Transformer core clamp piece breakage early warning method and device and computer equipment

Abstract

The application relates to a transformer core clamp wire breakage early warning method and device and computer equipment. The method comprises the following steps: obtaining parameters of a transformer core clamp in a normal grounding state and a wire breakage state, to obtain an original parameter-state data set; performing order raising processing on electric parameter characteristics in the original parameter-state data set, to obtain an order-raised parameter-state data set; calculating maximum mutual information coefficients of the order-raised parameter-state data set and the electric parameter characteristics representing the normal grounding state and the wire breakage state, to obtain a screening parameter-state data set; calculating Pearson correlation coefficients between electric parameter characteristics in the screening parameter-state data set, to obtain a final parameter-state data set; obtaining hyperparameters of the normal grounding state and the wire breakage state through a grid search method; and predicting whether the transformer core clamp has a wire breakage fault through a preset random forest model. The method can improve the accuracy of fault prediction.

Description

Technical Field

[0001] This application relates to the field of early warning technology, and in particular to a method, device, computer equipment, storage medium and computer program product for early warning of transformer core clamping wire breakage. Background Technology

[0002] Transformers are important equipment in power systems, and failures can have serious consequences. Therefore, it is crucial to implement fault monitoring and early warning. Common transformer faults include winding faults, core faults, tap changer faults, lead wire faults, insulation faults, and sealing faults, with core faults being the second most common.

[0003] During normal operation of a transformer, a non-uniform electric field is formed between the energized windings and components. The core clamps, situated within this field, will generate a floating potential. When this floating potential reaches the level that breaks down the transformer insulation gap, a localized, intermittent, minute discharge phenomenon will occur between the core and the grounded tank. Prolonged exposure to this state may lead to serious accidents. Therefore, relevant national standards stipulate that the transformer core clamps must be reliably grounded.

[0004] While traditional technologies include early warning schemes for transformer core clamp breaks, these schemes rely on analysis and early warning based on the experience of technical personnel and cannot provide accurate early warnings for transformer core clamp breaks. Summary of the Invention

[0005] Therefore, it is necessary to provide an accurate method, device, computer equipment, computer-readable storage medium, and computer program product for early warning of transformer core clamping wire breakage, in response to the above-mentioned technical problems.

[0006] Firstly, this application provides a method for early warning of wire breakage in transformer core clamps. The method includes:

[0007] Obtain the parameters of the transformer core clamps under normal grounding and open circuit conditions to obtain the original parameter-state dataset.

[0008] The electrical parameter features in the original parameter-state dataset are subjected to an upgrade process to obtain an upgraded parameter-state dataset.

[0009] Calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state. Based on the maximum mutual information coefficient, filter the upgraded parameter-state dataset to obtain the filtered parameter-state dataset.

[0010] Calculate the Pearson correlation coefficient between electrical parameter features in the filtering parameter-state dataset, and perform secondary filtering on the filtering parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset;

[0011] The parameters under the normal grounding state and the open circuit state are traversed and cross-validated by the grid search method to obtain the hyperparameters of the normal grounding state and the open circuit state.

[0012] Based on the hyperparameters and the final parameter-state dataset, a preset random forest model is used to predict whether there is a wire breakage fault in the transformer core clamping components.

[0013] In one embodiment, the step of performing an order-up process on the electrical parameter features in the original parameter-state dataset to obtain an ordered parameter-state dataset further includes:

[0014] The original parameter-state dataset is preprocessed.

[0015] In one embodiment, the step of calculating the maximum mutual information coefficient in the upgraded parameter-state dataset that represents the electrical parameter features of the normal grounding state and the open circuit state, and then filtering the upgraded parameter-state dataset based on the maximum mutual information coefficient, to obtain the filtered parameter-state dataset includes:

[0016] Obtain the target values ​​of preset electrical parameter characteristics representing normal grounding and open circuit states;

[0017] Based on the upgraded parameter-state dataset and the target value, calculate the maximum mutual information coefficient in the upgraded parameter-state dataset that represents the normal grounding state and the open circuit state in terms of electrical parameter characteristics.

[0018] The upgraded parameter-state dataset is obtained by filtering the upgraded parameter-state dataset based on the maximum mutual information coefficient.

[0019] In one embodiment, the step of calculating the Pearson correlation coefficient between electrical parameter features in the screening parameter-state dataset, and performing secondary screening on the screening parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset includes:

[0020] Calculate the Pearson correlation coefficients among the electrical parameter features in the selected parameter-state dataset;

[0021] If the Pearson correlation coefficient between the electrical parameter features in the selected parameter-state dataset is high, then the selected parameter-state dataset is further filtered based on the maximum mutual information coefficient to obtain the final parameter-state dataset.

