in the parameter sequence as the weight, weighting the difference between each data in the parameter sequence and the mean of all data, and obtaining the sequence stationarity of the parameter sequence;

[0009] S2, using the correlation coefficient between parameter sequences and the difference in sequence stationarity, the correlation between any two parameters in each fault category is calculated;

[0010] S3, combining the difference of the correlation between any two parameters among all fault categories and the difference of the sequence stationarity of the parameter sequences of various parameters among all fault categories, obtains the membership correction value of each working data in each clustering cluster;

[0011] S4, using the membership correction value to sum and correct the membership of each working data in each cluster;

[0012] Step 3: The fault category of the cluster with the largest membership degree in the current working data is used to diagnose the fault category of the flexible DC capacitor;

[0013] The method for obtaining the membership correction value of each working data in each cluster is as follows: the membership correction value of the working data s in the cluster C is recorded as ;

[0014] ; In the formula, exp is an exponential function with the natural constant e as the base, M represents the parameter type, The sequence stationarity of the parameter sequence of parameter a is the absolute value of the difference in identifiability between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s. represents the judgment weight of parameter a in cluster C on faults, Represents the synergy weight of the correlation between parameter c and parameter b in cluster C, The absolute value of the difference between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s, and the synergy of the correlation between parameter c and parameter b;

[0015] Among them, the sequence stationarity of the parameter sequence of parameter a is identifiable in fault category A The expression is: , T represents the total number of fault categories of the flexible capacitor, , They represent the mean values ​​of the sequence stationarity of the parameter sequences of all parameters a in fault category A and fault category B respectively;

[0016] Coordination of the correlation between parameter a and parameter b in fault category A The expression is: , , They represent the average values ​​of the correlations between all parameters a and b belonging to the same fault category A and fault category B, respectively.

[0017] Preferably, the operating data includes at least voltage, current, capacitance and surface temperature.

[0018] Preferably, the fault category includes all fault states and a no-fault state.

[0019] Preferably, the method for obtaining the sequence stationarity of the parameter sequence is:

[0020] Calculate the inverse of the weighted cumulative difference between each data in the parameter sequence and the mean of all data;

[0021] The inverse number is used to determine the sequence stationarity of the parameter sequence, and the inverse number is positively correlated with the sequence stationarity of the parameter sequence.

[0022] Preferably, the calculation method of the correlation between any two parameters in each fault category is:

[0023] Calculate the ratio of the correlation coefficient between the parameter sequences of any two parameters to the difference in the stationarity of the sequences;

[0024] The average of all ratios in the same fault category is taken as the correlation between any two parameters in each fault category.

[0025] Preferably, the initial membership degree is obtained by clustering the current and all historical collected working data in step 2 using an FCM clustering algorithm.

[0026] Preferably, the method for determining the fault category of the cluster is: taking the fault category with the majority number in the cluster as the fault category of the cluster.

[0027] Preferably, the method of using the membership correction value to correct the membership of each working data in each cluster is:

[0028] Sum the initial membership and membership correction value of each working data in each cluster;

[0029] The summation result is normalized to obtain the final corrected membership of each working data in each cluster.

[0030] In a second aspect, an embodiment of the present application further provides a flexible DC capacitor fault diagnosis system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned flexible DC capacitor fault diagnosis methods when executing the computer program.

[0031] This application has at least the following beneficial effects:

[0032] The present application first analyzes the change in the stationarity of each parameter sequence, and weights the data based on the distance between the data and the fault diagnosis, so that the sequence stationarity of the parameter sequence can better reflect the impact on the fault diagnosis; then, by combining the correlation coefficient between the parameter sequences and the difference in sequence stationarity, the correlation between the parameters under each fault category is constructed to evaluate whether the two parameters have a strong correlation effect under different fault categories; combining the difference in the correlation between any two parameters among all fault categories, and the difference in the sequence stationarity of the parameter sequences of various parameters among all fault categories, the membership correction value of each working data in each clustering cluster is obtained to illustrate the possibility that the working data belongs to the clustering cluster, so that the perspective of evaluating the fault category of the working data is more comprehensive; using the membership correction value to correct the membership of each working data in each clustering cluster, it greatly avoids the problem that the fault diagnosis of the flexible DC capacitor based only on instantaneous parameters may lead to unreliable diagnosis results, thereby improving the reliability and credibility of the flexible DC capacitor fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 A flow chart of a flexible capacitor fault diagnosis method provided in this application;

