Method and device for constructing error evaluation model of current transformer and method and device for evaluating error

Abstract

The invention provides a current transformer error evaluation model construction method and device, and an error evaluation method and device, and the method comprises the steps: obtaining a plurality of pieces of historical single-phase / three-phase amplitude initial data collected by a current transformer, and corresponding error measurement values; performing preprocessing and feature value extraction on the data to generate a historical sample set; and training and verifying the error evaluation initial model based on the stacking ensemble learning algorithm based on the historical sample set data, and generating a current device error evaluation model. According to the error evaluation model generated by the method, due to the fact that the advantages of different machine learning algorithm models are brought into full play, deviation and variance possibly existing in a single model are effectively reduced, generalization and robustness of the model are enhanced, and due to the fact that the model is trained and verified through a cross validation method, the error evaluation efficiency is improved. The training data is fully utilized, and the risk of over-fitting is reduced, so that the accuracy of error evaluation is improved.

Description

Technical Field

[0001] The present invention relates to the field of current transformer error evaluation, and more specifically, to the construction of a current transformer error evaluation model, an error evaluation method, and an apparatus. Background Art

[0002] In the era of big technology, fault diagnosis of power transformers based on the error evaluation of current transformers has integrated a variety of advanced technologies, such as Bayesian networks, deep learning, support vector machines, and extreme learning machines. These technologies have significantly improved the accuracy and effectiveness of diagnosis. With the continuous development of artificial intelligence, new methods such as artificial neural networks and support vector machines have been widely applied in this field, achieving intelligent fault diagnosis. The evaluation technology of current transformers includes a variety of classification technologies. There are mainly two methods to solve the fault diagnosis problem: one is through time-frequency analysis technologies of deep learning, such as wavelet transform, S transform, and Huang-Hilbert transform, etc.; the other is through traditional machine learning using classification algorithms and learning machines, such as artificial neural networks and tree-based algorithms, etc. Among them, for the error evaluation model obtained by traditional machine learning, a large amount of high-quality data is required for training. In the case of insufficient or low-quality data, the performance of the model may be significantly affected, and traditional machine learning generally uses a single learner for classification and prediction, resulting in the accuracy of prediction being affected as well. Summary of the Invention

[0003] In order to solve the technical problem in the prior art that traditional machine learning for constructing a current transformer error model generally uses a single learner, and has relatively high requirements for the quality and quantity of data, and is prone to low prediction accuracy when the data quality is average or the quantity is insufficient, the present invention provides a method for constructing a current transformer error evaluation model, an error evaluation method, and an apparatus.

[0004] According to one aspect of the present invention, the present invention provides a method for constructing a current transformer error evaluation model, the method comprising:

[0005] Obtaining a plurality of historical single-phase / three-phase amplitude initial data collected by a current transformer, and error measurement values corresponding to the historical single-phase / three-phase amplitude initial data, wherein the error measurement values are ratio errors and / or phase angle errors;

[0006] Performing data preprocessing on the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data;

[0007] Performing variational mode decomposition (VMD) on the historical single-phase / three-phase amplitude valid data, and extracting historical feature values;

[0008] Establish a historical sample set based on the historical characteristic values of the historical single-phase / three-phase amplitude effective data and the corresponding error measurement values;

[0009] Use the historical sample set to train an initial error evaluation model based on the stacking ensemble learning method to generate a current transformer error evaluation model. Among them, the initial error evaluation model includes a base learner composed of multiple first initial error prediction models and a meta-learner composed of a second initial error prediction model, and the base learner and the meta-learner are cascaded.

[0010] According to another aspect of the present invention, the present invention provides a device for constructing a current transformer error evaluation model, and the device includes:

[0011] A historical data module for obtaining a plurality of historical single-phase / three-phase amplitude initial data collected by a current transformer and error measurement values corresponding to the historical single-phase / three-phase amplitude initial data, where the error measurement values are ratio errors and / or phase angle errors;

[0012] A first preprocessing module for preprocessing the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude effective data;

[0013] A first extraction module for performing variational mode decomposition (VMD) on the historical single-phase / three-phase amplitude effective data and extracting historical characteristic values;

[0014] A sample set module for establishing a historical sample set based on the historical characteristic values of the historical single-phase / three-phase amplitude effective data and the corresponding error measurement values;

[0015] An optimal model module for using the historical sample set to train an initial error evaluation model based on the stacking ensemble learning method to generate a current transformer error evaluation model. Among them, the initial error evaluation model includes a base learner composed of multiple first initial error prediction models and a meta-learner composed of a second initial error prediction model, and the base learner and the meta-learner are cascaded.