[0022] In one embodiment, the hyperparameters for the normal grounding state and the open circuit state are obtained by traversing and cross-validating the parameters using a grid search method, including:

[0023] By using a grid search method, the parameters under the normal grounding state and the open circuit state are traversed to obtain all combinations of parameters;

[0024] Cross-validation of all the parameters was performed to obtain the hyperparameters for the normal grounding state and the open circuit state.

[0025] In one embodiment, the step of predicting whether there is a wire breakage fault in the transformer core clamping component using a preset random forest model based on the hyperparameters and the final parameter-state dataset further includes:

[0026] Divide the original parameter-state dataset into a training set and a test set;

[0027] The learning performance of the preset random forest model is evaluated based on the training and test sets.

[0028] Secondly, this application also provides a transformer core clamp breakage early warning device. The device includes:

[0029] The original parameter-state dataset acquisition module is used to acquire the parameters of the transformer core clamps under normal grounding and open circuit states, and obtain the original parameter-state dataset.

[0030] The upgraded parameter-state dataset acquisition module is used to perform upgraded processing on the electrical parameter features in the original parameter-state dataset to obtain the upgraded parameter-state dataset.

[0031] The parameter-state dataset acquisition module is used to calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state, and to filter the upgraded parameter-state dataset according to the maximum mutual information coefficient to obtain the filtered parameter-state dataset.

[0032] The final parameter-state dataset acquisition module is used to calculate the Pearson correlation coefficient between electrical parameter features in the filtered parameter-state dataset, and to perform secondary filtering on the filtered parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset.

[0033] The hyperparameter acquisition module is used to traverse and cross-validate the parameters under the normal grounding state and the open circuit state using a grid search method to obtain the hyperparameters under the normal grounding state and the open circuit state.

[0034] The open circuit fault prediction module is used to predict whether there is an open circuit fault in the transformer core clamping component based on the hyperparameters and the final parameter-state dataset and a preset random forest model.

[0035] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0036] Obtain the parameters of the transformer core clamps under normal grounding and open circuit conditions to obtain the original parameter-state dataset.

[0037] The electrical parameter features in the original parameter-state dataset are subjected to an upgrade process to obtain an upgraded parameter-state dataset.

[0038] Calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state. Based on the maximum mutual information coefficient, filter the upgraded parameter-state dataset to obtain the filtered parameter-state dataset.

[0039] Calculate the Pearson correlation coefficient between electrical parameter features in the filtering parameter-state dataset, and perform secondary filtering on the filtering parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset;

[0040] The parameters under the normal grounding state and the open circuit state are traversed and cross-validated by the grid search method to obtain the hyperparameters of the normal grounding state and the open circuit state.

[0041] Based on the hyperparameters and the final parameter-state dataset, a preset random forest model is used to predict whether there is a wire breakage fault in the transformer core clamping components.

[0042] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0043] Obtain the parameters of the transformer core clamps under normal grounding and open circuit conditions to obtain the original parameter-state dataset.

[0044] The electrical parameter features in the original parameter-state dataset are subjected to an upgrade process to obtain an upgraded parameter-state dataset.

[0045] Calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state. Based on the maximum mutual information coefficient, filter the upgraded parameter-state dataset to obtain the filtered parameter-state dataset.

[0046] Calculate the Pearson correlation coefficient between electrical parameter features in the filtering parameter-state dataset, and perform secondary filtering on the filtering parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset;

[0047] The parameters under the normal grounding state and the open circuit state are traversed and cross-validated by the grid search method to obtain the hyperparameters of the normal grounding state and the open circuit state.

[0048] Based on the hyperparameters and the final parameter-state dataset, a preset random forest model is used to predict whether there is a wire breakage fault in the transformer core clamping components.

[0049] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0050] Obtain the parameters of the transformer core clamps under normal grounding and open circuit conditions to obtain the original parameter-state dataset.

[0051] The electrical parameter features in the original parameter-state dataset are subjected to an upgrade process to obtain an upgraded parameter-state dataset.

[0052] Calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state. Based on the maximum mutual information coefficient, filter the upgraded parameter-state dataset to obtain the filtered parameter-state dataset.

[0053] Calculate the Pearson correlation coefficient between electrical parameter features in the filtering parameter-state dataset, and perform secondary filtering on the filtering parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset;

[0054] The parameters under the normal grounding state and the open circuit state are traversed and cross-validated by the grid search method to obtain the hyperparameters of the normal grounding state and the open circuit state.

[0055] Based on the hyperparameters and the final parameter-state dataset, a preset random forest model is used to predict whether there is a wire breakage fault in the transformer core clamping components.

