A method and system for evaluating the health status of a capacitor voltage transformer
By analyzing the secondary output signal of the capacitive voltage transformer SPE statistics and feature vectors, a health status evaluation model was established, which solved the problem that the health status of the capacitive voltage transformer cannot be evaluated online in the existing technology, and real-time evaluation was achieved and differentiated maintenance reduce operating and maintenance costs.
Patent Information
- Application Number
- CN202411356069.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-27
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-09-27
AI Technical Summary
The prior art cannot effectively evaluate the health status level of capacitive voltage transformers online, resulting in over- or insufficient maintenance, and the inability to perform differentiated maintenance for different fault scenarios.
By collecting the secondary output signal of the capacitive voltage transformer, calculating SPE statistics, extracting fault feature vectors, filtering the feature vector set using correlation analysis, establishing a health status evaluation model, training, and using this model to perform online data Process to determine the health status level.
Real-time health status evaluation of capacitive voltage transformers is realized, which can detect abnormal states in a timely manner, reduce maintenance costs, and adapt to the operation and maintenance needs of different fault scenarios.
Smart Images

Figure CN118884333B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of electric power technology, and in particular to a method and system for evaluating the health status of a capacitor voltage transformer. Background Art
[0002] Capacitive Voltage Transformer (CVT) is mainly used for measuring high-voltage signals in power systems. It has the advantages of high electrical strength, high precision, and large insulation margin, and plays an extremely important role in power systems. As a key device for signal acquisition in power systems, CVTs need to provide accurate primary voltage signals to secondary metering, measurement and control, and protection equipment to ensure fair electricity trade settlement and safe and reliable operation of the power system. However, in actual operation, affected by complex environments of varying degrees such as temperature, humidity, and frequency, the internal and external insulation resistance of CVTs and the equivalent parameters of the internal voltage divider capacitors are prone to change, and there are external leakage currents and capacitor breakdowns, which ultimately lead to reduced measurement stability, affecting the accuracy of energy metering and the reliability of the relay protection device. Therefore, it is necessary to conduct a timely and effective assessment of the health status of CVTs.
[0003] The traditional state assessment method is to conduct power outage detection regularly according to the verification procedures of the transformer, but the essence of regular detection is to conduct sampling evaluation of the equipment status in the time domain, which is relatively random and cannot timely evaluate the health status of the capacitive voltage transformer. There are problems of over-maintenance and under-maintenance, and the power outage maintenance operation is complicated, which will consume a lot of manpower and economic costs. Therefore, it is necessary to adopt an online evaluation method to evaluate the health status of the transformer in real time according to the operation data of the transformer. For example, the Chinese invention patent application with publication number CN116577718A discloses a method and system for online monitoring of CVT, which constructs a metering state evaluation model by acquiring the secondary analog voltage data of CVT and determines the metering state evaluation standard; then collects the secondary voltage data of CVT during the network operation; calculates the SPE statistics of the secondary voltage data, and performs online monitoring of the CVT in the network operation based on the metering state evaluation criterion. However, its online detection method only has two results of yes or no, and cannot specifically locate the health status level of CVT, and then make differentiated maintenance work, and has low practicality in dealing with different scene requirements. Summary of the invention
[0004] The technical problem to be solved by the present invention is how to realize online evaluation of the health status level of a capacitor voltage transformer.
[0005] The present invention solves the above technical problems through the following technical means: a method for evaluating the health status of a capacitor voltage transformer, the method comprising:
[0006] S1. Based on the accuracy level of capacitor voltage transformer, define the level of health status of capacitor voltage transformer;
[0007] S2, collect the secondary output signals of the three-phase capacitor voltage transformer in the normal operation state and different fault degree states after commissioning, and calculate the SPE statistics representing the overall health state of the capacitor voltage transformer in different operation states respectively;
[0008] S3, establishing an SPE statistic sequence that characterizes the overall health state of the capacitor voltage transformer under different operating conditions, and extracting a fault feature vector therefrom;
[0009] S4. Analyze the correlation between the feature vectors at different fault levels and their corresponding health status levels by using the correlation analysis method, and select the vector sequence with higher correlation as the feature vector set;
[0010] S5. Based on the feature vector set, a health status assessment model is established and trained;
[0011] S6. Collect online measurement data of the capacitor voltage transformer, extract a feature vector set from it, use the trained health status level assessment model to process the online data, and determine the health status level of the capacitor voltage transformer.