[0016] According to yet another aspect of the present invention, the present invention provides a method for evaluating the error of a current transformer, and the method includes:

[0017] Collect single-phase / three-phase amplitude data of the current transformer to be evaluated;

[0018] Preprocess the single-phase / three-phase amplitude data to generate effective single-phase / three-phase amplitude data;

[0019] Perform VMD decomposition on the effective single-phase / three-phase amplitude data and extract characteristic values;

[0020] Input the eigenvalue into the constructed error evaluation model of the current transformer. After being processed by the error evaluation model of the current transformer, an error prediction value of the current transformer to be evaluated is output, where the error prediction value includes ratio error and / or phase angle error. The error evaluation model of the current transformer is constructed based on any one of the construction methods of the error evaluation model of the current transformer described in the present invention, or the construction device of the error evaluation model of the current transformer.

[0021] According to another aspect of the present invention, the present invention provides an error evaluation device for a current transformer, and the device includes:

[0022] A data acquisition module, configured to acquire single-phase / three-phase amplitude data of the current transformer to be evaluated;

[0023] A second preprocessing module, configured to preprocess the single-phase / three-phase amplitude data to generate effective single-phase / three-phase amplitude data;

[0024] A second extraction module, configured to perform VMD decomposition on the effective single-phase / three-phase amplitude data and extract eigenvalues;

[0025] A result output module, configured to input the eigenvalue into the constructed error evaluation model of the current transformer. After being processed by the error evaluation model of the current transformer, an error prediction value of the current transformer to be evaluated is output, where the error prediction value includes ratio error and / or phase angle error. The error evaluation model of the current transformer is constructed based on any one of the construction methods of the error evaluation model of the current transformer described in the present invention, or the construction device of the error evaluation model of the current transformer.

[0026] According to another aspect of the present invention, the present invention provides a computer-readable storage medium, and the storage medium stores a computer program, and the computer program is used to execute the method described in any one of the above aspects of the present invention.

[0027] According to another aspect of the present invention, an electronic device is provided, and the electronic device includes: a processor; a memory for storing executable instructions of the processor; the processor is configured to read the executable instructions from the memory and execute the instructions to implement the method described in any one of the above aspects of the present invention.

[0028] Construction of current transformer error evaluation model, error evaluation method and device according to the present invention. In the model construction method, based on obtaining a number of historical single-phase / three-phase amplitude initial data collected by the current transformer and the corresponding error measurement values, a historical sample set is generated by preprocessing the data and extracting eigenvalue; based on the historical sample set data, an initial error evaluation model based on the stacking ensemble learning algorithm is trained and verified to generate a current transformer error evaluation model. The error evaluation model generated by the method of the present invention effectively reduces the bias and variance that may exist in a single model, enhances the generalization and robustness of the model, and because the method of cross-validation is used to train and verify the model, the training data is more fully utilized, reducing the risk of overfitting, thus improving the accuracy of error evaluation, as it gives full play to the advantages of different machine learning algorithm models. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The exemplary embodiments of the present invention can be more fully understood by referring to the following drawings:

[0030] Figure 1 It is a flowchart of the construction method of the current transformer error evaluation model according to the preferred embodiment of the present invention;

[0031] Figure 2 It is a schematic structural diagram of the construction device of the current transformer error evaluation model according to the preferred embodiment of the present invention;

[0032] Figure 3 It is a flowchart of the current transformer error evaluation method according to the preferred embodiment of the present invention;

[0033] Figure 4 It is a schematic structural diagram of the current transformer error evaluation device according to the preferred embodiment of the present invention;

[0034] Figure 5 It is a schematic structural diagram of the electronic device according to the preferred embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] Now, the exemplary embodiments of the present invention will be described with reference to the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely, and to fully convey the scope of the present invention to those skilled in the art. The terms in the exemplary embodiments shown in the drawings are not intended to limit the present invention. In the drawings, the same unit / element is denoted by the same reference numeral.

[0036] Unless otherwise specified, the terms used herein (including technical terms) have the ordinary meaning understood by those skilled in the relevant technical field. Additionally, it can be understood that terms defined in commonly used dictionaries should be construed to have a meaning consistent with the context of their relevant fields, and should not be construed as idealized or overly formal meanings.

[0037] Exemplary Method 1

[0038] Figure 1 It is a flowchart of a method for constructing an error evaluation model of a current transformer according to a preferred embodiment of the present invention. As Figure 1 shown, the method for constructing the error evaluation model of the current transformer described in this preferred embodiment starts from step 101.