[0056] The aforementioned transformer core clamping component disconnection early warning method, device, computer equipment, storage medium, and computer program products acquire parameters of the transformer core clamping component under normal grounding and disconnection states, obtaining an original parameter-state dataset; perform order-up processing on the electrical parameter features in the original parameter-state dataset to obtain an upgraded parameter-state dataset; calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding and disconnection states, and filter the upgraded parameter-state dataset based on the maximum mutual information coefficient to obtain a filtered parameter-state dataset; calculate the Pearson correlation coefficient between the electrical parameter features in the filtered parameter-state dataset, and perform a secondary filtering of the filtered parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain a final parameter-state dataset; traverse and cross-validate the parameters under normal grounding and disconnection states using a grid search method to obtain hyperparameters for normal grounding and disconnection states; and predict whether a disconnection fault exists in the transformer core clamping component using a preset random forest model based on the hyperparameters and the final parameter-state dataset. In this scheme, the electrical parameter features in the original parameter-state dataset are upgraded to a higher order. The maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state is calculated. The Pearson correlation coefficient between the electrical parameter features in the selected parameter-state dataset is calculated to obtain the final parameter-state dataset. A nonlinear relationship between the new feature parameters and the working state of the iron core clamp is established. The parameters in the normal grounding state and the open circuit state are traversed and cross-validated to obtain the hyperparameters of the normal grounding state and the open circuit state. The parameters closely related to the open circuit fault are selected, thereby improving the accuracy of fault prediction. Attached Figure Description

[0057] Figure 1 This is an application environment diagram of a transformer core clamp disconnection early warning method in one embodiment;

[0058] Figure 2 This is a flowchart illustrating a transformer core clamp disconnection early warning method in one embodiment;

[0059] Figure 3 This is a schematic diagram of the transformer core clamp port model under normal grounding conditions.

[0060] Figure 4 A schematic diagram of the transformer core clamp port model when the transformer core lead wire is broken;

[0061] Figure 5 This is a flowchart illustrating the random forest classification algorithm.

[0062] Figure 6 A schematic diagram illustrating the specific process of the early warning method for broken wires in transformer core clamps;

[0063] Figure 7 This is a flowchart illustrating a transformer core clamp breakage early warning method in another embodiment;

[0064] Figure 8 This is a structural block diagram of a transformer core clamp breakage early warning device in one embodiment;

[0065] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0067] The transformer core clamp breakage early warning method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. Terminal 102 acquires parameters of the transformer core clamp in normal grounding and open-circuit states and sends these parameters to server 104. Server 104 obtains the original parameter-state dataset by acquiring the parameters of the transformer core clamp in normal grounding and open-circuit states. It then performs an upgrade process on the electrical parameter features in the original parameter-state dataset to obtain an upgraded parameter-state dataset. It calculates the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding and open-circuit states, and filters the dataset based on the maximum mutual information coefficient. A selection of the upgraded parameter-state dataset yields a filtered parameter-state dataset. The Pearson correlation coefficients among the electrical parameter features in the filtered parameter-state dataset are calculated. A secondary filtering of the filtered parameter-state dataset is then performed based on the maximum mutual information coefficient and the Pearson correlation coefficients to obtain the final parameter-state dataset. A grid search method is used to traverse and cross-validate the parameters under normal grounding and open-circuit states to obtain the hyperparameters for these states. Based on the hyperparameters and the final parameter-state dataset, a pre-defined random forest model is used to predict whether an open-circuit fault exists in the transformer core clamping components. The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, etc. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0068] In one embodiment, such as Figure 2As shown, a method for early warning of wire breakage in transformer core clamps is provided, which is applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0069] S100: Obtain the parameters of the transformer core clamps under normal grounding and open circuit conditions to obtain the original parameter-state dataset.

[0070] Among them, the parameter-state dataset is a dataset established by converting the parameters of multiple sets of transformer core clamps under different working states as data features, and the corresponding states as target values ​​into numerical features.

[0071] Specifically, the parameters of the transformer core clamps are measured under normal grounding and open-circuit conditions. The measured parameters include the external measurable clamp-to-ground capacitance C when the core leads are not grounded. 夹-地 When the iron core lead wire is grounded, the externally measurable clamp capacitance to ground C 夹-地 ′、Clamping component to iron core capacitor C 夹-铁 Clamping grounding current I 夹 Insulation resistance R of iron core clamp 铁 R 夹 ,like Figure 3 and Figure 4 As shown, the following equation represents the capacitance value of the iron core clamp when it is normally grounded:

[0072]

[0073] C 夹-地 =C1+C3

[0074] C 夹-地 =C1

[0075] The capacitance value when the internal lead wire of the iron core grounding is broken is expressed by the following equation:

[0076]

[0077]

[0078] C 夹-地 =C 11

[0079] Wherein, C1 is the distributed capacitance between the iron core and the clamping member, C2 is the distributed capacitance between the iron core and other internal structures, and C3 is the distributed capacitance between the clamping member and other structures; C 11 C is the parasitic capacitance between the iron core lead wire disconnected from the iron core grounding bushing and the clamp. 21 This refers to the parasitic capacitance between the iron core lead wire disconnected from the iron core grounding bushing and the internal grounding structure of the main transformer.