[0012] Furthermore, the capacitor voltage transformer accuracy level is used as the basis for classification, and the levels of the capacitor voltage transformer health status level are defined as follows:
[0013] The measurement error of the capacitor voltage transformer is used as the basis for classifying the fault level. When the measurement error is within ±0.05%, it belongs to the normal fluctuation range, and the health status of the capacitor voltage transformer is judged to be normal operation;
[0014] When the measurement error is between ±0.05% and 0.20%, a detectable minor fault will occur, and the health status of the capacitor voltage transformer is judged to be a minor fault state;
[0015] When the measurement error is between ±0.20% and 0.5%, obvious abnormalities can be detected, and the health status of the capacitive voltage transformer is judged to be a high-risk state;
[0016] When the measurement error is ±0.5% or above, the degree of deviation is too large and the health status of the capacitor voltage transformer is judged to be a serious fault state.
[0017] Furthermore, the secondary output signals of the three-phase capacitor voltage transformer in the normal operating state and in the state of different fault degrees after commissioning are collected, and the SPE statistics characterizing the overall health state of the capacitor voltage transformer in different operating states are calculated respectively, specifically:
[0018] Based on the secondary output signal information, a data matrix of the output signal of the capacitor voltage transformer is constructed. ;
[0019] For the data matrix Perform standardization to obtain a standardized matrix , and then This breaks down into: ,in: is the true value information of the measured data of the capacitor voltage transformer, and E=TePeT is the comprehensive error information of the capacitor voltage transformer;
[0020] Calculate the modulus of the comprehensive error information of the capacitor voltage transformer, recorded as the SPE statistic: .
[0021] Furthermore, the SPE statistic sequence characterizing the overall health state of the capacitor voltage transformer under different operating conditions is established, and the fault feature vector is extracted therefrom, specifically:
[0022] The SPE statistic sequence is decomposed to obtain a low-frequency signal and a high-frequency signal after one layer of decomposition. The nth layer of decomposition obtains 2n nodes, and the decomposition coefficient of each node sub-signal is:
[0023]
[0024]
[0025] Where: , g is the filter function, After decomposition, the node The corresponding coefficients;
[0026] The energy value of the signal in the kth frequency band of the i-th layer is solved, and the corresponding energy is: ,in, is the length of the kth frequency band; Node in the sub-signal The corresponding decomposition coefficients;
[0027] The energy spectrum of all nodes in the i-th layer is used as the fault feature: ;
[0028] The feature vector T is normalized to obtain the fault feature vector, and the sum of the energy of each layer is: , then the normalized eigenvector is: .
[0029] Furthermore, the correlation analysis method is used to analyze the correlation between the feature vectors at different fault levels and their corresponding health status levels, and a vector sequence with a higher correlation is selected as a feature vector set, specifically:
[0030] Add fault level labels to the feature vector sequence and use the level label sequence as the reference sequence ;
[0031] Each of the feature vector sequences is dimensionlessly processed:
[0032]
[0033] In the formula, n is the number of rows of the feature vector sequence, m is the number of columns of the feature vector sequence, = ;
[0034] Calculate the difference between each sequence and the reference sequence The correlation between them is expressed by the correlation coefficient:
[0035]
[0036] In the formula, To compare sequences, , called the resolution coefficient; The smaller it is, the greater the resolution;
[0037] The correlation coefficients of each sequence are concentrated into one value, that is, the average value is calculated as the quantitative representation of the degree of correlation between the comparison sequence and the reference sequence. , ;
[0038] Sort the correlations by size and select the one with the largest correlation. The corresponding sequence as a set of feature vectors.
[0039] Furthermore, the health status level assessment model is established and trained based on the feature vector set, including:
[0040] The feature vector set is fuzzified, and the training fuzzy rule formula is:
[0041]
[0042] In the formula, is the input feature vector set; m is the number of input feature vector sequences, and is the rule parameter;
[0043] Calculate the incentive strength of the rule: ,in represents the membership function of the eigenvector;
[0044] Calculate the rule reasoning results: , y is the output result of the evaluated data;
[0045] Using the proximity principle, the output results are divided into the closest levels and health status labels are added.