[0035] Figure 2 A flow chart of the process of correcting the degree of membership of each working data to each cluster provided in this application. DETAILED DESCRIPTION

[0036] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the flexible capacitor fault diagnosis method and system proposed in the present application, its specific implementation, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0037] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0038] The following is a detailed description of a flexible capacitor fault diagnosis method and system provided by the present application in conjunction with the accompanying drawings.

[0039] An embodiment of the present application provides a flexible capacitor fault diagnosis method and system.

[0040] Specifically, a flexible capacitor fault diagnosis method is provided as follows, please refer to Figure 1 , the method comprises the following steps:

[0041] Step 1: Collect the working data of the flexible DC capacitor in the current working process; the working data includes a parameter sequence of corresponding item parameters composed of the same parameter data; obtain the fault types and corresponding working data of the same model of flexible DC capacitor in several historical working processes.

[0042] The voltage and current of the flexible capacitor during the current working process are collected using a voltage sensor and a current sensor, and the capacitance of the flexible capacitor is obtained at the same time. The surface temperature of the flexible capacitor during the current working process is collected using a temperature sensor. Among them, the voltage, current, capacitance and surface temperature are all parameters in the working data. In other embodiments of the present application, the working data of other relevant parameters of the flexible capacitor during the working process can also be collected, such as stray inductance, inter-electrode withstand voltage, etc.

[0043] At the same time, historical data is analyzed to obtain the fault categories and corresponding working data of the same model of flexible DC capacitor in several historical working processes. In this embodiment, the total number of fault categories of the flexible DC capacitor is preset to be T, which includes all fault states and no fault state. The fault category of the flexible DC capacitor is determined by professionals in the relevant field, and the specific number of categories is no longer limited.

[0044] Among them, the various parameter data collected in the current working process of the flexible capacitor in this embodiment are the working data within 10 minutes before the current preparation for diagnosis, the frequency of collecting data is one per second, and the same parameter data are arranged in the order of collection time to form a parameter sequence of various parameters; for the collected historical parameter data, among them, for the working data in the fault state, the working data within 10 minutes before the fault occurs are collected; for the working data in the fault-free state, the working data within 10 consecutive minutes during the working process are randomly collected. In other embodiments of the present application, the frequency and time of collecting data can be set by the implementer.

[0045] For failures of flexible DC capacitors, since there may be ambiguity in the definitions between different fault presentation results, it is necessary to combine the signs before the flexible DC capacitor fails, that is, to use the change rules of various parameter data of the flexible DC capacitor to reduce the degree of ambiguity, thereby increasing the reliability of the flexible DC capacitor fault diagnosis results.

[0046] This embodiment collects various parameter data before the current flexible capacitor diagnosis, performs a comprehensive analysis on it and the working data of various fault categories in all the acquired historical data, and obtains the fault diagnosis result by analyzing the change correlation between different parameters. In this embodiment, the fault diagnosis of the flexible capacitor is performed once every minute.

[0047] Step 2: Cluster the current and all historical collected working data, and modify the membership degree of each working data to each cluster cluster by analyzing the characteristics of working data of different fault categories.

[0048] By analyzing the data features of different fault categories, the working data of different fault categories are clustered, thereby increasing the reliability and credibility of the fault category diagnosis results of the current working data.

[0049] This embodiment uses the Euclidean distance between the parameter sequences of the corresponding parameters of the flexible capacitor as the clustering distance of FCM, clusters all the working data, and analyzes the clustering results obtained by using the FCM clustering algorithm, where the K value in the FCM clustering algorithm is T, and the purpose is to divide the clusters into T categories, that is, each cluster corresponds to a fault category. At this time, the initial membership of the working data of the same type of flexible capacitor belonging to different state categories can be obtained. Among them, the FCM clustering algorithm is a well-known technology and will not be repeated.