[0080] When the iron core clamp is normally grounded, the corresponding target value is set to 1. When the iron core clamp experiences a wire breakage fault, the corresponding target value is set to 0. A parameter-state dataset D is then established.

[0081] S200, the electrical parameter features in the original parameter-state dataset are upgraded to obtain the upgraded parameter-state dataset.

[0082] The order-up process involves squaring the values ​​of the electrical parameter features to obtain new sample features.

[0083] Specifically, in the parameter-state dataset D, the feature X is known. i Based on this, the eigenvalues ​​are raised by one order to obtain a new set of features. At this point, the sample features become 2k, resulting in the upgraded parameter-state dataset.

[0084] S300: Calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter feature characterization of normal grounding state and open circuit state. Based on the maximum mutual information coefficient, filter the upgraded parameter-state dataset to obtain the filtered parameter-state dataset.

[0085] Among them, the maximum mutual information coefficient is used to measure the degree of correlation between two feature variables, so as to screen out features that are highly correlated with the target values ​​of electrical parameter features representing normal grounding state and open circuit state.

[0086] Specifically, the MIC (Maximal Information Coefficient) of each electrical parameter feature and the target value of the electrical parameter feature representing the normal grounding state and the open circuit state is calculated in the upgraded parameter-state dataset, and the top m electrical parameter features with high correlation are selected to obtain the filtered parameter-state dataset.

[0087] S400: Calculate the Pearson correlation coefficient between electrical parameter features in the screening parameter-state dataset, and perform secondary screening on the screening parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset.

[0088] The Pearson correlation coefficient is used to measure the linear correlation between two features in a filter parameter-state dataset to eliminate redundant features.

[0089] Specifically, the Pearson correlation coefficients between features in the screening parameter-state dataset are calculated, as shown in Table 1. For two strongly correlated sets of features, the one with the larger MIC is retained to obtain the final parameter-state dataset.

[0090] Table 1. Criteria for Judging Pearson Correlation Coefficient

[0091]

[0092] The S500 uses a grid search method to traverse and cross-validate the parameters under normal grounding and open circuit conditions to obtain the hyperparameters for these conditions.

[0093] Among them, grid search is an exhaustive search method with specified parameter values. It obtains the optimal learning algorithm by optimizing the parameters of the estimated function through cross-validation. Cross-validation divides the dataset into two parts: one part is used as the training set and the other part is used as the validation set. First, the classifier is trained using the training set, and then the trained model is tested using the validation set.

[0094] Specifically, the parameters under normal grounding and open circuit conditions are traversed using a grid search method, and multiple sets of different parameters are obtained iteratively. In the cross-validation stage, the obtained multiple sets of different parameters are validated using the training set and the validation set to obtain the hyperparameters for normal grounding and open circuit conditions.

[0095] S600, based on hyperparameters and final parameter-state datasets, uses a pre-defined random forest model to predict whether there is a broken wire fault in the transformer core clamping components.

[0096] Specifically, the hyperparameters and final parameter-state dataset are used as input to a pre-defined random forest model to predict whether there is a broken wire fault in the transformer core clamping component. While decision tree algorithms are prone to overfitting with increasing tree depth, random forest algorithms utilize multiple subsets to generate classifiers, further improving the model's generalization ability. Information entropy, used as the basis for branching in the decision tree, can measure the mixing degree of the sample set, such as... Figure 5 As shown, its mathematical expression is:

[0097]

[0098]

[0099] In the formula, H(X) is the information entropy of a feature X, and p i Let X be the possible values ​​{x1, x2, ..., x}. n The probability of X taking a value is given by}, where n is the number of possible values ​​for X; H(D) is the information entropy of the sample set D under feature X, |x i | is the category x in feature X i The number of samples, where |D| is the total number of samples in the set;

[0100] When generating a decision tree, appropriate features need to be selected for branching, as shown in the following formula. The indicators that can be used include:

[0101] Information gain g(D,A): Branch points preferentially select features with large information gain. The larger the gain, the higher the information purity.

[0102] Information gain ratio g R (D,A): Instead of directly selecting the feature with the highest information gain, the feature is selected based on the proportion of information gain under feature A's classification.

