Distributed new energy power distribution network fault diagnosis system

Abstract

The invention relates to the technical field of distributed new energy power distribution network fault diagnosis, and discloses a distributed new energy power distribution network fault diagnosis system. The distributed new energy power distribution network fault diagnosis system comprises a data integration module, a sorting and classifying module, an information fusion diagnosis module and an intelligent terminal. The system analyzes data distribution features through a waveform data feature extraction module by using a time domain, frequency domain and sample entropy feature extraction method, realizes preliminary screening of a fault diagnosis model, outputs a low-dimensional fault feature set, inputs the low-dimensional fault feature set to a sample training module and a sorting module, realizes training and sorting of fault data, outputs a diagnosis result, and realizes fault diagnosis. And the information fusion diagnosis module is further combined with the protection action condition and the circuit breaker switch displacement information to sort suspicious fault elements for information fusion diagnosis, and outputs a diagnosis result to the intelligent terminal. Automatic fault diagnosis, analysis means and rapid active decision making of the distributed new energy power distribution network are realized, and the adaptability of a fault diagnosis model algorithm to complex characteristics of the power distribution network is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed new energy distribution network fault diagnosis, and in particular to a distributed new energy distribution network fault diagnosis system. Background Art

[0002] In recent years, with the access of a high proportion of new energy and power electronic equipment to the power grid, the randomness, volatility, intermittency and heterogeneous characteristics of power grid faults have become increasingly obvious, posing a comprehensive and systematic test pressure on all links of dispatch decision support. The algorithm iteration and support means of the traditional fault identification model are unsustainable, which can easily affect the efficiency and quality of dispatch decision. First, the information that dispatch decision relies on is characterized by massive, multi-point, large and crowded. The judgment of real power grid faults is highly dependent on the experience and contingency handling ability of the dispatcher. The intelligent and automated development in the field of fault identification is seriously lagging behind. Secondly, the existing distribution network fault identification model algorithm is extremely dependent on the primary system state information and has poor anti-interference performance. In recent years, the development trend of secondary system evidence information has become increasingly similar to that of the primary system, with poor timeliness and independence. The quality level of fault identification directly determines the level of dispatch decision. Finally, the algorithm update and iteration algorithm of the existing fault identification model have significant predictability and periodicity characteristics, which cannot well adapt to the dispatch decision needs of the new power system background with intermittent, fluctuating output and complex, changeable and unstable situations where new energy occupies a dominant position in the power structure, and needs to be optimized urgently.

[0003] The main contradictions in fault identification and fault sorting at the dispatching level of distributed renewable energy distribution networks are concentrated in the dispatching decision support system characterized by big data and horizontal integration. In the past, it has been developing in the aspects of data collection and visualization, ignoring data mining and fault sorting. Too much information has also caused unnecessary interference, and put forward stringent requirements on the fault identification and handling experience of dispatching managers. The technical principle of the existing dispatching decision system is based on the fault identification system based on impedance analysis. The technical means is the unified modeling of fixed value parameters and equipment ledgers. This model architecture with strong expectations cannot cope with the multi-state, heterogeneous and unpredictable fault characteristics of renewable energy as the main body. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies in the prior art, the present invention provides a distributed new energy distribution network fault diagnosis system, which has the advantages of improving the adaptability of the fault diagnosis model algorithm to the complex characteristics of the distribution network, and solves the above-mentioned technical problems.

