CVT error state prediction method and device based on incremental ensemble learning model

Abstract

The present invention discloses a CVT error state prediction method and device based on an incremental ensemble learning model, the method comprising the steps of: dividing the CVT historical data of power outage verification into a number of data blocks, generating a corresponding number of base models according to the data blocks, and fusing the base models into a reference state prediction model; acquiring the first CVT real-time data, and detecting whether concept drift occurs between the first CVT real-time data and the CVT historical data; when concept drift occurs, acquiring the incremental data of the first CVT real-time data relative to the CVT historical data, and generating an incremental base model according to the incremental data; generating an adaptive incremental ensemble learning model according to the incremental base model and the reference state prediction model, acquiring the second CVT real-time data, and performing error state prediction on the second CVT real-time data according to the adaptive incremental ensemble learning model. The present invention improves the accuracy of CVT error state prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of CVT error state prediction, and in particular to a CVT error state prediction method and device based on an incremental ensemble learning model. Background Art

[0002] The voltage transformer is an important measuring device in the power system. Its primary winding is connected to the high-voltage power grid, and the secondary winding is connected to the measurement, metering, protection and other devices. It is used to convert the high-voltage signal on the primary side into a low-voltage signal for use by the secondary equipment.

[0003] Long-term operating experience shows that due to the increase in the service life of voltage transformers, there is a certain risk of out-of-tolerance for voltage transformers after several years of operation.

[0004] The continued operation of out-of-tolerance voltage transformers will bring huge losses to the trade settlement of the three parties of power generation, supply and use, and even affect the stable operation of the power system. Therefore, in order to ensure the accuracy of measurement and the safe operation of the power system, it is necessary to timely evaluate and replace voltage transformers with abnormal error states. There are mature offline evaluation methods for periodic offline evaluation of voltage transformers, but it is difficult to perform non-fault power outages on high-voltage transmission networks, making it difficult for this method to cover all voltage transformers to be tested within the specified period; and there is a difference between the environmental electromagnetic field during offline evaluation and that during online operation, resulting in a certain deviation between the evaluation results and the actual situation, which further leads to a large number of voltage transformers in operation in substations being overdue for inspection and unknown errors.

[0005] In order to solve the shortcomings of the periodic offline evaluation method, the existing technology adopts an online evaluation method under non-stop power supply conditions to realize real-time online monitoring of the error state of the voltage transformer. The existing online evaluation technology is based on the signals collected by each device in the power system and is analyzed and processed based on the principle of data-driven, so as to evaluate the error state of the voltage transformer, that is, by constructing an approximate model with the help of historical data, real-time data and relational data, and relying on a large amount of data and calculations to characterize the real error state of the voltage transformer in real time. However, the existing technology has the following shortcomings: In terms of data, the existing technology does not consider the concept drift in the data set. Concept drift refers to the change of the concepts contained in the data set, such as equipment aging, sudden changes in operating conditions, etc., which makes the concepts contained in the new and old data no longer consistent. Once the concept drift occurs in the data set, it will affect the accuracy of characterizing the real error state of the voltage transformer based on the data-driven principle. Summary of the invention

[0006] The present invention provides a CVT error state prediction method and device based on an incremental ensemble learning model, which improves the accuracy of CVT error state prediction.

[0007] An embodiment of the present invention provides a CVT error state prediction method based on an incremental ensemble learning model, comprising the following steps:

[0008] Divide the CVT historical data of power outage inspection into a number of data blocks, generate a corresponding number of base models according to the data blocks, and fuse the base models into a reference state prediction model;

[0009] Acquire first CVT real-time data, and detect whether concept drift occurs between the first CVT real-time data and CVT historical data;

[0010] When concept drift occurs, obtaining incremental data of the first CVT real-time data relative to the CVT historical data, and generating an incremental base model according to the incremental data;

[0011] An adaptive incremental integrated learning model is generated according to the incremental base model and the reference state prediction model, second CVT real-time data is acquired, and error state prediction is performed on the second CVT real-time data according to the adaptive incremental integrated learning model.

[0012] Further, generating a corresponding number of base models according to the plurality of data blocks and fusing the base models into a reference state prediction model comprises the following steps:

[0013] The k data blocks are correspondingly formed into k first base models, and then the k first base models are updated into corresponding k second base models by a cross-validation method; wherein k is a positive integer greater than 3;

[0014] The k second base models are fused into a reference state prediction model.

[0015] Furthermore, for any of the first base models, the first base model is generated according to one data block among the k data blocks, and the first base model is cross-validated using the remaining k-1 data blocks except the one data block to obtain the corresponding second base model.