[0103] Gini coefficient G: measures the probability that a sample in a set is misclassified, so features with small G values ​​should be preferred when selecting branch points;

[0104] Decision tree algorithms are prone to overfitting as the tree depth increases, while random forest algorithms use multiple subsets to generate classifiers, further improving the model's generalization ability.

[0105]

[0106] g(D,A)=H(D)-H(DA)

[0107]

[0108]

[0109] In the formula, H(D) and H(D|A) are the information entropy of D before and after classification by feature A, and |a i | represents the i-th possible value of feature A; G(D) is the Gini coefficient of D under the classification of feature A. For sample D, which belongs to category a i The probability of.

[0110] The entire process of the transformer core clamp breakage early warning method is as follows: Figure 6 As shown.

[0111] In the aforementioned transformer core clamp open circuit warning method, parameters of the transformer core clamp under normal grounding and open circuit states are obtained to obtain an original parameter-state dataset. The electrical parameter features in the original parameter-state dataset are then subjected to an upgrade process to obtain an upgraded parameter-state dataset. The maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding and open circuit states is calculated. Based on the maximum mutual information coefficient, the upgraded parameter-state dataset is filtered to obtain a filtered parameter-state dataset. Finally, the Pearson correlation coefficient between the electrical parameter features in the filtered parameter-state dataset is calculated. The parameter-state dataset is further filtered based on the maximum mutual information coefficient and Pearson correlation coefficient to obtain the final parameter-state dataset. The parameters under normal grounding and open circuit conditions are traversed and cross-validated using a grid search method to obtain hyperparameters for normal grounding and open circuit conditions. Based on the hyperparameters and the final parameter-state dataset, a pre-set random forest model is used to predict whether there is an open circuit fault in the transformer core clamp. This establishes a nonlinear relationship between new feature parameters and the working state of the core clamp, filters out parameters closely related to open circuit faults, and improves the accuracy of fault prediction.

[0112] In one embodiment, the electrical parameter features in the original parameter-state dataset are subjected to an order-up process to obtain an ordered parameter-state dataset, which further includes:

[0113] Preprocess the original parameter-state dataset.

[0114] The preprocessing involves removing outliers from the original parameter-state dataset and normalizing it, and then dividing it into training and test sets.

[0115] Specifically, to avoid errors caused by excessive differences in the dimensions between features, and to balance the influence of features on the target value, the data is normalized as shown in the following formula:

[0116]

[0117] In the formula, x and x′ represent the data before and after normalization, and x... max and x min These are the maximum and minimum values ​​of the parameter characteristics, respectively.

[0118] The normalized feature data ranges from [0,1]. The distribution of each feature's corresponding value is statistically analyzed. Based on the mean feature values ​​of the core clamp under normal grounding and open-circuit conditions, the feature reference point C is calculated. A reasonable threshold is set based on practical considerations, as shown in the following formula. The Euclidean distance L(X) from the remaining samples to C is calculated according to the working state. iC) As a criterion for outlier identification, if the sample size is sufficient, obvious outlier samples can be directly removed; otherwise, to ensure the accuracy of model training, mean imputation can be used for replacement.

[0119]

[0120] In the formula, X i C represents sample i and its corresponding reference point under the same operating conditions. Let k be the Euclidean distance between sample i and reference points on the first feature, and k be the number of features.

[0121] In this embodiment, by preprocessing the original parameter-state dataset, removing outliers and normalizing the dataset, it is easier to train the model subsequently.

[0122] In one embodiment, such as Figure 7 As shown, S300 includes:

[0123] S320, obtain the target values ​​of preset electrical parameter characteristics representing normal grounding state and open circuit state.

[0124] S340, based on the upgraded parameter-state dataset and the target value, calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter characteristics representing the normal grounding state and the open circuit state.

[0125] S360, based on the maximum mutual information coefficient, filters the upgraded parameter-state dataset to obtain the filtered parameter-state dataset.

[0126] The target value is set to the value corresponding to the electrical parameter characteristics representing the normal grounding state and the open circuit state. When the electrical parameter characteristics represent the normal grounding state, it is set to 1, and when the electrical parameter characteristics represent the open circuit state, it is set to 0.

[0127] Specifically, the MIC (Minimum Interpretation Value) of each feature with respect to the target value is calculated and compared, and the top m highly correlated features are selected for the next stage. For two variables that are related, mutual information can be used to measure the degree of their association, and the MIC formed based on this has higher accuracy; the larger the value, the stronger the association between the variables. The mathematical expression for this process is:

[0128]

[0129]

[0130] Where X and Y are two variables with a certain correlation, I[X; Y] is the relative entropy between the joint distribution p(X, Y) and the marginal distributions p(X)p(Y); |X||Y| < B is the condition for partitioning the grid for variables X and Y, and MIC[x; y] is the maximum normalized mutual information;

[0131] If the correlation between two parameter features is relatively high, then the one with a higher MIC is retained.