[0046] Furthermore, the online measurement data of the capacitor voltage transformer is collected, and a feature vector set is extracted therefrom, and the online data is processed using a trained health status assessment model to determine the health status level of the capacitor voltage transformer, specifically:
[0047] Collect online operation data of capacitor voltage transformer and construct data matrix of secondary output signal amplitude:
[0048]
[0049] In the formula, , , are the secondary output signal amplitudes of the three phases of the capacitor voltage transformer, 1≤i≤t, t is the number of times the output amplitude parameter of the capacitor voltage transformer is sampled;
[0050] Based on the data matrix , calculate the SPE statistic and obtain the statistic sequence , t is the number of times the output amplitude parameter of the capacitive voltage transformer is sampled;
[0051] For the statistical series Decompose to obtain sub-signals, calculate energy values and filter out energy feature vector sets;
[0052] The energy feature set is trained using the trained health status assessment model to calculate the inference result. , and obtain the online evaluation results of the health status of the capacitor voltage transformer.
[0053] Further, the inference result is calculated , the online evaluation results of the health status of the capacitor voltage transformer are obtained, specifically:
[0054] Output value Compared with the grade label, when or When the health status level of the capacitor voltage transformer is determined to be a normal state;
[0055] when When the health status level of the capacitor voltage transformer is determined to be a minor fault;
[0056] when , determining that the health status level of the capacitor voltage transformer is a high risk;
[0057] when , determining that the health status level of the capacitor voltage transformer is a serious fault.
[0058] The present invention also provides a capacitor voltage transformer health status assessment system corresponding to the above assessment method, the system comprising:
[0059] A health level definition module is used to define the level of health status of capacitor voltage transformers based on the accuracy level of capacitor voltage transformers;
[0060] The acquisition module is used to collect the secondary output signal of the three-phase capacitor voltage transformer in the running state after commissioning;
[0061] A feature extraction module, used for extracting a fault feature vector based on the secondary output signal;
[0062] The correlation analysis module is used to analyze the correlation between the feature vectors under different fault levels and their corresponding health status levels, and screen out the vector sequences with higher correlation;
[0063] A training module, used to establish and train a health status assessment model based on a feature vector set;
[0064] The evaluation module is used to process online data and perform online evaluation of the health status level of the capacitor voltage transformer using the trained health status level evaluation model.
[0065] Furthermore, the association analysis module includes:
[0066] The feature vector processing unit is used to perform dimensionless processing on each feature vector sequence: , where n is the number of rows of the feature vector sequence, and m is the number of columns of the feature vector sequence. = ;
[0067] A feature set screening unit, used to screen out vector sequences with high correlation as feature vector sets;
[0068] The step of selecting a vector sequence with high correlation as a feature vector set includes:
[0069] Calculate the difference between each sequence and the reference sequence The correlation between them is expressed by the correlation coefficient: ,in, To compare sequences, , called the resolution coefficient; where The smaller it is, the greater the resolution;
[0070] The correlation coefficients of each sequence are concentrated into one value, that is, the average value is calculated as the quantitative representation of the degree of correlation between the comparison sequence and the reference sequence. , ;
[0071] Sort the correlations by size and select the one with the largest correlation. The corresponding sequence as a set of feature vectors.
[0072] The advantages of the present invention are:
[0073] (1) The present invention analyzes the secondary output data of the capacitor voltage transformer in different health states, divides the health state of the capacitor voltage transformer into different levels, and collects and processes the online operation data to achieve real-time evaluation of the overall operation state of the capacitor voltage transformer. It can help the staff to detect the abnormal state of the transformer in time and perform differentiated operation and maintenance work for the transformers with different fault degrees, which is conducive to reducing the operation and maintenance cost of the transformer and has good economic benefits.
[0074] (2) By analyzing the output data of the capacitor voltage transformer, a SPE (Squared Prediction Error) statistic sequence that represents the health status is constructed. The SPE statistic sequence is processed to obtain a fault feature vector. The fault feature vector is used to identify the fault degree of the online collected data of the capacitor voltage transformer, thereby realizing online evaluation of the health status of the capacitor voltage transformer. BRIEF DESCRIPTION OF THE DRAWINGS
[0075] Figure 1 This is a flow chart of a method for evaluating the health status of a capacitor voltage transformer according to Embodiment 1 of the present invention;
[0076] Figure 2 A schematic diagram of the process of training and evaluating the energy feature set by the health status assessment model in Example 1 of the present invention;
[0077] Figure 3This is a schematic diagram of the health status evaluation result of the health status level in Example 1 of the present invention;
[0078] Figure 4 This is a structural schematic diagram of a capacitor voltage transformer health status assessment system according to Embodiment 2 of the present invention;
[0079] Figure 5 This is a schematic diagram of an application of a capacitor voltage transformer health status assessment system according to Embodiment 2 of the present invention. DETAILED DESCRIPTION
[0080] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described in combination with the embodiments of the present invention. Obviously, the described embodiments are 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.