[0006] (II) Technical solution

[0007] To achieve the above-mentioned object, the present invention provides the following technical solutions: a distributed new energy distribution network fault diagnosis system, comprising a data integration module, a data cleaning module, a data standardization processing module, a waveform data feature extraction module, a sample training module, a sorting and classification module, an information fusion diagnosis module and an intelligent terminal;

[0008] The data integration module obtains the recording master station data and the security master station data to obtain the credibility of the distributed new energy distribution network fault information and perform weighted fusion calculation;

[0009] The data cleaning module cleans invalid and redundant data generated during data generation, data collection and data transmission, and the data standardization processing module processes the data and converts it into standard data;

[0010] The waveform data feature extraction module is used to process the information of changes before, during and after the occurrence of a new energy distribution network fault, and filter out valid data, extract fault data features, and transmit them to the sample training module to generate a fault sample library;

[0011] The sample training module obtains the processed fault data of the recording master station and the security master station to establish a fault diagnosis sample set, and compares and analyzes the fault data features in the fault sample library. After the processing is completed, fault feature sample training is performed and the prediction results are output to the sorting and classification module;

[0012] The sorting and classification module is used to sort the fault data and transmit it to the information fusion diagnosis module;

[0013] The signal fusion diagnosis module realizes the standardization of multi-heterogeneous data and fault identification after filtering, and outputs the fault diagnosis and judgment results after identification to the intelligent terminal;

[0014] The intelligent terminal is used to provide a decision-making plan for fault identification.

[0015] As a preferred technical solution of the present invention, the recording master station data includes the recording morphological characteristics of the fault on this side and the recording morphological characteristics of the fault on the opposite side, and the protection master station data includes the protection setting value, the protection action signal and the switch position signal;

[0016] The specific workflow of the data integration module is as follows:

[0017] A1. Perform a preliminary fault morphology feature judgment and fault credibility assessment on the fault recording morphology features on the local side and the fault recording morphology features on the opposite side in the recording master station data to confirm whether the fault information features on both sides are consistent;

[0018] A2. Obtain protection setting value, protection action signal and switch position signal from the security master station, sort out the suspected fault component set, obtain the fault credibility index for each component in the suspected fault component set one by one, and obtain the fault characterization probability of each component;

[0019] A3. Calculate the fault credibility index E of the distributed new energy distribution network. The specific expression is as follows:

[0020] E=P(E L，t <E0)

[0021]

[0022] Among them, E is the fault credibility index, P(E L，t <E0) means E L，t The probability of occurrence of <E0, E L，t is the fault credibility index of the system fault loss load ratio L at time t, E0 represents the credibility index of the limit state 0 point, Q r，t , Q n，t They represent the post-fault load and normal operating load of the system at time t respectively.

[0023] As the preferred technical solution of the present invention, the data cleaning module performs quality detection on the data output by the data integration module, identifies outlier data, fills in missing data and smoothes noise data, uses swarm intelligence to optimize the association rule algorithm for data cleaning, and calculates the accuracy of the cleaned distributed new energy distribution network fault data. The specific expression is as follows:

[0024]

[0025] Among them, TP means the true value is positive, TN means the true value is negative, M is the total number of sample data, and the accuracy of the cleaned data is The fault data of the distributed new energy distribution grid is output to the data standardization processing module. When the accuracy of the cleaned data is lower than When the ions are detected, they are filtered out after reconfirmation.

[0026] As a preferred technical solution of the present invention, the data standardization processing module includes a data fault-tolerant processing unit and a data standardization conversion unit. The data fault-tolerant processing unit is used to process the situation where there is missing or redundancy in the data, and specifically includes the following steps:

[0027] B1. Scan the fault file to make an initial judgment;

[0028] B2. If any information attribute is missing, query the attribute list, which contains the attribute type and compensation information, and supplement the missing attribute;

[0029] B3. If there are missing data, use cubic spline interpolation or linear averaging to fill in the gaps;

[0030] B4. Delete the redundant data;

[0031] The data standardization conversion unit is used to standardize the data processed by the data fault-tolerant processing unit. The specific calculation expression is as follows:

[0032]

[0033] in, Represents input The z-th real vector function of dimension, C represents the number of fault classes, e represents a natural constant, Indicates the sum of C data. Indicated in The z-th real vector in dimension .