[0039] In step 101, a plurality of historical single-phase / three-phase amplitude initial data collected by the current transformer, and error measurement values corresponding to the historical single-phase / three-phase amplitude initial data are obtained, where the error measurement values are ratio errors and / or phase angle errors.

[0040] In step 102, data preprocessing is performed on the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data.

[0041] Preferably, the data preprocessing of the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data includes:

[0042] Performing data cleaning on the historical single-phase / three-phase amplitude initial data to remove data that does not meet the set experimental requirements and / or deviation thresholds, and generating first historical data;

[0043] Performing missing value processing on the first historical data to generate second historical data;

[0044] Performing outlier elimination on the second historical data to identify and delete data that affects the performance of the proposed current transformer error evaluation model, and generating third historical data;

[0045] Performing standardization processing on the third historical data to generate historical single-phase / three-phase amplitude valid data.

[0046] Data preprocessing is a crucial step in ensuring the performance and stability of the model. This stage includes multiple steps such as data cleaning, handling missing values, and dealing with outliers, like elimination and data standardization or normalization. Data cleaning mainly involves removing data points that do not meet the experimental requirements or have obvious deviations from the actual situation. Handling missing values is usually accomplished through interpolation or methods based on statistical models. Outlier elimination is achieved through a series of statistical tests to identify and remove data points that may affect the model's performance. Data standardization or normalization is to eliminate the influence of different dimensions and data ranges on the model's performance, enabling the model to more accurately capture the relationships between various feature symbols. The data standardization formula adopted in this preferred embodiment is as follows:

[0047]

[0048] In the formula, x is the original data, mean is the mean of the original data, and std is the standard deviation of the original data.

[0049] In step 103, perform variational mode decomposition (VMD) on the historical single-phase / three-phase amplitude valid data and extract historical feature values.

[0050] Preferably, performing variational mode decomposition (VMD) on the historical single-phase / three-phase amplitude valid data and extracting historical feature values includes:

[0051] Use the grey wolf optimization algorithm to determine the number of decomposition layers C for performing VMD on the historical single-phase / three-phase amplitude valid data;

[0052] Decompose the historical single-phase / three-phase amplitude valid data into sub-signals of C decomposition layers through VMD;

[0053] Respectively extract historical feature values from the sub-signals of the C decomposition layers using electromagnetic signal processing methods to obtain the historical feature values corresponding to the historical single-phase / three-phase amplitude valid data.

[0054] In this preferred embodiment, the VMD used for eigenvalue extraction is an adaptive and completely non - recursive modal decomposition and signal processing method. Its adaptability lies in determining the number of modal decompositions of a given sequence according to the actual situation. In this preferred embodiment, the grey wolf optimization algorithm is used to determine the number of modal decompositions. In the grey wolf optimization algorithm, the fitness value reflects the quality of each individual in the current solution space. Its value indicates the higher the quality of the individual solution according to whether the optimization goal is smaller or larger. In this preferred embodiment, the selection of the fitness function is based on the mean of permutation entropy, which is regarded as an objective evaluation criterion for the target problem. When using permutation entropy as the goal of the optimization algorithm, it is usually better to be smaller. Permutation entropy is an index used to measure the degree of disorder or randomness of a set of data. A smaller permutation entropy indicates that the arrangement of data is more orderly and regular, while a larger permutation entropy indicates that the arrangement of data is more disorderly and random. Here, the optimization aims to minimize the mean of permutation entropy, that is, to make the data arrangement more orderly and regular. Solving according to the grey wolf optimization algorithm, the number of modal decompositions when using VMD in this preferred embodiment is 2, that is, the collected amplitude - effective data is divided into two decomposition layers through VMD decomposition. Then, the eigenvalue extraction of the vibration signal is carried out on the two sub - signals of the amplitude - effective data by using common electromagnetic signal processing methods. Then, 12 widely used eigenvalues are extracted in the time domain and frequency domain, which are the mean value, standard deviation, skewness, kurtosis, maximum value, minimum value, peak - to - peak value, mean square deviation, amplitude factor, waveform factor, impact factor, and margin factor respectively.

[0055] In step 104, a historical sample set is established according to the historical eigenvalues and the corresponding error measurement values of the historical single - phase / three - phase amplitude - effective data.

[0056] In step 105, the initial error evaluation model based on the stacking ensemble learning method is trained using the historical sample set to generate a current transformer error evaluation model. Among them, the initial error evaluation model includes a base learner composed of multiple first initial error prediction models and a meta - learner composed of a second initial error prediction model, and the base learner and the meta - learner are cascaded.