[0081] Example 1
[0082] like Figure 1 As shown, Embodiment 1 of the present invention proposes a method for evaluating the health status of a capacitor voltage transformer, the method comprising the following steps:
[0083] S1. Based on the accuracy level of capacitor voltage transformer, the health status level of capacitor voltage transformer is defined, including:
[0084] S11. The capacitor voltage transformer measurement error is used as the basis for classifying the fault level. When the measurement error is within ±0.05%, which is within the normal fluctuation range, the capacitor voltage transformer is judged to be in normal operation.
[0085] S12. When the measurement error is between ±0.05% and 0.20%, a detectable minor fault will occur, and the health state of the capacitive voltage transformer at this time is judged to be a minor fault state;
[0086] S13. When the measurement error is between ±0.20% and 0.5%, an obvious abnormality can be detected, and the health status of the capacitive voltage transformer is judged to be a high-risk state;
[0087] S14. When the measurement error is ±0.5% or above, the degree of deviation is too large and the health status of the capacitive voltage transformer is judged to be a serious fault state.
[0088] Specifically, the health status level of the capacitor voltage transformer is shown in the following table:
[0089]
[0090] S2. Collect the secondary output signals of the three-phase capacitor voltage transformer in normal operation and different fault severity after commissioning, and calculate the SPE statistics that characterize the overall health status of the capacitor voltage transformer in different operating states, including:
[0091] S21, constructing a data matrix of the output signal of the capacitor voltage transformer based on the secondary output signal information , n is the number of samples collected, m is the number of output signals collected, for:
[0092]
[0093] Where: Indicates the nth sampling point of the mth output signal collected; n is the number of samples collected, which can be taken as 1000;
[0094] S22, the data matrix Perform standardization to obtain a standardized matrix , and then This breaks down into: ; Among them: T and T e is the standardized matrix The principal component score matrix and residual score matrix are obtained after the singular value decomposition of the covariance matrix, P and is the standardized matrix The principal component loading matrix and residual loading matrix are obtained after the singular value decomposition of the covariance matrix. is the transposed matrix of matrix P, For the matrix The transposed matrix of is the true value information of the measured data of the capacitive voltage transformer, E=T e P e T It is the comprehensive error information of the capacitive voltage transformer.
[0095] For the data matrix The method for standardization is:
[0096]
[0097] In the formula, I n = [1,1,…,1] T , b= ( X 0 ) T / n is the mean vector of the data, Σ=diag ( σ 1 2 ,…, σ m 2 ) is the variance matrix of the data.
[0098] S23, calculate the module value of the comprehensive error information of the capacitor voltage transformer, recorded as SPE statistic: .
[0099] S3. Establish an SPE statistic sequence that characterizes the overall health status of the capacitor voltage transformer under different operating conditions, and extract the fault feature vector from it, including:
[0100] S31, decomposing the SPE statistic sequence, obtaining a low-frequency signal and a high-frequency signal after one layer of decomposition, and the nth layer of decomposition can obtain 2n nodes, and the decomposition coefficient of each node sub-signal is:
[0101]
[0102]
[0103] Where: , is the filter function, After decomposition, the node The corresponding coefficients.
[0104] S32. Solve the energy value of the signal in the kth frequency band of the i-th layer, and the corresponding energy is: ,in: is the length of the kth frequency band; Node in the sub-signal The corresponding decomposition coefficients;
[0105] S33, the energy spectrum of all nodes in the i-th layer is used as the fault feature vector: ;
[0106] S34, normalize the feature vector F to obtain a fault feature vector. Let the sum of the energy of each layer be: , then the normalized eigenvector is: .
[0107] S4. Analyze the correlation between the feature vectors at different fault levels and their corresponding health status levels through correlation analysis, and select vector sequences with higher correlation as feature vector sets, including:
[0108] S41. Add fault level labels to the feature vector sequence and use the level label sequence as a reference sequence ;
[0109] S42, performing dimensionless processing on each of the feature vector sequences:
[0110]
[0111] In the formula, n is the number of rows of the feature vector sequence, m is the number of columns of the feature vector sequence, = ;
[0112] S43, calculate each sequence and reference sequence The correlation between them is expressed by the correlation coefficient:
[0113]
[0114] In the formula, To compare sequences, , called the resolution coefficient, The smaller the resolution, the greater the When the resolution is the best, usually 0.5;
[0115] S44. The correlation coefficients of each sequence are concentrated into one value, that is, the average value is calculated as the quantitative representation of the degree of correlation between the comparison sequence and the reference sequence. , ;
[0116] S45, sorting the correlations by size, and selecting the one with the largest correlation The corresponding sequence as a set of feature vectors.