[0132] In this embodiment, by obtaining the target values characterizing the normal grounding state and the disconnection state of the preset electrical parameter features, according to the order-raising parameter-state data set and the target values, calculating the maximum mutual information coefficient in the order-raising parameter-state data set that characterizes the normal grounding state and the disconnection state of the electrical parameter features, and screening the order-raising parameter-state data set according to the maximum mutual information coefficient, a screened parameter-state data set can be obtained, and the electrical parameter features with a relatively high degree of association with the target values can be screened out.

[0133] In one embodiment, calculate the Pearson correlation coefficient between the electrical parameter features in the screened parameter-state data set, and perform a secondary screening on the screened parameter-state data set according to the maximum mutual information coefficient and the Pearson correlation coefficient, and the final parameter-state data set obtained includes:

[0134] Calculate the Pearson correlation coefficient between the electrical parameter features in the screened parameter-state data set; if the Pearson correlation coefficient between the electrical parameter features in the screened parameter-state data set is relatively high, then perform a secondary screening on the screened parameter-state data set according to the maximum mutual information coefficient to obtain the final parameter-state data set.

[0135] Specifically, calculate the Pearson correlation coefficient ρ between the first m features of the MIC values, as shown in the following formula:

[0136]

[0137]

[0138]

[0139]

[0140] Where E(X), E(Y) and Cov(X, Y) are the expectations and covariance of variables X and Y respectively, X i and Y i are the i-th in n samples; ρ XY is the Pearson correlation coefficient between variables X and Y, σ X and σ Y are the standard deviations of variables X and Y

[0141] In this embodiment, the Pearson correlation coefficient between electrical parameter features in the screening parameter-state dataset is calculated. If the Pearson correlation coefficient between electrical parameter features in the screening parameter-state dataset is high, the screening parameter-state dataset is further screened based on the maximum mutual information coefficient to obtain the final parameter-state dataset, which can screen strongly correlated electrical parameter features in the parameter-state dataset.

[0142] In one embodiment, the parameters for both normal grounding and open-circuit states are traversed and cross-validated using a grid search method to obtain the hyperparameters for both states, including:

[0143] By using a grid search method, the parameters under normal grounding and open circuit conditions are traversed to obtain all parameter combinations; cross-validation of all parameter combinations is then performed to obtain the hyperparameters for normal grounding and open circuit conditions.

[0144] Specifically, the selection of hyperparameters for the random forest model includes two stages: traversal and cross-validation. In the traversal stage, the range of values ​​for the number of tree parameters T and the depth N needs to be set according to the structure of the processed iron core clamp dataset, and all combinations of parameters are obtained through iteration. In the cross-validation stage, the sub-training set and sub-validation set are divided for the sub-decision tree samples generated by the random forest. The sub-validation set is used to measure the training effect, and finally the hyperparameters with the best performance are selected to train the model.

[0145] In this embodiment, the parameters under normal grounding and open circuit conditions are traversed using a grid search method to obtain all parameter combinations. Cross-validation of all parameter combinations is then performed to obtain the hyperparameters for normal grounding and open circuit conditions. This allows for the acquisition of the optimal parameters for the random forest model, thereby improving the accuracy of fault prediction.

[0146] In one embodiment, based on hyperparameters and the final parameter-state dataset, a pre-defined random forest model is used to predict whether there is a wire breakage fault in the transformer core clamping component. Prior to this, the method also includes:

[0147] Divide the original parameter-state dataset into training and test sets; and evaluate the learning performance of the pre-defined random forest model based on the training and test sets.

[0148] Specifically, the preprocessed original parameter-state dataset is divided into training and test sets to test the learning effect of the preset random forest model.

[0149] In this embodiment, by dividing the original parameter-state dataset into training and test sets, the learning effect of the preset random forest model is tested based on the training and test sets, which facilitates the subsequent model's prediction of disconnection faults.

[0150] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0151] Based on the same inventive concept, this application also provides an early warning device for implementing the above-mentioned method for early warning of transformer core clamp disconnection. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more early warning device embodiments provided below can be found in the limitations of the transformer core clamp disconnection early warning method described above, and will not be repeated here.