[0057] Preferably, before training the initial error evaluation model based on the stacking ensemble learning method using the historical sample set to generate a current transformer error evaluation model, it also includes establishing the initial error evaluation model and using the Bayesian parameter tuning method to determine the hyperparameters of the initial error evaluation model. Among them, the base learner in the initial error evaluation model includes 3 first initial error prediction models, which are the first linear support vector machine model, the Gaussian support vector machine model, and the decision tree model, and the second initial error prediction model used by the meta - learner is the second linear support vector machine model.

[0058] In this preferred embodiment, three first initial error prediction models are selected in the base learner, namely, the first linear support vector machine model using a linear support vector machine, the Gaussian support vector machine model using a Gaussian support vector machine, and the decision tree model using a decision tree algorithm. Among them, the linear support vector machine is suitable for high-dimensional data applications and can better process the extracted multi-dimensional eigenvalue. The Gaussian support vector machine can effectively solve the problem that the standard support vector machine has poor ability to process white noise in the input sequence and can improve the classification accuracy when training with a small dataset. The decision tree algorithm can better reflect the mapping relationship between the target feature and the target value, so as to select the feature that has the most significant impact on the target variable for prediction. Therefore, the combination of the above three initial error prediction models can improve the classification accuracy when the training data is small.

[0059] Preferably, training the initial error evaluation model based on the stacking ensemble learning method with the historical sample set to generate a current transformer error evaluation model includes:

[0060] Dividing the historical sample set into K equal parts, where 1 part is randomly selected as the test set, and the remaining K - 1 parts are used as the training set to generate K groups of data;

[0061] Training the initial error evaluation model based on the K - 1 parts of the training set data in the K groups of data to generate an error evaluation verification model;

[0062] Performing K - fold cross - validation on the error evaluation verification model based on the 1 part of the test set data in the K groups of data to generate a current transformer error evaluation model.

[0063] In this preferred embodiment, K is set to 5. The historical sample set is evenly divided into 5 parts, and 1 part is randomly selected from the 5 parts as the test set, generating a total of 5 combinations of training set data and test set data. The training set data in the 5 groups are respectively input into each first initial error prediction model in the base learner, and then the results output by each first initial error prediction model are combined with the corresponding error measurement values in the training set as the input of the second initial error prediction model in the meta - learner for training to generate an error evaluation verification model. Then, the test set in the 5 groups of data is used to perform 5 - fold cross - validation on the error evaluation verification model to generate a current transformer error evaluation model. Using the cross - validation method to train the initial error evaluation model based on the stacking algorithm can make full use of the training data, thereby reducing the risk of overfitting.

[0064] The method for constructing the current transformer error evaluation model according to the preferred embodiment obtains a number of historical single-phase / three-phase amplitude initial data collected by the current transformer and the error measurement values corresponding to the historical single-phase / three-phase amplitude initial data, and generates a historical sample set by preprocessing the data and extracting eigenvalue; based on the data of the historical sample set, the initial error evaluation model based on the stacking ensemble learning algorithm is trained and verified to generate the current transformer error evaluation model. The error evaluation model generated by the method of the present invention gives full play to the advantages of different machine learning algorithm models, effectively reduces the bias and variance that may exist in a single model, enhances the generalization and robustness of the model, and because the cross-validation method is used to train and verify the model, the training data is more fully utilized, reducing the risk of overfitting, thereby improving the accuracy of error evaluation.

[0065] In order to evaluate the difference in the prediction effect of the current transformer error evaluation model generated by the traditional single machine learning algorithm and the cascaded error evaluation model using the stacking ensemble algorithm on the same current amplitude data, in this preferred embodiment, the single linear support vector machine model, the Gaussian kernel support vector machine, the decision tree, and the cascaded error evaluation model based on the stacking ensemble algorithm are respectively used to perform the error prediction of the ratio error (Table 1) and the error prediction of the angular error (Table 2) on the single-phase amplitude data, as shown below.