[0034] As the preferred technical solution of the present invention, the waveform data feature extraction module determines the fault type and nature of the fault recording master station data and the security master station data processed by the standardization processing module: respectively extracts the time domain features of the waveform data signals of normal operation, internal faults, and external faults. Frequency domain characteristics And sample entropy feature SE constructs a multi-domain fault feature set of distributed renewable energy distribution network, and performs low-dimensional mapping on the multi-domain fault feature set of distributed renewable energy distribution network to obtain a low-dimensional fault feature set;

[0035] The time domain characteristics Through empirical mode decomposition, the instantaneous frequency and amplitude that change with time are extracted to extract the time domain feature vector composed of ten time domain features, including peak value, peak-to-peak value, root mean square, absolute mean, root square amplitude, variance, skewness, kurtosis, waveform index, and margin index.

[0036] The frequency domain characteristics The three-dimensional frequency domain fault feature vector is formed by extracting three time domain features: mean square frequency, frequency variance and peak frequency.

[0037] The sample entropy feature SE performs VMD decomposition on the fault signal of the distributed renewable energy distribution network, extracts the first four IMF components as effective components, extracts sample entropy and constructs a four-dimensional sample entropy feature vector:

[0038] SE=[SampEn1, SampEn2, SampEn3, SampEn4]

[0039] Among them, SampEn1, SampEn2, SampEn3, and SampEn4 represent the four-dimensional sample entropy.

[0040] As a preferred technical solution of the present invention, the sorting and classification module uses the K-means clustering algorithm for the pre-diagnosis results output by the sample training module to complete the sorting and classification of the distributed new energy distribution network fault data. The specific steps are as follows:

[0041] C1. Randomly select k initial cluster centers from the fault data set for sample training

[0042] C2. Calculate the remaining data objects and cluster centers The Euclidean distance is used to find the cluster center closest to the target data object. At the same time, the waveform data objects are assigned to the cluster centers In the corresponding cluster, the specific calculation expression is as follows:

[0043]

[0044] in, represents μ and The distance between the similarities, For the cluster centers, μ represents data objects, Represents the dimension of the data object, is μ and No. attribute value;

[0045] C3. Calculate the average value of the data objects in each cluster again as the new cluster center and perform the next iteration until the cluster center no longer changes.

[0046] Compared with the prior art, the present invention provides a distributed new energy distribution network fault diagnosis system, which has the following beneficial effects:

[0047] The present invention uses the waveform data feature extraction module to analyze the data distribution characteristics using the time domain, frequency domain, and sample entropy feature extraction methods, realizes the preliminary screening of the fault diagnosis model, outputs a low-dimensional fault feature set, and inputs it into the sample training module and the sorting module to realize the training and sorting of fault data, outputs the diagnosis results, and further combines the protection action situation and the circuit breaker switch displacement information through the information fusion diagnosis module, sorts out the suspected fault components for information fusion diagnosis, and outputs the diagnosis results to the intelligent terminal. The distributed new energy distribution network automation fault diagnosis, analysis means and rapid active decision-making are realized, and the adaptability of the fault diagnosis model algorithm to the complex characteristics of the distribution network is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic diagram of the structure of the present invention;

[0049] Figure 2 This is a schematic diagram of standard data file conversion in the present invention;

[0050] Figure 3 A schematic diagram of multi-domain feature extraction of distributed distribution network faults according to the present invention;

[0051] Figure 4 Schematic diagram of the fault data enhancement process of the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] See also Figure 1-4 , a distributed new energy distribution network fault diagnosis system, including a data integration module, a data cleaning module, a data standardization processing module, a waveform data feature extraction module, a sample training module, a sorting and classification module, an information fusion diagnosis module and an intelligent terminal;

[0054] The data integration module obtains the data from the recording master station and the security master station to obtain the credibility of the fault information of the distributed renewable energy distribution network. The formula for weighted fusion calculation is: F = W·S, where F is the fusion result, W is the weight, and S is the data source. By assigning different weights to different data sources, the data from multiple data sources can be integrated to improve the accuracy and reliability of data fusion.