[0016] Furthermore, after replacing the base model with the worst classification effect among the k second base models with the incremental base model, the incremental base model and the remaining k-1 second base models are fused to obtain an adaptive incremental ensemble learning model.

[0017] Furthermore, after replacing the base model with the worst classification effect among the k first base models with the incremental base model, the remaining k-1 first base models and the incremental base model are updated into corresponding k second base models by using a cross-validation method;

[0018] The k second base models are fused into a reference state prediction model.

[0019] Furthermore, the incremental base model and the reference state prediction model are fused to obtain an adaptive incremental ensemble learning model.

[0020] Further, calculating the KL divergence and drift threshold according to the first CVT real-time data and CVT historical data;

[0021] Whether concept drift occurs is determined according to the KL divergence and the drift threshold.

[0022] Furthermore, mean shift clustering is performed on the entire CVT data set formed by the CVT historical data and the first CVT real-time data, and the drift threshold is calculated according to the formula R=D-2r, where D is the average distance between cluster centers and r is the average radius of the cluster.

[0023] Furthermore, an adversarial neural network algorithm is used to oversample the minority class samples in the incremental data, and the incremental base model is generated according to the incremental data after the oversampling process.

[0024] Another embodiment of the present invention provides a CVT error state prediction device based on an incremental ensemble learning model, comprising a base model generation module, a concept drift detection module, an incremental base model generation module, and an error state prediction module;

[0025] The base model generation module is used to divide the CVT historical data of power outage inspection into several data blocks, generate a corresponding number of base models according to the several data blocks, and fuse the base models into a reference state prediction model;

[0026] The concept drift detection module is used to obtain first CVT real-time data and detect whether concept drift occurs between the first CVT real-time data and CVT historical data;

[0027] The incremental base model generation module is used for acquiring incremental data of the first CVT real-time data relative to the CVT historical data when concept drift occurs, and generating an incremental base model according to the incremental data;

[0028] The error state prediction module is used to generate an adaptive incremental integrated learning model according to the incremental base model and the reference state prediction model, obtain the second CVT real-time data, and perform error state prediction on the second CVT real-time data according to the adaptive incremental integrated learning model.

[0029] The embodiments of the present invention have the following beneficial effects:

[0030] The present invention provides a CVT error state prediction method and device based on an incremental ensemble learning model. The method detects the concept drift phenomenon that may occur in CVT historical data and CVT real-time data, and according to the concept drift detection result, adopts an adaptive incremental ensemble classification model to cope with dynamic data streams, performs adaptive error state evaluation, improves the accuracy of the error state evaluation model and the adaptability range of the model, and thus improves the accuracy of the CVT error state evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a flow chart of a CVT error state prediction method based on an incremental ensemble learning model provided by an embodiment of the present invention;

[0032] Figure 2 It is a structural schematic diagram of a CVT error state prediction device based on an incremental ensemble learning model provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0033] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings 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.

[0034] like Figure 1 As shown, a CVT error state prediction method based on an incremental ensemble learning model provided by an embodiment of the present invention includes the following steps:

[0035] Step S101: divide the CVT historical data of power outage inspection into a number of data blocks, generate a corresponding number of base models according to the data blocks, and fuse the base models into a reference state prediction model.

[0036] As one embodiment, generating a corresponding number of base models according to the plurality of data blocks and fusing the base models into a reference state prediction model includes the following steps:

[0037] Step S11: k data blocks are correspondingly formed into k first base models, and then the k first base models are updated into corresponding k second base models by cross-validation method; wherein k is a positive integer greater than 3. Specifically, the k first base models are cross-validated K times to generate k second base models, and for any of the first base models, the first base model is generated according to one data block among the k data blocks, and the remaining k-1 data blocks except the one data block are used to cross-validate the first base model to obtain the corresponding second base model.

[0038] Step S12: fusing the k second base models into a reference state prediction model.

[0039] As one embodiment, the CVT historical data of power failure detection is divided into k data blocks D = {D1, D2, D3, ..., D k}, where k = 1, 2, …, a; a ≥ 1.

[0040] D t Represents the window data at historical time t, and defines each window data:

[0041]

[0042] The underlying data distribution is:

[0043]

[0044] Where n represents the number of samples, represents the feature vector, The feature vectors of CVT selected by the present invention are the inter-group amplitude ratio difference f and the inter-group phase difference The class labels of CVT samples are: normal, abnormal, and warning.