[0066] Table 1: Ratio error prediction results

[0067]

[0068] Table 2: Angular error prediction results

[0069]

[0070] As shown in Table 1 and Table 2, in this preferred embodiment, the mean square error (MSE), mean absolute error (MAE), and root mean square error (RMSE) are used to evaluate the error between the error prediction value and the error measurement value output by different models. According to the calculation results, when a model generated by a single machine learning algorithm, such as a linear support vector machine model, a Gaussian kernel support vector machine model, and a decision tree model, is used for error evaluation, compared with the error evaluation model generated by cascading the base learners composed of the first linear support vector machine model, Gaussian support vector machine model, and decision tree model and the meta-learner composed of the second linear support vector machine based on stacking ensemble learning, its MSE, MAE, and RMSE are all larger, indicating that the error evaluation model generated by the stacking ensemble learning algorithm has higher accuracy in predicting the precision error of the current transformer. Among the models obtained by a single machine learning algorithm, the linear support vector machine model has higher accuracy in predicting the precision error of the current transformer compared with the Gaussian kernel support vector machine model and the decision tree model. This is also the reason why the meta-learner uses a linear support vector machine when constructing the initial error evaluation model by the stacking ensemble learning algorithm.

[0071] On the basis of obtaining a number of historical single-phase / three-phase amplitude initial data collected by the current transformer and the corresponding error measurement values, the method for constructing the current transformer error evaluation model in this preferred embodiment generates a historical sample set by preprocessing the data and extracting eigenvalue; based on the historical sample set data, the initial error evaluation model based on the stacking ensemble learning algorithm is trained and verified to generate the current transformer error evaluation model. The error evaluation model generated by the method of the present invention gives full play to the advantages of different machine learning algorithm models, effectively reduces the bias and variance that may exist in a single model, enhances the generalization and robustness of the model, and because the cross-validation method is used to train and verify the model, makes more full use of the training data, reduces the risk of overfitting, and thus improves the accuracy of error evaluation.

[0072] Exemplary Device 1

[0073] Figure 2 FIG. is a schematic structural diagram of a device for constructing a current transformer error evaluation model according to a preferred embodiment of the present invention. As Figure 2 shown, the device 200 for constructing the current transformer error evaluation model described in this preferred embodiment includes:

[0074] The historical data module 201 is used to obtain a number of historical single-phase / three-phase amplitude initial data collected by a current transformer, and error measurement values corresponding to the historical single-phase / three-phase amplitude initial data, where the error measurement values are ratio errors and / or phase angle errors;

[0075] The first preprocessing module 202 is used to perform data preprocessing on the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data;

[0076] The first extraction module 203 is used to perform variational mode decomposition (VMD) on the historical single-phase / three-phase amplitude valid data and extract historical feature values;

[0077] The sample set module 204 is used to establish a historical sample set according to the historical feature values of the historical single-phase / three-phase amplitude valid data and the corresponding error measurement values;

[0078] The optimal model module 205 is used to train an initial error evaluation model based on the stacking ensemble learning method using the historical sample set to generate a current transformer error evaluation model, where the initial error evaluation model includes a base learner composed of a plurality of first initial error prediction models and a meta-learner composed of a second initial error prediction model, and the base learner and the meta-learner are cascaded.

[0079] Preferably, the first preprocessing module 202 performs data preprocessing on the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data, including:

[0080] Performing data cleaning on the historical single-phase / three-phase amplitude initial data to remove data that does not meet the set experimental requirements and / or deviation thresholds, and generating first historical data;

[0081] Performing missing value processing on the first historical data to generate second historical data;

[0082] Performing outlier elimination on the second historical data to identify and delete data that affects the performance of the proposed current transformer error evaluation model, and generating third historical data;

[0083] Performing standardization processing on the third historical data to generate historical single-phase / three-phase amplitude valid data.

[0084] Preferably, the first extraction module 203 performs variational mode decomposition (VMD) on the historical single-phase / three-phase amplitude valid data and extracts historical feature values, including:

[0085] Using the grey wolf optimization algorithm to determine the number of layers C for performing VMD decomposition on the historical single-phase / three-phase amplitude valid data;

[0086] Decompose the historical single-phase / three-phase amplitude effective data into sub-signals of C decomposition layers through VMD decomposition;

[0087] Respectively extract historical eigenvalues from the sub-signals of C decomposition layers by using electromagnetic signal processing methods to obtain the historical eigenvalues corresponding to the historical single-phase / three-phase amplitude effective data.

[0088] Preferably, the device further includes an initial model unit, which is used to establish an initial error evaluation model and determine the hyperparameters of the initial error evaluation model by using the Bayesian parameter tuning method before training the initial error evaluation model based on the stacking ensemble learning method with a historical sample set to generate a current transformer error evaluation model. Among them, the base learners in the initial error evaluation model include 3 first initial error prediction models, namely the first linear support vector machine model, the Gaussian support vector machine model and the decision tree model, and the second initial error prediction model used by the meta-learner is the second linear support vector machine model.