[0055] The recording master station data includes the recording morphological characteristics of the fault on this side and the fault on the opposite side, and the protection master station data includes the protection setting value, protection action signal and switch position signal;

[0056] The specific workflow of the data integration module is as follows:

[0057] A1. Perform a preliminary fault morphology feature judgment and fault credibility assessment on the fault recording morphology features on the local side and the fault recording morphology features on the opposite side in the recording master station data to confirm whether the fault information features on both sides are consistent;

[0058] A2. Obtain protection setting value, protection action signal and switch position signal from the security master station, sort out the suspected fault component set, obtain the fault credibility index for each component in the suspected fault component set one by one, and obtain the fault characterization probability of each component;

[0059] A3. Calculate the fault credibility index E of the distributed new energy distribution network. The specific expression is as follows:

[0060] E=P(E L，t <E0)

[0061]

[0062] Among them, E is the fault credibility index, P(E L，t <E0) means E L，t The probability of occurrence of <E0, E L，t is the fault credibility index of the system fault loss load ratio L at time t, E0 represents the credibility index of the limit state 0 point, Q r，t , Q n，t They represent the post-fault load and normal operating load of the system at time t respectively. The higher the fault credibility index E, the more realistic the fault waveform data feature extraction is. On the contrary, the lower the fault credibility index E, the more serious the distortion of the fault waveform data feature extraction is. Therefore, weighted fault diagnosis should be performed on each suspected fault component with low credibility index and transmitted to the data cleaning module.

[0063] The data cleaning module cleans the invalid and redundant data generated during data generation, data collection and data transmission, and the data standardization processing module processes the data and converts it into standard data;

[0064] The data cleaning module performs quality inspection on the data output by the data integration module, identifies outliers, fills in missing data, smoothes noise data, eliminates invalid data, and completes valuable data. It uses swarm intelligence to optimize association rule algorithms for data cleaning, and mines the association rules between data fault information feature indicators and distributed new energy distribution network faults, removes invalid and redundant data in the process of data generation, data collection, and data transmission, and calculates the accuracy of the distributed new energy distribution network fault data after cleaning. The specific expression is as follows:

[0065]

[0066] Among them, TP means the true value is positive, TN means the true value is negative, M is the total number of sample data, and the accuracy of the cleaned data is The fault data of the distributed renewable energy distribution network is output to the data standardization processing module. When the accuracy of the cleaned data is lower than When the test result is 0, it is filtered out after reconfirmation;

[0067] The data standardization processing module includes a data fault-tolerant processing unit and a data standardization conversion unit. The data fault-tolerant processing unit is used to process the situation where there is missing or redundancy in the data, and specifically includes the following steps:

[0068] B1. Scan the fault file to make an initial judgment;

[0069] B2. If any information attribute is missing, query the attribute list, which contains the attribute type and compensation information, and supplement the missing attribute;

[0070] B3. If there are missing data, use cubic spline interpolation or linear averaging to fill in the gaps;

[0071] Assume S(x) is a cubic spline interpolation function, x represents the interpolation node, then the data function S(X) must satisfy: in each small interval [xj, xj + 1] is a cubic polynomial, where j = 0, 1, ..., n, x is a given (n+1) interpolation node in a specific data interval; S(x) has a continuous second-order derivative in a certain data interval; S(xj) = yj, j = 0, 1, 2, ..., n, y is the value of the corresponding data function in a specific data interval. By performing cubic spline interpolation on the missing data, the mean error of the interpolation can reach ±1.78mm. For the case where there are fewer missing data, the linear average method is used to fill in the gaps;

[0072] B4. Delete the redundant data;