[0152] In one embodiment, such as Figure 8 As shown, a transformer core clamp wire breakage early warning device is provided, comprising: a raw parameter-state dataset acquisition module 100, an upgraded parameter-state dataset acquisition module 200, a filtered parameter-state dataset acquisition module 300, a final parameter-state dataset acquisition module 400, a hyperparameter acquisition module 500, and a wire breakage fault prediction module 600, wherein:

[0153] The original parameter-state dataset acquisition module 100 is used to acquire the parameters of the transformer core clamps under normal grounding and open circuit states, and obtain the original parameter-state dataset.

[0154] The upgraded parameter-state dataset acquisition module 200 is used to upgrade the electrical parameter features in the original parameter-state dataset to obtain the upgraded parameter-state dataset.

[0155] The parameter-state dataset acquisition module 300 is used to calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter characteristics representing the normal grounding state and the open circuit state. The upgraded parameter-state dataset is then filtered based on the maximum mutual information coefficient to obtain the filtered parameter-state dataset.

[0156] The final parameter-state dataset acquisition module 400 is used to calculate the Pearson correlation coefficient between electrical parameter features in the filtered parameter-state dataset, and to perform secondary filtering on the filtered parameter-state dataset based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset.

[0157] The hyperparameter acquisition module 500 is used to traverse and cross-validate the parameters under normal grounding and open circuit conditions using a grid search method to obtain the hyperparameters under normal grounding and open circuit conditions.

[0158] The open circuit fault prediction module 600 is used to predict whether there is an open circuit fault in the transformer core clamping parts based on the hyperparameters and the final parameter-state dataset and through a preset random forest model.

[0159] In one embodiment, the parameter-state dataset acquisition module 300 is further configured to acquire target values ​​representing the normal grounding state and the open circuit state of the preset electrical parameter features; calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state based on the upgraded parameter-state dataset and the target values; and filter the upgraded parameter-state dataset based on the maximum mutual information coefficient to obtain the filtered parameter-state dataset.

[0160] In one embodiment, the final parameter-state dataset acquisition module 400 is further used to calculate the Pearson correlation coefficient between electrical parameter features in the filtered parameter-state dataset; if the Pearson correlation coefficient between electrical parameter features in the filtered parameter-state dataset is high, the filtered parameter-state dataset is further filtered based on the maximum mutual information coefficient to obtain the final parameter-state dataset.

[0161] In one embodiment, the hyperparameter acquisition module 500 is further configured to traverse the parameters under normal grounding and open circuit conditions using a grid search method to obtain a combination of all parameters; and to perform cross-validation on the combination of all parameters to obtain the hyperparameters under normal grounding and open circuit conditions.

[0162] Each module in the aforementioned transformer core clamp breakage early warning device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0163] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9As shown, the computer device includes a processor, memory, and network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores parameter data of the transformer core clamps under normal grounding and open-circuit conditions. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a transformer core clamp open-circuit warning method.

[0164] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0165] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0166] The parameters of the transformer core clamping components under normal grounding and open-circuit conditions are obtained to obtain the original parameter-state dataset. The electrical parameter features in the original parameter-state dataset are then upgraded to obtain the upgraded parameter-state dataset. The maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding and open-circuit conditions is calculated. Based on the maximum mutual information coefficient, the upgraded parameter-state dataset is filtered to obtain the filtered parameter-state dataset. The Pearson correlation coefficient between the electrical parameter features in the filtered parameter-state dataset is calculated, and the filtered parameter-state dataset is further filtered based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset. A grid search method is used to traverse and cross-validate the parameters under normal grounding and open-circuit conditions to obtain the hyperparameters for these conditions. Based on the hyperparameters and the final parameter-state dataset, a pre-defined random forest model is used to predict whether an open-circuit fault exists in the transformer core clamping components.

[0167] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0168] Preprocess the original parameter-state dataset.

[0169] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0170] Obtain the target values ​​of the preset electrical parameter features representing the normal grounding state and the open circuit state; calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state based on the upgraded parameter-state dataset and the target values; filter the upgraded parameter-state dataset based on the maximum mutual information coefficient to obtain the filtered parameter-state dataset.

[0171] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0172] Calculate the Pearson correlation coefficient among the electrical parameter features in the screening parameter-state dataset; if the Pearson correlation coefficient among the electrical parameter features in the screening parameter-state dataset is high, then perform a second screening on the screening parameter-state dataset based on the maximum mutual information coefficient to obtain the final parameter-state dataset.

[0173] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0174] By using a grid search method, the parameters under normal grounding and open circuit conditions are traversed to obtain all parameter combinations; cross-validation of all parameter combinations is then performed to obtain the hyperparameters for normal grounding and open circuit conditions.

[0175] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0176] Divide the original parameter-state dataset into training and test sets; and evaluate the learning performance of the pre-defined random forest model based on the training and test sets.