[0050] Since similar fault categories may lead to similar parameters of the flexible DC capacitor at the time of the fault, resulting in similar membership degrees of multiple fault categories obtained through clustering, the present application further analyzes the faults of the flexible DC capacitor, and corrects the membership degrees of the clustering results according to the characteristics of different fault categories, so as to obtain more accurate distance results.

[0051] The process flow chart for correcting the degree of membership of each working data to each cluster is shown in the attached figure. Figure 2 The specific analysis process is as follows:

[0052] S1, takes the time interval between each data and the last data in the parameter sequence as the weight, weights the difference between each data in the parameter sequence and the mean of all data, and obtains the sequence stationarity of the parameter sequence.

[0053] Firstly, by calculating the similarity between different fault categories, the difficulty of fault differentiation between different fault categories is determined, and then the more difficult fault categories are analyzed to reduce the difficulty of differentiation and increase the reliability of the final diagnosis result of the current working data obtained by using the clustering algorithm.

[0054] Since the precursors of different faults may be different, it is necessary to analyze the changing trends of different parameters and the correlation between different parameters based on the data obtained at adjacent moments, and accurately determine different types of flexible DC capacitor faults based on these changing characteristics.

[0055] This embodiment takes the surface temperature of the flexible capacitor during operation as an example for analysis. First, the stability change of the temperature series is analyzed. Since the parameters collected in this application are data before fault detection, when analyzing the data, the closer the time of data acquisition is to the time of fault detection, the more convincing the data is. Therefore, when analyzing the stability of the temperature series, this application weights the temperature data acquired at different times, and then obtains the final stability analysis result of the temperature series.

[0056] Preferably, the method for obtaining the sequence stationarity of the parameter sequence is: calculating the inverse of the weighted cumulative result of the difference between each data in the parameter sequence and the mean of all data; using the inverse number to determine the sequence stationarity of the parameter sequence, and the inverse number is positively correlated with the sequence stationarity of the parameter sequence.

[0057] It can be understood that a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. This is determined by actual application and is not specifically limited in this application.

[0058] In one implementation of this embodiment, the sequence stationarity of the temperature series is recorded as , ;Wherein, w represents the number of elements in the temperature sequence, represents the time weight of the i-th temperature data in the temperature sequence, Represents the value of the i-th temperature data in the temperature sequence, It indicates the mean temperature in the temperature sequence. That is, when the required distance to the flexible capacitor diagnosis time is closer and the difference between the corresponding temperature value and the obtained temperature mean is smaller, it means that the stability of the temperature sequence is more stable in the precursor of the fault.

[0059] The weighted value is calculated as follows: , ; In the formula, represents the time weight of the i-th temperature data in the temperature sequence, represents the time interval between the last data of the i-th temperature data in the temperature sequence, w represents the number of elements in the temperature sequence, Represents the time distance weight of the i-th temperature data in the temperature sequence. In order to ensure that the sum of the weights is 1, it is normalized to obtain the final time weight That is, the closer the acquisition time of the required temperature data is to the fault diagnosis time, the greater the corresponding time weight.

[0060] According to the above method of calculating the sequence stationarity of the temperature parameter, the sequence stationarity of other parameter sequences is obtained.

[0061] S2, using the correlation coefficient between parameter sequences and the difference in sequence stationarity, calculates the correlation between any two parameters in each fault category.

[0062] Before different faults occur, various parameters will show a certain trend of change, which will affect some parameters accordingly, resulting in the relationship between some parameters being inconsistent with that in normal times. Therefore, during fault diagnosis, it is possible to analyze the correlation between previous parameter data, and then analyze whether the parameter correlation has changed from that in normal times.

[0063] Preferably, the method for calculating the correlation between any two parameters in each fault category is: calculating the ratio of the correlation coefficient between the parameter sequences of any two parameters and the difference in sequence stationarity; and taking the average of all ratios in the same fault category as the correlation between any two parameters in each fault category.