[0117] S5. Based on the feature vector set, a health status assessment model is established and trained, including:
[0118] S51, fuzzifying the feature vector set, and training the fuzzy rule formula is:
[0119]
[0120] In the formula, is the input feature vector set; m is the number of input feature vector sequences, and is a rule parameter.
[0121] S52. Calculate the incentive intensity of the rule: ,in represents the membership function of the eigenvector;
[0122] S53, calculation rule reasoning result: , y is the output result of the evaluated data;
[0123] It should be noted that since the y value is a comprehensive evaluation result based on the membership function, that is, the output data may be in the interval between the divided categories, the proximity principle is adopted to divide the output results into the closest levels and add health status level labels. For example, when the output y=2.7, the fault level of the capacitive voltage transformer is judged to be 3, which is in a high-risk state.
[0124] S6. Collect online measurement data of the capacitor voltage transformer and extract a feature vector set from it. Use the trained health status level assessment model to process the online data and determine the health status level of the capacitor voltage transformer, such as Figure 2 As shown, specifically including
[0125] S61. Collect online operation data of the capacitor voltage transformer and construct a data matrix of the secondary output signal amplitude:
[0126]
[0127] In the formula, , , are the secondary output signal amplitudes of the three phases of the capacitor voltage transformer, 1≤i≤t, t is the number of times the output amplitude parameter of the capacitor voltage transformer is sampled;
[0128] S62, based on the data matrix , calculate the SPE statistic and obtain the statistic sequence , t is the number of times the output amplitude parameter of the capacitive voltage transformer is sampled.
[0129] S63, the statistical sequence Decompose to obtain sub-signals, calculate the energy value and filter out the energy feature vector set.
[0130] S64. Use the trained health status assessment model to train the energy feature set, calculate the inference result, and obtain an online assessment result of the health status of the capacitor voltage transformer.
[0131] like Figure 3 As shown, the reasoning result is calculated in step S64 to obtain the online evaluation result of the health status of the capacitor voltage transformer, which is specifically:
[0132] Output value Compared with the grade label, when or When the health status level of the capacitor voltage transformer is determined to be a normal state;
[0133] when When the health status level of the capacitor voltage transformer is determined to be a minor fault;
[0134] when , determining that the health status level of the capacitor voltage transformer is a high risk;
[0135] when , determining that the health status level of the capacitor voltage transformer is a serious fault.
[0136] This embodiment aims at the different degrees of fault conditions that may occur during the operation of the capacitor voltage transformer, and realizes the online evaluation of the health status of the capacitor voltage transformer by collecting and analyzing its output data. First, the output data is correlated using the statistical analysis method, and then the characteristic signal that can reflect the fault information is obtained through signal decomposition, and finally the health status of the capacitor voltage transformer is evaluated. This method can not only determine whether the capacitor voltage transformer is faulty, but also diagnose the degree of fault of the capacitor voltage transformer, so that the staff can take different coping methods to perform differentiated operation and maintenance of the transformer, which greatly reduces the waste of manpower, material resources and financial resources.
[0137] In this embodiment, the evaluation of the health status of the capacitor voltage transformer does not need to rely on other sensing equipment. The degree of transformer fault can be determined only by processing and analyzing the secondary output data of the transformer, and the health status of the capacitor voltage transformer can be detected in real time.
[0138] Example 2
[0139] It should be further explained that, based on the same inventive concept of Example 1, Example 2 of the present invention further proposes a capacitor voltage transformer health status assessment system, such as Figure 4 As shown, the system comprises:
[0140] The health level definition module 10 is used to define the level of the health status level of the capacitor voltage transformer based on the accuracy level of the capacitor voltage transformer, specifically including:
[0141] The measurement error of the capacitor voltage transformer is used as the basis for classifying the fault level. When the measurement error is within ±0.05%, it belongs to the normal fluctuation range, and the health status of the capacitor voltage transformer is judged to be normal operation;
[0142] When the measurement error is between ±0.05% and 0.20%, a detectable minor fault will occur, and the health state of the capacitive voltage transformer at this time is judged to be a minor fault state;
[0143] When the measurement error is between ±0.20% and 0.5%, obvious abnormalities can be detected, and the health status of the capacitive voltage transformer is judged to be a high-risk state;
[0144] When the measurement error is ±0.5% or above, the degree of deviation is too large and the health status of the capacitor voltage transformer is judged to be a serious fault state.