[0045]

[0046]

[0047] Among them, A, B, and C represent the three phases of the CVT, i represents the group, j represents the row of the vector, and V Aij represents the amplitude of the A-phase CVT in the j-th row of the i-th group; the concept drift refers to the change in the underlying data distribution, that is:

[0048]

[0049] At time t, the goal of the ensemble model is to construct a The optimal model of distribution M t Therefore, the first base model is generated according to formula (6):

[0050]

[0051] The first base model is updated according to the loss function (7):

[0052]

[0053] (7); where χ(·) represents an arbitrary distribution function, E(·) represents the expected value of a random variable, L is the loss function, where k = 1, 2, ..., N; MSE represents the predicted value and the true value , λ is the learning parameter (initial value is 0), θ(M_new) is the parameter of the new model, and θ(M_orign) is the parameter of the original model.

[0054] Step S102: Acquire first CVT real-time data, and detect whether concept drift occurs between the first CVT real-time data and CVT historical data.

[0055] As one embodiment, a KL divergence and a drift threshold are calculated according to the first CVT real-time data and the CVT historical data; and whether a concept drift occurs is determined according to the KL divergence and the drift threshold.

[0056] The CVT data set formed by the CVT historical data and the first CVT real-time data is subjected to mean shift clustering as a whole, and the drift threshold is calculated according to the formula R=D-2r, where D is the average distance between cluster centers and r is the average radius of the cluster.

[0057] The KL divergence is calculated according to formula (8):

[0058]

[0059] In the formula, the CVT historical data is recorded as D old , whose mean vector and covariance matrix are μ old and η old ; The first CVT real-time data is D new , whose mean vector and covariance matrix are μ new and η new ; d is the input sample dimension, preferably, d=2; tr represents the trace of the matrix; T represents the transpose of the matrix.

[0060] Step S103: When concept drift occurs, incremental data of the first CVT real-time data relative to the CVT historical data is obtained, and an incremental base model is generated according to the incremental data.

[0061] An adversarial neural network algorithm is used to perform oversampling processing on minority class samples in the incremental data, and the incremental base model is generated according to the incremental data after the oversampling processing.

[0062] Step S104: Generate an adaptive incremental ensemble learning model based on the incremental base model and the reference state prediction model, obtain the second CVT real-time data, and perform error state prediction on the second CVT real-time data based on the adaptive incremental ensemble learning model. The error state includes normal, abnormal and warning.

[0063] As one of the embodiments, after replacing the base model with the worst classification effect among the k second base models with the incremental base model, the incremental base model and the remaining k-1 second base models are fused to obtain an adaptive incremental ensemble learning model.

[0064] As one of the embodiments, after replacing the base model with the worst classification effect among the k first base models with the incremental base model, the remaining k-1 first base models and the incremental base model are updated into corresponding k second base models by the cross-validation method; and the k second base models are fused into a baseline state prediction model.

[0065] As one of the embodiments, the incremental base model and the reference state prediction model are fused to obtain an adaptive incremental ensemble learning model.

[0066] When the calculation added by the incremental base model does not exceed the maximum memory and limited time of the computer, the adaptive incremental ensemble learning model is obtained by fusing the incremental base model with the reference state prediction model. When the calculation added by the incremental base model exceeds the maximum memory and limited time of the computer, the adaptive incremental ensemble learning model is obtained by replacing the base model with the incremental base model and then fusing it.

[0067] The present invention detects the possible concept drift phenomenon of CVT historical data and CVT real-time data, and adopts an adaptive incremental integrated classification model to cope with dynamic data streams based on the concept drift detection results, thereby improving the accuracy of the error state evaluation model and the adaptability range of the model, thereby improving the accuracy of the CVT error state evaluation.

[0068] Based on the above-mentioned embodiments of the invention, the present invention provides corresponding device embodiments, such as Figure 2 As shown;

[0069] Another embodiment of the present invention provides a CVT error state prediction device based on an incremental ensemble learning model, comprising a base model generation module 101, a concept drift detection module 102, an incremental base model generation module 103 and an error state prediction module 104;

[0070] The base model generation module is used to divide the CVT historical data of power outage inspection into several data blocks, generate a corresponding number of base models according to the several data blocks, and fuse the base models into a reference state prediction model;

[0071] The concept drift detection module is used to obtain first CVT real-time data and detect whether concept drift occurs between the first CVT real-time data and CVT historical data;

[0072] The incremental base model generation module is used for acquiring incremental data of the first CVT real-time data relative to the CVT historical data when concept drift occurs, and generating an incremental base model according to the incremental data;

[0073] The error state prediction module is used to generate an adaptive incremental integrated learning model according to the incremental base model and the reference state prediction model, obtain the second CVT real-time data, and perform error state prediction on the second CVT real-time data according to the adaptive incremental integrated learning model.

[0074] For the convenience and brevity of description, the device item embodiment of the present invention includes all the implementation methods in the above-mentioned CVT error state prediction method embodiment based on the incremental ensemble learning model, which will not be repeated here.