[0064] As an implementation method, the correlation between different parameters is obtained, such as the correlation between parameter a and parameter b. Take this as an example, where the correlation calculation method is as follows: ; In the formula, represents the correlation between parameter a and parameter b, represents the Pearson correlation coefficient between the corresponding sequences of parameter a and parameter b, The absolute value of the difference between the parameter sequences of parameter a and parameter b in terms of sequence stationarity, It is a very small positive number. In this embodiment, the value is 0.01, in order to prevent the denominator from being 0. That is, when the Pearson correlation coefficient between the two parameter sequences is larger and the corresponding difference in the stationarity of the two sequences is smaller, it means that the correlation between the two sequences is stronger.

[0065] The above method is used to obtain the correlation between all acquired pairwise parameter sequences.

[0066] At the same time, the fault categories in the historical data are obtained, and the average value of the correlation between all parameters a and b belonging to the same fault category A is recorded as , which represents the correlation between parameter a and parameter b in fault category A, and is used to evaluate whether there is a strong correlation effect between these two parameters under different fault categories.

[0067] S3, combining the difference in the correlation between any two parameters among all fault categories and the difference in the sequence stationarity of the parameter sequences of various parameters among all fault categories, obtains the membership correction value of each working data in each clustering cluster.

[0068] The present application calculates the change trend of each parameter sequence, calculates the relevant changes of the parameter sequence of the same parameter of the same fault category, and then analyzes the correlation between the relevant change characteristics and the fault category.

[0069] First, we analyze the similarity between the same parameters of different fault categories. That is, the higher the similarity between two parameters of different fault types, the less reference these two parameters have for judging the fault category. The calculation method for obtaining the change correlation between the same parameters of different fault categories is as follows: the synergy of the correlation between parameter a and parameter b in fault category A For example, the calculation formula is:

[0070]

[0071] In the formula, represents the synergy of the correlation between parameter a and parameter b in fault category A, T represents the total number of fault categories of the flexible capacitor, , They represent the average value of the correlation between all parameters a and b belonging to the same fault category A and fault category B. That is, the greater the difference between the correlation between parameters a and b in fault category A and the correlation between parameters a and b in other faults, the more likely it is to judge whether fault category A occurs based on the correlation between parameters a and b.

[0072] Next, the synergy weight of the correlation between parameter a and parameter b in fault category A is calculated as follows:

[0073]

[0074] In the formula, represents the synergy weight of the correlation between parameter a and parameter b in fault category A, represents the synergy of the correlation between parameter a and parameter b in fault category A, M represents the parameter type, It indicates the synergy of the correlation between parameter c and parameter d in fault category A. That is, when the correlation between the required parameter a and parameter b is stronger in fault category A relative to all other fault categories, the corresponding synergy weight of the obtained parameter a and parameter b is greater.

[0075] Similarly, the identifiability of the sequence stationarity of the parameter sequence of each parameter in the fault category is calculated, and the identifiability of the sequence stationarity of the parameter sequence of parameter a in the fault category A is calculated. For example, the calculation expression is:

[0076]

[0077] In the formula, represents the identifiability of the sequence stationarity of the parameter sequence of parameter a in fault category A, T represents the total number of fault categories of the flexible capacitor, , They represent the sequence stationarity means of the parameter sequences of all parameters a in fault category A and fault category B respectively.

[0078] That is, the greater the difference between the sequence stationarity of the parameter sequence of the desired parameter a in fault category A and the sequence stationarity of the same parameter sequence in other fault categories, the more the sequence stationarity of the parameter sequence of parameter a can be used as an indicator for identifying fault A, that is, the identifiability The bigger.

[0079] Similarly, the corresponding weights of the single fault judgment results are obtained, where the judgment weight of parameter a in fault category A for the fault is The method is as follows:

[0080]

[0081] In the formula, It represents the judgment weight of parameter a in fault category A. represents the identifiability of the sequence stationarity of the parameter sequence of parameter a in fault category A, and M represents the parameter type.

[0082] That is, when the fault category A is being sought, the stronger the identifiability of the sequence stationarity of the parameter sequence of parameter a in fault category A is, the more parameter a can be used to judge fault category A.