[0145] Specifically, the health status level of the capacitor voltage transformer is shown in the following table:
[0146]
[0147] The acquisition module 20 is used to collect and analyze the secondary output signal of the three-phase capacitor voltage transformer during operation after commissioning, including:
[0148] A collection unit, used to collect the secondary output signal of the three-phase capacitive voltage transformer during operation after commissioning;
[0149] The analysis unit is used to analyze the output signal and construct the data matrix of the output signal of the capacitive voltage transformer ;
[0150] Calculation unit, used to calculate the SPE statistic reflecting the degree of error: .
[0151] The feature extraction module 30 is used to extract a fault feature vector based on the secondary output data, including:
[0152] The signal decomposition unit is used to decompose the SPE statistic sequence, and obtain a low-frequency signal and a high-frequency signal after one layer of decomposition. The nth layer of decomposition can obtain 2n nodes, and the decomposition coefficient of each node sub-signal is:
[0153]
[0154]
[0155] Where: , g is the filter function, After decomposition, the node The corresponding coefficients.
[0156] The energy value solving unit is used to solve the energy value of the signal in the kth frequency band of the i-th layer. The corresponding energy is ,in: is the length of the kth frequency band; Node in the sub-signal The corresponding decomposition coefficients;
[0157] Construct an energy spectrum unit to form an energy spectrum of all the energies of the i-th layer nodes as a fault feature: ;
[0158] The normalization unit is used to normalize the feature vector T to obtain the fault feature vector. Let the sum of the energy of each layer be: , then the normalized eigenvector is: ;
[0159] The correlation analysis module 40 is used to analyze the correlation between the feature vectors at different fault levels and their corresponding health status levels, and screen out vector sequences with high correlation, including:
[0160] The feature vector processing unit is used to perform dimensionless processing on each feature vector sequence: , where n is the number of rows of the feature vector sequence, and m is the number of columns of the feature vector sequence. = ;
[0161] The feature set screening unit is used to screen out vector sequences with high correlation as feature vector sets.
[0162] Specifically, the step of selecting a vector sequence with a high correlation as a feature vector set includes:
[0163] Calculate the difference between each sequence and the reference sequence The correlation between them is expressed by the correlation coefficient: ,in, To compare sequences, , called the resolution coefficient. The smaller the value, the greater the resolution. When the resolution is the best, usually 0.5;
[0164] The correlation coefficients of each sequence are concentrated into one value, that is, the average value is calculated as the quantitative representation of the degree of correlation between the comparison sequence and the reference sequence. , ;
[0165] Sort the correlations by size and select the one with the largest correlation. The corresponding sequence as a set of feature vectors.
[0166] The training module 50 is used to establish and train a health status assessment model based on the feature vector set, including:
[0167] The fuzzification unit is used to perform fuzzification processing on the feature vector set. The training rule formula is: ,in, is the input feature vector set; m is the number of input feature vector sequences, and is the rule parameter;
[0168] The incentive unit is used to calculate the incentive strength of the rule: ,in represents the membership function of the eigenvector;
[0169] Output unit, used to calculate rule reasoning results: , y is the output result of the health status level of the evaluated data.
[0170] It should be noted that since the y value is a comprehensive evaluation result based on the membership function, that is, the output data may be in the interval between the divided categories, the proximity principle is adopted to divide the output results into the closest levels and add health status level labels. For example, when the output y=2.7, the fault level of the capacitive voltage transformer is judged to be 3, which is in a high-risk state.
[0171] The evaluation module 60 is used to process the online data and perform online evaluation of the health status level of the capacitor voltage transformer using the trained health status level evaluation model, including:
[0172] The energy feature set is trained using a trained health status assessment model, and an inference result is calculated to obtain an online assessment result of the health status of the capacitor voltage transformer.
[0173] More specifically, the step calculates the inference result to obtain the online evaluation result of the health status of the capacitor voltage transformer, and the process includes:
[0174] Output value Compared with the grade label, when or When the health status level of the capacitor voltage transformer is determined to be a normal state;
[0175] when When the health status level of the capacitor voltage transformer is determined to be a minor fault;
[0176] when , determining that the health status level of the capacitor voltage transformer is a high risk;
[0177] when , determining that the health status level of the capacitor voltage transformer is a serious fault.