[0073] The data standardization conversion unit receives the data processed by the data fault-tolerant processing unit, and converts the multi-source data into a unified standard for inconsistent data from multiple manufacturers or systems. Non-numerical information must be digitized and converted into dimensionless pure numerical values ​​to facilitate unified analysis and measurement. For non-standard data, granularity conversion is required, which can be divided into coarser granularity and finer granularity, so that it falls into a small specific interval. Although some details will be discarded, the granularized data is more meaningful and easier to obtain effective features. Granularity conversion is completed for attributes such as current amplitude, voltage amplitude, and time interval using fixed-distance or fixed-ratio classification methods. The standard data file conversion process is shown in the figure;

[0074] The standard spatial data exchange file interacts in real time with the recording master station data and the security master station data processed by the data cleaning module. The identified non-standard or non-standard data needs to be sent to the data cleaning module for processing again. The data standardization method is implemented using the softmax function, which can "compress" the k-dimensional vector z containing any real number into another k-dimensional real vector σ(z). Make each element range between (0, 1), and the sum of all elements is 1;

[0075] The data standardization conversion unit is used to standardize the data processed by the data fault-tolerant processing unit. The specific calculation expression is as follows:

[0076]

[0077] in, Represents input The z-th real vector function of dimension, C represents the number of fault classes, e represents a natural constant, Indicates the sum of C data. Indicated in The zth real vector in dimension;

[0078] The waveform data feature extraction module is used to process the information of changes before, during and after the occurrence of a fault in the new energy distribution network, filter out valid data, extract fault data features, and transmit them to the sample training module to generate a fault sample library;

[0079] The waveform data feature extraction module uses the fault recording master station data and the security master station data processed by the standardization processing module to determine the fault type and nature of the fault. 1000 groups of waveform data signals of normal operation, internal faults, and external faults are taken respectively, and each group of data has 1200 sampling points. The time domain features of the waveform signals under the three states are extracted. Frequency domain characteristics And sample entropy feature SE constructs a multi-domain fault feature set of distributed renewable energy distribution network. The fault feature information mainly includes fault interval, fault time, fault phase, tripping time, coincidence time, second fault time, second tripping time and second fault phase.

[0080] The time domain features are extracted by empirical mode decomposition (EMD) of the instantaneous frequency and amplitude that change with time. The extraction method selects the time domain feature vector composed of ten time domain features: peak value, peak-to-peak value, root mean square, absolute mean, root mean amplitude, variance, skewness, kurtosis, waveform index, and margin index. 1000 groups of data are taken for each operating state of the distributed new energy distribution network system, and a total of 3000 groups of samples for the three states are formed into a dimensioned time domain fault feature set and a dimensionless time domain fault feature set. It is found that in the early stage of the fault, the signal mutation caused by the fault is weak and difficult to identify in a noisy environment. The time-frequency representation of the signal is obtained by using HHT, and the signal energy distribution can be analyzed to extract the time domain characteristics of the fault data;

[0081] Frequency domain feature extraction method: Mean square frequency, frequency variance and peak frequency constitute a three-dimensional frequency domain fault feature vector. 1000 groups of data are taken for each operating state of the distributed new energy distribution network system, and a total of 3000 groups of samples for the three states form a frequency domain feature set. When a fault occurs in the distribution network, there are a large number of high-frequency transient components in the recorded fault signal, and the high-frequency and low-frequency energy ratio of the faulty component is much greater than that of the non-faulty component. The frequency domain is integrated to obtain high-resolution and statistical energy distribution characteristics of the data in the frequency domain, such as frequency variation degree (FVD), energy distortion degree (EDD), and amplitude fault degree (AFD), so as to extract the frequency domain characteristics of the fault data;

[0082] The sample entropy feature extraction method extracts sample entropy by selecting the first four IMF components as effective components to construct a four-dimensional sample entropy feature vector: SE = [SampEn1, SampEn2, SampEn3, SampEn4], where SampEn represents sample entropy, SampEn1, SampEn2, SampEn3, SampEn4 represent four-dimensional sample entropy, and 1000 groups of data are taken for each operating state of the distributed new energy distribution network system, and a total of 3000 groups of samples in three states form an entropy feature set. Obtain indicators that characterize the degree of component failure, thereby extracting the entropy characteristics of fault data.