[0083] Finally, the calculation method for the membership correction value of each working data in each cluster is as follows:

[0084] First, for each cluster, count the fault categories that belong to the majority in all working data as the fault category of the cluster. Then, based on the fault categories corresponding to different clusters, analyze the membership correction value of each working data in each cluster, and use the membership correction value of working data s in cluster C as the membership correction value. For example, the calculation expression is:

[0085]

[0086] In the formula, It represents the membership correction value of the working data s in the cluster C, exp is an exponential function with the natural constant e as the base, M represents the parameter type, The sequence stationarity of the parameter sequence of parameter a is the absolute value of the difference in identifiability between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s. represents the judgment weight of parameter a in cluster C on faults, Represents the synergy weight of the correlation between parameter c and parameter b in cluster C, It represents the absolute value of the difference between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s, and the synergy of the correlation between parameter c and parameter b.

[0087] That is, when the sequence stationarity of the parameter sequence of the parameter a is sought, the smaller the difference in identifiability between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s, and the smaller the difference in the synergy of the correlation between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s, the smaller the difference in the synergy of the correlation between parameter c and parameter b, the stronger the possibility that the working data s belongs to cluster C, and the larger the corresponding membership correction value of the working data s belonging to cluster C. The membership correction value is used to illustrate the possibility of the working data belonging to the cluster, so that the perspective of evaluating the fault category of the working data is more comprehensive.

[0088] S4, using the membership correction value to sum and correct the membership of each working data in each cluster.

[0089] Preferably, the method of using the membership correction value to sum and correct the membership of each working data in each cluster cluster is: summing the initial membership of each working data in each cluster cluster with the membership correction value; normalizing the sum result to obtain the final corrected membership of each working data in each cluster cluster.

[0090] The final modified membership of the working data s in the cluster C For example, the expression for calculation is:

[0091]

[0092]

[0093] In the formula, It represents the modified membership of the working data s in the cluster C. It represents the initial membership of the working data s in the cluster C when clustering using the FCM algorithm. It represents the membership correction value of the working data s in the cluster C. At the same time, in order to ensure that the sum of the membership of each working data in all clusters is 1, it is normalized to obtain the final corrected membership , represents the membership of the working data s in the cluster C after final correction, T represents the total number of fault categories of the flexible capacitor, Represents the corrected membership of the working data s in the cluster D.

[0094] The membership correction value is used to correct the membership of each working data in each cluster, so as to obtain the final corrected membership. This greatly avoids the problem that the fault diagnosis of the flexible DC capacitor based only on instantaneous parameters may lead to unreliable diagnosis results, and improves the reliability and credibility of the flexible DC capacitor fault diagnosis.

[0095] Step 3: The fault category of the cluster with the largest membership degree in the current working data is used to diagnose the fault category of the flexible DC capacitor.

[0096] The current working data and all the working data collected in the past are clustered and analyzed according to the above method, and then the membership degree of each working data in each cluster is corrected.

[0097] The corrected membership degree can be used to better evaluate the fault category of each working data, making the fuzzy boundary of data analysis due to different types of faults clearer, so as to more accurately judge the fault category of the flexible capacitor in the current working data. Specifically, the fault category of the cluster with the largest membership degree in the current working data is used as the fault category of the current flexible capacitor, thereby completing the fault diagnosis of the flexible capacitor.

[0098] Based on the same inventive concept as the above method, an embodiment of the present application also provides a flexible DC capacitor fault diagnosis system, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of a flexible DC capacitor fault diagnosis method described in any one of the above methods are implemented.

[0099] The various embodiments in the present application are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0100] It should be noted that, unless otherwise specified and limited, terms such as "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such article or device. In the absence of further restrictions, an element defined by the sentence "including one..." does not exclude the existence of other identical elements in the article or device including the element. In addition, the term "and\or" used herein includes any and all combinations of one or more related listed items.

[0101] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not invented by the present application.