[0178] The capacitor voltage transformer health status assessment system proposed in this embodiment has a specific application system structure as shown in the following figure. Figure 5 As shown, by collecting the operating data of the capacitor voltage transformer, evaluating it in the health status assessment system, and feeding back the results to the on-site staff, the health status of the capacitor voltage transformer can be detected in real time, which can ensure timely maintenance when the capacitor voltage transformer has an abnormality, prevent the capacitor voltage transformer from further deterioration, ensure the safe and stable operation of the power system, and have very good economic benefits.
[0179] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for evaluating the health status of a capacitor voltage transformer, characterized in that: The method comprises: S1. Based on the accuracy level of capacitor voltage transformer, define the level of health status of capacitor voltage transformer; S2, collect the secondary output signals of the three-phase capacitor voltage transformer in the normal operation state and different fault degree states after commissioning, and calculate the SPE statistics representing the overall health state of the capacitor voltage transformer in different operation states respectively; S3, establishing an SPE statistic sequence that characterizes the overall health state of the capacitor voltage transformer under different operating conditions, and extracting a fault feature vector therefrom; S4. Analyze the correlation between the feature vectors at different fault levels and their corresponding health status levels by using the correlation analysis method, and select the vector sequence with higher correlation as the feature vector set; S5. Based on the feature vector set, a health status assessment model is established and trained; S6. Collect online measurement data of the capacitor voltage transformer, extract a feature vector set from it, use the trained health status level assessment model to process the online data, and determine the health status level of the capacitor voltage transformer.
2. A method for evaluating the health status of a capacitor voltage transformer according to claim 1, characterized in that: The capacitor voltage transformer accuracy level is used as the classification basis to define the capacitor voltage transformer health status level as follows: The measurement error of the capacitor voltage transformer is used as the basis for classifying the fault level. When the measurement error is within ±0.05%, it belongs to the normal fluctuation range, and the health status of the capacitor voltage transformer is judged to be normal operation; When the measurement error is between ±0.05% and 0.20%, a detectable minor fault will occur, and the health status of the capacitor voltage transformer is judged to be a minor fault state; When the measurement error is between ±0.20% and 0.5%, obvious abnormalities can be detected, and the health status of the capacitive voltage transformer is judged to be a high-risk state; When the measurement error is ±0.5% or above, the degree of deviation is too large and the health status of the capacitor voltage transformer is judged to be a serious fault state.
3. A method for evaluating the health status of a capacitor voltage transformer according to claim 1, characterized in that: The secondary output signals of the three-phase capacitor voltage transformer in the normal operation state and in the state of different fault degrees after commissioning are collected, and the SPE statistics characterizing the overall health state of the capacitor voltage transformer in different operation states are calculated respectively, which are specifically: Based on the secondary output signal information, a data matrix of the output signal of the capacitor voltage transformer is constructed. ; For the data matrix Perform standardization to obtain a standardized matrix , and then This breaks down into: ,in: is the true value information of the measured data of the capacitor voltage transformer, and E=TePeT is the comprehensive error information of the capacitor voltage transformer; Calculate the modulus of the comprehensive error information of the capacitor voltage transformer, recorded as the SPE statistic: .
4. A method for evaluating the health status of a capacitor voltage transformer according to claim 1, characterized in that: The SPE statistic sequence characterizing the overall health state of the capacitor voltage transformer under different operating states is established, and the fault feature vector is extracted therefrom, specifically: The SPE statistic sequence is decomposed to obtain a low-frequency signal and a high-frequency signal after one layer of decomposition. The nth layer of decomposition obtains 2n nodes, and the decomposition coefficient of each node sub-signal is: Where: , g is the filter function, After decomposition, the node The corresponding coefficients; The energy value of the signal in the kth frequency band of the i-th layer is solved, and the corresponding energy is: ,in, is the length of the kth frequency band; Node in the sub-signal The corresponding decomposition coefficients; The energy spectrum of all nodes in the i-th layer is used as the fault feature: ; The feature vector T is normalized to obtain the fault feature vector, and the sum of the energy of each layer is: , then the normalized eigenvector is: .