[0083] The steps of multi-domain fault feature extraction method are as follows:

[0084] Step 1: Extract the time domain and frequency domain characteristics of the distributed renewable energy distribution network operation status signal;

[0085] Step 2: Perform VMD decomposition on the fault signal of the distributed renewable energy distribution network to extract the IMF component sample entropy characteristics, and select effective features based on the correlation coefficient method;

[0086] Step 3: Combine time domain features, frequency domain features and sample entropy features to construct a multi-domain feature set;

[0087] Step 4: Extract and fuse global information of multi-domain fault feature sets and map multi-dimensional features into low-dimensional space;

[0088] Step 5: End the algorithm and output a new feature set, i.e., a low-dimensional fault feature set;

[0089] The sample training module obtains the processed fault data of the recording master station and the security master station to establish a fault diagnosis sample set, and compares and analyzes the fault data features in the fault sample library. After the processing is completed, the fault feature sample training is performed and the prediction results are output to the sorting and classification module;

[0090] The sample training module includes fault diagnosis sample set, fault sample library, KELM model training, output pre-diagnosis results, and receives data from the waveform data feature extraction module. The overlapping sampling data enhancement technology is used for the fault features of the received low-dimensional data. Each sample will have an overlapping part with the previous sample. First, the sample length l is given, and then the moving step length r is set. The overlapping sample length is lr. The smaller r is, the more overlapping parts there are, and the more samples can be collected.

[0091] The sample size is calculated as follows:

[0092]

[0093] Where m represents the number of samples, s represents the length of the fault feature signal of the collected low-dimensional data, l represents the sample length, and r represents the moving step length.

[0094] The fault diagnosis sample set, after the sample entropy feature extraction in the waveform data feature extraction module, found that the sample entropy value of the fault phase is much smaller than the sample entropy value of the non-fault phase, and the fault phase is preliminarily judged. After the VMD decomposition is completed, the sample entropy value fault characteristics of the four inherent modal components (IMF) of each phase voltage are calculated respectively to form the fault diagnosis sample set.

[0095] The fault sample library, through the collected and processed waveform information, screens out the effective data reflecting the fault process by analyzing the data changes before, during and after the fault occurs, and extracts and processes the fault features by the waveform data feature extraction module to generate the fault sample library. The 12 sample entropy values ​​of each group of three-phase fault components extracted by the sample entropy feature in the waveform data feature extraction module are used as fault features, and are input into the KELM model to select the data processed by the waveform data feature extraction module for training. At time t, the fault feature quantities of each data of the recording master station and the security master station constitute a data set, named: X t =[x 1， t ...x i，t ...x n，t ], where n represents the number of sample categories; i represents the fault feature of the i-th data; x i，t It indicates that the value of the fault feature quantity of the i-th data at time t is trained, and the fault type matching is obtained and input into the pre-diagnosis result;

[0096] The sorting and classification module is used to sort the fault data and transmit it to the information fusion diagnosis module;

[0097] The sorting and classification module uses the K-means clustering algorithm to sort and classify the fault data of the distributed new energy distribution network for the pre-diagnosis results output by the sample training module. The specific steps are as follows:

[0098] C1. Randomly select k initial cluster centers from the fault data set for sample training

[0099] C2. Calculate the remaining data objects and cluster centers The Euclidean distance is used to find the cluster center closest to the target data object. At the same time, the waveform data objects are assigned to the cluster centers In the corresponding cluster, the specific calculation expression is as follows:

[0100]

[0101] in, represents μ and The distance between the similarities, For the cluster centers, μ represents the data object, m represents the dimension of the data object, is μ and No. attribute value;

[0102] C3, calculate the average value of the data objects in each cluster again as the new cluster center, and perform the next iteration until the cluster center no longer changes;