[0075] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0076] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

[0077] Those skilled in the art can understand that all or part of the processes in the above embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above embodiments. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

Claims

1. A CVT error state prediction method based on an incremental ensemble learning model, characterized in that: The following steps are involved: Divide the CVT historical data of power outage inspection into a number of data blocks, generate a corresponding number of base models according to the data blocks, and fuse the base models into a reference state prediction model; Acquire first CVT real-time data, and detect whether concept drift occurs between the first CVT real-time data and CVT historical data; When concept drift occurs, obtaining incremental data of the first CVT real-time data relative to the CVT historical data, and generating an incremental base model according to the incremental data; Generate an adaptive incremental integrated learning model according to the incremental base model and the reference state prediction model, obtain second CVT real-time data, and perform error state prediction on the second CVT real-time data according to the adaptive incremental integrated learning model; The step of generating a corresponding number of base models according to the plurality of data blocks and fusing the base models into a reference state prediction model comprises the following steps: The k data blocks are correspondingly formed into k first base models, and then the k first base models are updated into corresponding k second base models by a cross-validation method; wherein k is a positive integer greater than 3; fusing the k second base models into a reference state prediction model; After replacing the base model with the worst classification effect among the k second base models with the incremental base model, the incremental base model and the remaining k-1 second base models are merged to obtain an adaptive incremental ensemble learning model; After replacing the base model with the worst classification effect among the k first base models with the incremental base model, the CVT error state prediction method further includes: updating the remaining k-1 first base models and the incremental base model into corresponding k second base models by a cross-validation method; fusing the k second base models into a reference state prediction model; The CVT error state prediction method also includes: using an adversarial neural network algorithm to oversample minority class samples in the incremental data, and generating the incremental basis model according to the incremental data after the oversampling process.

2. The CVT error state prediction method based on the incremental ensemble learning model according to claim 1 is characterized in that: For any of the first base models, the first base model is generated according to one data block among the k data blocks, and the remaining k-1 data blocks except the one data block are used to cross-validate the first base model to obtain the corresponding second base model.

3. The CVT error state prediction method based on the incremental ensemble learning model according to claim 2 is characterized in that: The incremental base model and the reference state prediction model are integrated to obtain an adaptive incremental ensemble learning model.

4. The CVT error state prediction method based on the incremental ensemble learning model according to claim 1 or 3, characterized in that: Calculating a KL divergence and a drift threshold according to the first CVT real-time data and the CVT historical data; Whether concept drift occurs is determined according to the KL divergence and the drift threshold.

5. The CVT error state prediction method based on the incremental ensemble learning model according to claim 4 is characterized in that: The CVT data set formed by the CVT historical data and the first CVT real-time data is clustered by mean shift as a whole, and the drift threshold is calculated according to the formula R=D-2r, where D is the average distance between cluster centers and r is the average radius of the cluster.

6. A CVT error state prediction device based on an incremental ensemble learning model, characterized in that: It includes a base model generation module, a concept drift detection module, an incremental base model generation module and an error state prediction module; The base model generation module is used to divide the CVT historical data of power outage inspection into several data blocks, generate a corresponding number of base models according to the several data blocks, and fuse the base models into a reference state prediction model; The concept drift detection module is used to obtain first CVT real-time data and detect whether concept drift occurs between the first CVT real-time data and CVT historical data; The incremental base model generation module is used for acquiring incremental data of the first CVT real-time data relative to the CVT historical data when concept drift occurs, and generating an incremental base model according to the incremental data; The error state prediction module is used to generate an adaptive incremental integrated learning model according to the incremental base model and the reference state prediction model, obtain the second CVT real-time data, and perform error state prediction on the second CVT real-time data according to the adaptive incremental integrated learning model; The step of generating a corresponding number of base models according to the plurality of data blocks and fusing the base models into a reference state prediction model comprises the following steps: The k data blocks are correspondingly formed into k first base models, and then the k first base models are updated into corresponding k second base models by a cross-validation method; wherein k is a positive integer greater than 3; fusing the k second base models into a reference state prediction model; After replacing the base model with the worst classification effect among the k second base models with the incremental base model, the incremental base model and the remaining k-1 second base models are merged to obtain an adaptive incremental ensemble learning model; After replacing the base model with the worst classification effect among the k first base models with the incremental base model, the method further includes: updating the remaining k-1 first base models and the incremental base model into corresponding k second base models by a cross-validation method; fusing the k second base models into a reference state prediction model; The incremental base model generation module is also used to: use an adversarial neural network algorithm to oversample the minority class samples in the incremental data, and generate the incremental base model according to the incremental data after the oversampling process.

Images