[0089] Preferably, before the optimal model module 205 trains the initial error evaluation model based on the stacking ensemble learning method with a historical sample set to generate a current transformer error evaluation model, it also includes establishing an initial error evaluation model and determining the hyperparameters of the initial error evaluation model by using the Bayesian parameter tuning method. Among them, the base learners in the initial error evaluation model include 3 first initial error prediction models, namely the first linear support vector machine model, the Gaussian support vector machine model and the decision tree model, and the second initial error prediction model used by the meta-learner is the second linear support vector machine model.

[0090] The steps of processing the basic historical data of the current transformer error evaluation model construction device described in this preferred embodiment, extracting features to generate a historical sample set, and then training and validating the initial error evaluation model generated based on the stacking ensemble learning algorithm with the historical sample set by using the K-fold cross-validation method to obtain the current transformer error evaluation model are the same as the steps of the current transformer error evaluation model construction method described in the present invention, and the achieved technical effects are also the same, which will not be elaborated here.

[0091] Exemplary Method 2

[0092] Figure 3 Is a flowchart of the error evaluation method for a current transformer according to a preferred embodiment of the present invention. As Figure 3 shown, the flowchart of the error evaluation method for the current transformer described in this preferred embodiment starts from step 301.

[0093] In step 301, collect single-phase / three-phase amplitude data of the current transformer to be evaluated.

[0094] In step 302, preprocess the single-phase / three-phase amplitude data to generate effective single-phase / three-phase amplitude data.

[0095] In step 303, perform VMD decomposition on the effective single-phase / three-phase amplitude data and extract eigenvalues.

[0096] In step 304, input the eigenvalues into the constructed current transformer error evaluation model. After being processed by the current transformer error evaluation model, output the error prediction value of the current transformer to be evaluated, where the error prediction value includes ratio error and / or phase angle error, and the current transformer error evaluation model is constructed based on the construction method of the current transformer error evaluation model of the present invention or the device for the construction method of the current transformer error evaluation model.

[0097] Compared with the traditional method, the error evaluation method of the current transformer in this preferred embodiment gives full play to the advantages of different machine learning algorithm models by using the error evaluation model, effectively reduces the bias and variance that may exist in a single model, enhances the generalization and robustness of the model, and moreover, due to using the cross-validation method to train and verify the model, makes more full use of the training data, reduces the risk of overfitting, and thus improves the accuracy of error evaluation.

[0098] Exemplary Device 2

[0099] Figure 4 It is a schematic structural diagram of an error evaluation device for a current transformer according to a preferred embodiment of the present invention. As Figure 4 shown, the error evaluation device 400 for the current transformer in this preferred embodiment includes:

[0100] A data acquisition module 401, configured to collect single-phase / three-phase amplitude data of the current transformer to be evaluated;

[0101] A second preprocessing module 402, configured to preprocess the single-phase / three-phase amplitude data to generate effective single-phase / three-phase amplitude data;

[0102] A second extraction module 403, configured to perform VMD decomposition on the effective single-phase / three-phase amplitude data and extract eigenvalues;

[0103] The result output module 404 is configured to input the eigenvalue into the constructed current transformer error evaluation model. After being processed by the current transformer error evaluation model, an error prediction value of the current transformer to be evaluated is output, where the error prediction value includes ratio error and / or phase angle error. The current transformer error evaluation model is constructed based on the construction method of the current transformer error evaluation model of the present invention or the device for constructing the current transformer error evaluation model.

[0104] The steps of predicting the current transformer error by the current transformer error evaluation device according to this preferred embodiment are the same as those of the current transformer error evaluation method of the present invention, and the achieved technical effects are also the same, which will not be elaborated here.

[0105] Exemplary Electronic Device

[0106] Figure 5 It is a schematic structural diagram of an electronic device according to a preferred embodiment of the present invention. The electronic device can be any one or both of the first device and the second device, or a stand-alone device independent of them. The stand-alone device can communicate with the first device and the second device to receive the input signals collected from them. Figure 5 It illustrates a block diagram of an electronic device according to an embodiment of the present disclosure. As Figure 5 shown, the electronic device includes one or more processors 501 and a memory 502.

[0107] The processor 501 can be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and can control other components in the electronic device to perform desired functions.