5. A method for evaluating the health status of a capacitor voltage transformer according to claim 1, characterized in that: The correlation analysis method is used to analyze the correlation between the feature vectors under different fault degrees and their corresponding health status levels, and select the vector sequence with higher correlation as the feature vector set, which is specifically: Add fault level labels to the feature vector sequence and use the level label sequence as the reference sequence ; Each of the feature vector sequences is dimensionlessly processed: In the formula, n is the number of rows of the feature vector sequence, m is the number of columns of the feature vector sequence, = ; Calculate the difference between each sequence and the reference sequence The correlation between them is expressed by the correlation coefficient: In the formula, To compare sequences, , called the resolution coefficient; The smaller it is, the greater the resolution; The correlation coefficients of each sequence are concentrated into one value, that is, the average value is calculated as the quantitative representation of the degree of correlation between the comparison sequence and the reference sequence. , ; Sort the correlations by size and select the one with the largest correlation. The corresponding sequence as a set of feature vectors.
6. A method for evaluating the health status of a capacitor voltage transformer according to claim 1, characterized in that: The health status level assessment model is established and trained based on the feature vector set, including: The feature vector set is fuzzified, and the training fuzzy rule formula is: In the formula, is the input feature vector set; m is the number of input feature vector sequences, and is the rule parameter; Calculate the incentive strength of the rule: ,in represents the membership function of the eigenvector; Calculate the rule reasoning results: , y is the output result of the evaluated data; Using the proximity principle, the output results are divided into the closest levels and health status labels are added.
7. A method for evaluating the health status of a capacitor voltage transformer according to claim 1, characterized in that: The online measurement data of the capacitor voltage transformer is collected, and a feature vector set is extracted therefrom. The online data is processed using a trained health status assessment model to determine the health status level of the capacitor voltage transformer, specifically: Collect online operation data of capacitor voltage transformer and construct data matrix of secondary output signal amplitude: In the formula, , , are the secondary output signal amplitudes of the three phases of the capacitor voltage transformer, 1≤i≤t, t is the number of times the output amplitude parameter of the capacitor voltage transformer is sampled; Based on the data matrix , calculate the SPE statistic and obtain the statistic sequence , t is the number of times the output amplitude parameter of the capacitive voltage transformer is sampled; For the statistical series Decompose to obtain sub-signals, calculate energy values and filter out energy feature vector sets; The energy feature set is trained using the trained health status assessment model to calculate the inference result. , and obtain the online evaluation results of the health status of the capacitor voltage transformer.
8. A method for evaluating the health status of a capacitor voltage transformer according to claim 7, characterized in that: The inference result is calculated , the online evaluation results of the health status of the capacitor voltage transformer are obtained, specifically: Output value Compared with the grade label, when or When the health status level of the capacitor voltage transformer is determined to be a normal state; when When the health status level of the capacitor voltage transformer is determined to be a minor fault; when , determining that the health status level of the capacitor voltage transformer is a high risk; when , determining that the health status level of the capacitor voltage transformer is a serious fault.
9. A capacitor voltage transformer health status assessment system, characterized in that: The system comprises: A health level definition module is used to define the level of health status of capacitor voltage transformers based on the accuracy level of capacitor voltage transformers; The acquisition module is used to collect the secondary output signal of the three-phase capacitor voltage transformer in the running state after commissioning; A feature extraction module, used for extracting a fault feature vector based on the secondary output signal; The correlation analysis module is used to analyze the correlation between the feature vectors under different fault levels and their corresponding health status levels, and screen out the vector sequences with higher correlation; A training module, used to establish and train a health status assessment model based on a feature vector set; The evaluation module is used to process online data and online evaluate the health status level of the capacitor voltage transformer using the trained health status level evaluation model.
10. A capacitor voltage transformer health status assessment system according to claim 9, characterized in that: The association analysis module comprises: The feature vector processing unit is used to perform dimensionless processing on each feature vector sequence: , where n is the number of rows of the feature vector sequence, and m is the number of columns of the feature vector sequence. = ; A feature set screening unit, used to screen out vector sequences with high correlation as feature vector sets; The step of selecting a vector sequence with a high correlation as a feature vector set includes: Calculate the difference between each sequence and the reference sequence The correlation between them is expressed by the correlation coefficient: ,in, To compare sequences, , called the resolution coefficient; where The smaller it is, the greater the resolution; The correlation coefficients of each sequence are concentrated into one value, that is, the average value is calculated as the quantitative representation of the degree of correlation between the comparison sequence and the reference sequence. , ; Sort the correlations by size and select the one with the largest correlation. The corresponding sequence as a set of feature vectors.