[0102] It should be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. A method for diagnosing faults of flexible capacitors, characterized in that: The method comprises the following steps: Step 1: Collect working data of the flexible DC capacitor in the current working process; the working data includes a parameter sequence of corresponding item parameters composed of the same parameter data; obtain the fault types and corresponding working data of the same model of flexible DC capacitor in several historical working processes; Step 2: Cluster the current and all historical collected working data, analyze the characteristics of working data of different fault categories, and modify the membership of each working data to each cluster cluster; specifically: S1, taking the time interval between each data and the last data in the parameter sequence as the weight, weighting the difference between each data in the parameter sequence and the mean of all data, and obtaining the sequence stationarity of the parameter sequence; S2, using the correlation coefficient between parameter sequences and the difference in sequence stationarity, the correlation between any two parameters in each fault category is calculated; S3, combining the difference of the correlation between any two parameters among all fault categories and the difference of the sequence stationarity of the parameter sequences of various parameters among all fault categories, obtains the membership correction value of each working data in each clustering cluster; S4, using the membership correction value to sum and correct the membership of each working data in each cluster; Step 3: The fault category of the cluster with the largest membership degree in the current working data is used to diagnose the fault category of the flexible DC capacitor; The method for obtaining the membership correction value of each working data in each cluster is as follows: the membership correction value of the working data s in the cluster C is recorded as ; ; In the formula, exp is an exponential function with the natural constant e as the base, M represents the parameter type, The sequence stationarity of the parameter sequence of parameter a is the absolute value of the difference in identifiability between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s. represents the judgment weight of parameter a in cluster C on faults, Represents the synergy weight of the correlation between parameter c and parameter b in cluster C, The absolute value of the difference between the fault category of cluster C and the fault category of the cluster corresponding to the maximum initial membership of the working data s, and the synergy of the correlation between parameter c and parameter b; Among them, the sequence stationarity of the parameter sequence of parameter a is identifiable in fault category A The expression is: , T represents the total number of fault categories of the flexible capacitor, , They represent the mean values ​​of the sequence stationarity of the parameter sequences of all parameters a in fault category A and fault category B respectively; Coordination of the correlation between parameter a and parameter b in fault category A The expression is: , , They represent the average values ​​of the correlations between all parameters a and b belonging to the same fault category A and fault category B, respectively.

2. A method for diagnosing a fault of a flexible capacitor as claimed in claim 1, characterized in that: The operating data includes at least voltage, current, capacitance and surface temperature.

3. A method for diagnosing a fault of a flexible capacitor as claimed in claim 1, characterized in that: The fault category includes all fault states and no-fault states.

4. A method for diagnosing a fault of a flexible capacitor as claimed in claim 1, characterized in that: The method for obtaining the sequence stationarity of the parameter sequence is: Calculate the inverse of the weighted cumulative difference between each data in the parameter sequence and the mean of all data; The inverse number is used to determine the sequence stationarity of the parameter sequence, and the inverse number is positively correlated with the sequence stationarity of the parameter sequence.

5. A method for diagnosing a fault of a flexible capacitor as claimed in claim 1, characterized in that: The calculation method of the correlation between any two parameters in each fault category is: Calculate the ratio of the correlation coefficient between the parameter sequences of any two parameters to the difference in the stationarity of the sequences; The average of all ratios in the same fault category is taken as the correlation between any two parameters in each fault category.

6. A method for diagnosing a fault of a flexible capacitor as claimed in claim 1, characterized in that: The initial membership degree is obtained by clustering the current and all historical collected working data using the FCM clustering algorithm in step 2.

7. A method for diagnosing a fault of a flexible capacitor as claimed in claim 1, characterized in that: The method for determining the fault category of the cluster is: taking the fault category with the majority in the cluster as the fault category of the cluster.

8. A method for diagnosing a fault of a flexible capacitor as claimed in claim 6, characterized in that: The method of using the membership correction value to correct the membership of each working data in each cluster is as follows: Sum the initial membership and membership correction value of each working data in each cluster; The summation result is normalized to obtain the final corrected membership of each working data in each cluster.

9. A flexible capacitor fault diagnosis system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of a flexible capacitor fault diagnosis method as described in any one of claims 1-8 are implemented.

Images