[0103] The data after classification diagnosis can be roughly divided into three categories: short circuit fault, complex fault and disturbance data. The short circuit fault and complex fault data are input into the information fusion diagnosis module for processing; the credibility of the faulty component is obtained, the diagnosis result is output, the disturbance data is filed, and the information fusion request of other related data is accepted;

[0104] The signal fusion diagnosis module realizes the standardization of multi-heterogeneous data and fault identification after filtering, and outputs the fault diagnosis and judgment results after identification to the intelligent terminal;

[0105] The signal fusion diagnosis module adopts the intelligent fault-tolerance and lossless compatibility method of recorded data based on the COMTRADE standard to solve the problems of multi-source heterogeneity, inconsistent sampling rates, asynchronous time scales, units and other interferences on the accuracy of information fusion of different system data, standardize multi-source heterogeneous data, output the pre-diagnosis results of distributed distribution network faults and send them to the intelligent terminal. The accuracy of fault identification of the information fusion diagnosis module is greater than 99.5%;

[0106] The intelligent terminal is used to provide a decision-making plan for fault identification;

[0107] The intelligent terminal obtains data processed by the signal fusion diagnosis module, and automatically analyzes new fault characteristics caused by factors such as limited fault current amplitude, phase change, and change in equivalent impedance properties, which are affected by the intermittent, fluctuating output and complex and changeable instability of distributed new energy. It is adapted to the fault characteristic analysis at different stages of the whole process of high-proportion distributed new energy and high-proportion power electronic equipment characteristics, and the time for automatically generating a fault identification decision plan is less than 10 minutes.

[0108] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A distributed new energy distribution network fault diagnosis system, characterized by: It includes data integration module, data cleaning module, data standardization processing module, waveform data feature extraction module, sample training module, sorting and classification module, information fusion diagnosis module and intelligent terminal; The data integration module obtains the recording master station data and the security master station data to obtain the credibility of the distributed new energy distribution network fault information and perform weighted fusion calculation; The data cleaning module cleans invalid and redundant data generated during data generation, data collection and data transmission, and the data standardization processing module processes the data and converts it into standard data; The waveform data feature extraction module is used to process the information of changes before, during and after the occurrence of a new energy distribution network fault, and filter out valid data, extract fault data features, and transmit them to the sample training module to generate a fault sample library; The sample training module obtains the processed fault data of the recording master station and the security master station to establish a fault diagnosis sample set, and compares and analyzes the fault data features in the fault sample library. After the processing is completed, fault feature sample training is performed and the prediction results are output to the sorting and classification module; The sorting and classification module is used to sort the fault data and transmit it to the information fusion diagnosis module; The signal fusion diagnosis module realizes the standardization of multi-heterogeneous data and fault identification after filtering, and outputs the fault diagnosis and judgment results after identification to the intelligent terminal; The intelligent terminal is used to provide a decision-making plan for fault identification.

2. A distributed new energy distribution network fault diagnosis system according to claim 1, characterized in that: The recording master station data includes the recording morphological characteristics of the fault on this side and the recording morphological characteristics of the fault on the opposite side, and the protection master station data includes the protection setting value, the protection action signal and the switch position signal; The specific workflow of the data integration module is as follows: A1. Perform a preliminary fault morphology feature judgment and fault credibility assessment on the fault recording morphology features on the local side and the fault recording morphology features on the opposite side in the recording master station data to confirm whether the fault information features on both sides are consistent; A2. Obtain protection setting value, protection action signal and switch position signal from the security master station, sort out the suspected fault component set, obtain the fault credibility index for each component in the suspected fault component set one by one, and obtain the fault characterization probability of each component; A3. Calculate the fault credibility index E of the distributed new energy distribution network. The specific expression is as follows: E=P(E L，t <E0) Among them, E is the fault credibility index, and P(E L，t <E0) represents the occurrence probability of E L，e < E0, where E L，t is the fault credibility index of the system fault loss load ratio L at time t, E0 represents the credibility index at the limit state 0 point, and Q r，t and Q n，t represent the post-fault load and the normal operating load of the system at time t, respectively.