[0108] The memory 502 can include one or more computer program products, and the computer program products can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory can include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory can include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions can be stored on the computer-readable storage medium, and the processor 501 can run the program instructions to implement the energy consumption anomaly diagnosis method based on the enterprise energy consumption space and / or other desired functions described above in the disclosed embodiments. In one example, the electronic device can further include: an input device 503 and an output device 504, and these components are interconnected through a bus system and / or other forms of connection mechanisms (not shown).

[0109] In addition, the input device 503 can further include, for example, a keyboard, a mouse, etc.

[0110] The output device 504 can output various information to the outside. The output device 504 can include, for example, a display, a speaker, a printer, a communication network, and remote output devices connected thereto, etc.

[0111] Of course, for simplicity, Figure 5 only some of the components related to the present disclosure in the electronic device are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application scenarios, the electronic device may further include any other appropriate components.

[0112] Exemplary Computer Program Product and Computer Readable Storage Medium

[0113] In addition to the above methods and devices, an embodiment of the present disclosure may also be a computer program product, which includes computer program instructions that, when run by a processor, cause the processor to execute the steps in the method for constructing a current transformer error evaluation model according to various embodiments of the present disclosure described in the "Exemplary Method 1" section of this specification, and the steps in the method for evaluating the aging degree according to various embodiments of the present disclosure described in the "Exemplary Method 2" section.

[0114] The computer program product can be written in any combination of one or more programming languages for the program code to execute the operations of the embodiments of the present disclosure. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, executed as an independent software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0115] In addition, an embodiment of the present disclosure may also be a computer-readable storage medium, on which computer program instructions are stored, and the computer program instructions, when run by a processor, cause the processor to execute the steps in the method for constructing a current transformer error evaluation model according to various embodiments of the present disclosure described in the "Exemplary Method 1" section of this specification, and the steps in the method for evaluating the aging degree according to various embodiments of the present disclosure described in the "Exemplary Method 2" section.

[0116] The computer-readable storage medium may adopt any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0117] The basic principles of the present disclosure have been described in conjunction with specific embodiments. However, it should be noted that the advantages, benefits, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, benefits, effects, etc. are essential for each embodiment of the present disclosure. In addition, the above-mentioned specific details are only for illustrative and easy-to-understand purposes, rather than limitations, and the above details do not limit the present disclosure to necessarily adopt the above specific details for implementation.

[0118] Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For system embodiments, since they basically correspond to method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments.

[0119] The block diagrams of the devices, apparatuses, equipment, and systems involved in the present disclosure are only illustrative examples and do not intend to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open-ended terms, meaning "including but not limited to", and can be used interchangeably with each other. The word "or" and "and" used herein refer to the word "and / or" and can be used interchangeably with each other, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to" and can be used interchangeably with each other.

[0120] The apparatuses and methods of the present disclosure may be implemented in many ways. For example, the apparatuses and methods of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of the steps for the method is for illustration only, and the steps of the method of the present disclosure are not limited to the specific order described above, unless otherwise specifically stated. In addition, in some embodiments, the present disclosure may also be implemented as a program recorded in a recording medium, and these programs include machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

[0121] It should also be noted that in the apparatuses, devices, and methods of the present disclosure, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent solutions of the present disclosure. The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0122] The above description has been presented for purposes of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the form disclosed herein. Although several example aspects and embodiments have been discussed above, those skilled in the art will recognize some of their variations, modifications, alterations, additions, and subcombinations.

Claims

1. A method for constructing a current transformer error evaluation model, characterized in that: The method comprises: Acquire a number of historical single-phase / three-phase initial amplitude data collected by the current transformer, and error measurement values ​​corresponding to the historical single-phase / three-phase initial amplitude data, wherein the error measurement values ​​are ratio difference and / or angle difference; Performing data preprocessing on the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data; Performing a variable analysis mode VMD decomposition on the historical single-phase / three-phase amplitude effective data, and extracting historical characteristic values; Establishing a historical sample set according to the historical characteristic values ​​of the historical single-phase / three-phase amplitude valid data and the corresponding error measurement values; A historical sample set is used to train an initial error evaluation model based on a stacking ensemble learning method to generate a current transformer error evaluation model, wherein the initial error evaluation model includes a basic learner consisting of multiple first error prediction initial models and a meta learner consisting of a second error prediction initial model, and the basic learner and the meta learner are cascaded.

2. The method according to claim 1, characterized in that The data preprocessing of the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data includes: Cleaning the historical single-phase / three-phase amplitude initial data to remove data that does not meet the set experimental requirements and / or deviation thresholds, and generating first historical data; Performing missing value processing on the first historical data to generate second historical data; Eliminating outliers from the second historical data, identifying and deleting data that affects the performance of the current transformer error evaluation model to be constructed, and generating third historical data; The third historical data is standardized to generate historical single-phase / three-phase amplitude valid data.

3. The method according to claim 1, characterized in that The performing of a variable analysis mode VMD decomposition on the historical single-phase / three-phase amplitude effective data and extracting historical characteristic values ​​comprises: The Grey Wolf optimization algorithm is used to determine the number of layers C for VMD decomposition of the historical single-phase / three-phase amplitude valid data; Decomposing the historical single-phase / three-phase amplitude effective data into sub-signals of C decomposition layers by VMD; The sub-signals of the C decomposition layers are respectively subjected to historical eigenvalue extraction by adopting an electromagnetic signal processing method to obtain historical eigenvalues ​​corresponding to the historical single-phase / three-phase amplitude valid data.

4. The method according to claim 1, characterized in that: Before using historical sample sets to train an initial error assessment model based on a stacking ensemble learning method to generate a current transformer error assessment model, the method also includes establishing an initial error assessment model and determining hyperparameters of the initial error assessment model using a Bayesian parameter tuning method, wherein the basic learner in the initial error assessment model includes three first error prediction initial models, namely a first linear support vector machine model, a Gaussian support vector machine model and a decision tree model, and the second error prediction initial model used by the meta-learner is a second linear support vector machine model.

5. The method according to claim 1, characterized in that The method of using a historical sample set to train an initial error evaluation model based on a stacking ensemble learning method to generate a current transformer error evaluation model includes: Divide the historical sample set into K parts equally, randomly select 1 part as the test set, and the remaining K-1 parts as the training set to generate K groups of data; The error assessment initial model is trained based on K-1 sets of training set data in the K groups of data to generate an error assessment verification model; The error evaluation verification model is cross-validated K times based on one test set data in the K groups of data to generate a current transformer error evaluation model.

6. A device for constructing a current transformer error evaluation model, characterized in that: The device comprises: A historical data module, used to obtain a number of historical single-phase / three-phase amplitude initial data collected by the current transformer, and an error measurement value corresponding to the historical single-phase / three-phase amplitude initial data, wherein the error measurement value is a ratio difference and / or an angle difference; A first preprocessing module, used for performing data preprocessing on the historical single-phase / three-phase amplitude initial data to generate historical single-phase / three-phase amplitude valid data; A first extraction module is used to perform a VMD decomposition on the historical single-phase / three-phase amplitude effective data and extract historical eigenvalues; A sample set module, used to establish a historical sample set according to the historical characteristic values ​​of the historical single-phase / three-phase amplitude valid data and the corresponding error measurement values; The optimal model module is used to use a historical sample set to train an initial error evaluation model based on a stacking ensemble learning method to generate a current transformer error evaluation model, wherein the initial error evaluation model includes a basic learner composed of multiple first error prediction initial models and a meta learner composed of a second error prediction initial model, and the basic learner and the meta learner are cascaded.

7. A method for evaluating an error of a current transformer, characterized in that: The method comprises: Collect single-phase / three-phase amplitude data of the current transformer to be evaluated; Preprocessing the single-phase / three-phase amplitude data to generate effective single-phase / three-phase amplitude data; Perform VMD decomposition on the effective single-phase / three-phase amplitude data and extract eigenvalues; The characteristic value is input into the constructed current transformer error evaluation model, and after being processed by the current transformer error evaluation model, an error prediction value of the current transformer to be evaluated is output, wherein the error prediction value includes a ratio difference and / or an angle difference, and the current transformer error evaluation model is constructed based on the method described in any one of claims 1 to 5 above, or the device described in claim 6.

8. An error evaluation device for a current transformer, characterized in that: The device comprises: A data acquisition module, used to collect single-phase / three-phase amplitude data of the current transformer to be evaluated; A second preprocessing module, used for preprocessing the single-phase / three-phase amplitude data to generate effective single-phase / three-phase amplitude data; A second extraction module is used to perform VMD decomposition on the effective single-phase / three-phase amplitude data and extract characteristic values; A result output module is used to input the characteristic value into a constructed current transformer error evaluation model, and output an error prediction value of the current transformer to be evaluated after being processed by the current transformer error evaluation model, wherein the error prediction value includes a ratio difference and / or an angle difference, and the current transformer error evaluation model is constructed based on the method described in any one of claims 1 to 5 above, or the device described in claim 6.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and the computer program is used to execute the method described in any one of claims 1 to 5 or 7 above.

10. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the method described in any one of claims 1 to 5 or 7 above.

Images