3. A distributed new energy distribution network fault diagnosis system according to claim 2, characterized in that: The data tendency module performs quality detection on the data output by the data integration module, identifies outlier data, fills in missing data and smoothes noise data, uses swarm intelligence to optimize the association rule algorithm for data cleaning, and calculates the accuracy of the cleaned distributed new energy distribution network fault data. The specific expression is as follows: Among them, TP means the true value is positive, TN means the true value is negative, M is the total number of sample data, and the accuracy of the cleaned data is The fault data of the distributed renewable energy distribution network is output to the data standardization processing module. When the accuracy of the cleaned data is lower than When the ions are detected, they are filtered out after reconfirmation.

4. A distributed new energy distribution network fault diagnosis system according to claim 1, characterized in that: The data standardization processing module includes a data fault-tolerant processing unit and a data standardization conversion unit. The data fault-tolerant processing unit is used to process the situation where there is missing or redundancy in the data, and specifically includes the following steps: B1. Scan the fault file to make an initial judgment; B2. If any information attribute is missing, query the attribute list, which contains the attribute type and compensation information, and supplement the missing attribute; B3. If there are missing data, use cubic spline interpolation or linear averaging to fill in the gaps; B4. Delete the redundant data; The data standardization conversion unit is used to standardize the data processed by the data fault-tolerant processing unit. The specific calculation expression is as follows: in, Represents input The z-th real vector function of dimension, C represents the number of fault classes, e represents a natural constant, Indicates the sum of C data. Indicated in The z-th real vector in dimension .

5. A distributed new energy distribution network fault diagnosis system according to claim 1, characterized in that: The waveform data feature extraction module determines the fault type and nature of the fault recording master station data and the security master station data processed by the standardization processing module: extracts the time domain features of the waveform data signals of normal operation, internal faults, and external faults respectively. Frequency domain characteristics And sample entropy feature SE constructs a multi-domain fault feature set of distributed renewable energy distribution network, and performs low-dimensional mapping on the multi-domain fault feature set of distributed renewable energy distribution network to obtain a low-dimensional fault feature set; The time domain characteristics Through empirical mode decomposition, the instantaneous frequency and amplitude that change with time are extracted to extract the time domain feature vector composed of ten time domain features, including peak value, peak-to-peak value, root mean square, absolute mean, root square amplitude, variance, skewness, kurtosis, waveform index, and margin index. The frequency domain characteristics The three-dimensional frequency domain fault feature vector is formed by extracting three time domain features: mean square frequency, frequency variance and peak frequency. The sample entropy feature SE performs VMD decomposition on the fault signal of the distributed renewable energy distribution network, extracts the first four IMF components as effective components, extracts sample entropy and constructs a four-dimensional sample entropy feature vector: SE=[SampEn1, SampEn2, SampEn3, SampEn4] Among them, SampEn1, SampEn2, SampEn3, and SampEn4 represent the four-dimensional sample entropy.

6. A distributed new energy distribution network fault diagnosis system according to claim 1, characterized in that: The sorting and classification module uses K-means to calculate the pre-diagnosis results output by the sample training module. s Clustering algorithm is used to sort and classify the fault data of distributed new energy distribution network. The specific steps are as follows: C1. Randomly select k initial cluster centers from the fault data set for sample training C2. Calculate the remaining data objects and cluster centers The Euclidean distance is used to find the cluster center closest to the target data object. At the same time, the waveform data objects are assigned to the cluster centers In the corresponding cluster, the specific calculation expression is as follows: in, represents μ and The distance between the similarities, For the cluster centers, μ represents data objects, Represents the dimension of the data object, is μ and No. attribute value; C3. Calculate the average value of the data objects in each cluster again as the new cluster center and perform the next iteration until the cluster center no longer changes.

Images