[0177] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0178] The parameters of the transformer core clamping components under normal grounding and open-circuit conditions are obtained to obtain the original parameter-state dataset. The electrical parameter features in the original parameter-state dataset are then upgraded to obtain the upgraded parameter-state dataset. The maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding and open-circuit conditions is calculated. Based on the maximum mutual information coefficient, the upgraded parameter-state dataset is filtered to obtain the filtered parameter-state dataset. The Pearson correlation coefficient between the electrical parameter features in the filtered parameter-state dataset is calculated, and the filtered parameter-state dataset is further filtered based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset. A grid search method is used to traverse and cross-validate the parameters under normal grounding and open-circuit conditions to obtain the hyperparameters for these conditions. Based on the hyperparameters and the final parameter-state dataset, a pre-defined random forest model is used to predict whether an open-circuit fault exists in the transformer core clamping components.

[0179] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0180] Preprocess the original parameter-state dataset.

[0181] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0182] Obtain the target values ​​of the preset electrical parameter features representing the normal grounding state and the open circuit state; calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state based on the upgraded parameter-state dataset and the target values; filter the upgraded parameter-state dataset based on the maximum mutual information coefficient to obtain the filtered parameter-state dataset.

[0183] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0184] Calculate the Pearson correlation coefficient among the electrical parameter features in the screening parameter-state dataset; if the Pearson correlation coefficient among the electrical parameter features in the screening parameter-state dataset is high, then perform a second screening on the screening parameter-state dataset based on the maximum mutual information coefficient to obtain the final parameter-state dataset.

[0185] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0186] By using a grid search method, the parameters under normal grounding and open circuit conditions are traversed to obtain all parameter combinations; cross-validation of all parameter combinations is then performed to obtain the hyperparameters for normal grounding and open circuit conditions.

[0187] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0188] Divide the original parameter-state dataset into training and test sets; and evaluate the learning performance of the pre-defined random forest model based on the training and test sets.

[0189] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0190] The parameters of the transformer core clamping components under normal grounding and open-circuit conditions are obtained to obtain the original parameter-state dataset. The electrical parameter features in the original parameter-state dataset are then upgraded to obtain the upgraded parameter-state dataset. The maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding and open-circuit conditions is calculated. Based on the maximum mutual information coefficient, the upgraded parameter-state dataset is filtered to obtain the filtered parameter-state dataset. The Pearson correlation coefficient between the electrical parameter features in the filtered parameter-state dataset is calculated, and the filtered parameter-state dataset is further filtered based on the maximum mutual information coefficient and the Pearson correlation coefficient to obtain the final parameter-state dataset. A grid search method is used to traverse and cross-validate the parameters under normal grounding and open-circuit conditions to obtain the hyperparameters for these conditions. Based on the hyperparameters and the final parameter-state dataset, a pre-defined random forest model is used to predict whether an open-circuit fault exists in the transformer core clamping components.

[0191] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0192] Preprocess the original parameter-state dataset.

[0193] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0194] Obtain the target values ​​of the preset electrical parameter features representing the normal grounding state and the open circuit state; calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state based on the upgraded parameter-state dataset and the target values; filter the upgraded parameter-state dataset based on the maximum mutual information coefficient to obtain the filtered parameter-state dataset.

[0195] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0196] Calculate the Pearson correlation coefficient among the electrical parameter features in the screening parameter-state dataset; if the Pearson correlation coefficient among the electrical parameter features in the screening parameter-state dataset is high, then perform a second screening on the screening parameter-state dataset based on the maximum mutual information coefficient to obtain the final parameter-state dataset.

[0197] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0198] By using a grid search method, the parameters under normal grounding and open circuit conditions are traversed to obtain all parameter combinations; cross-validation of all parameter combinations is then performed to obtain the hyperparameters for normal grounding and open circuit conditions.

[0199] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0200] Divide the original parameter-state dataset into training and test sets; and evaluate the learning performance of the pre-defined random forest model based on the training and test sets.

[0201] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0202] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0203] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0204] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for early warning of wire breakage in transformer core clamps, characterized in that, The method includes: Obtain the parameters of the transformer core clamps under normal grounding and open circuit conditions to obtain the original parameter-state dataset. The electrical parameter features in the original parameter-state dataset are subjected to an upgrade process to obtain an upgraded parameter-state dataset. Obtain the target values ​​of preset electrical parameter characteristics representing normal grounding and open circuit states; Based on the upgraded parameter-state dataset and the target value, calculate the maximum mutual information coefficient in the upgraded parameter-state dataset that represents the normal grounding state and the open circuit state in terms of electrical parameter characteristics. The upgraded parameter-state dataset is obtained by filtering the upgraded parameter-state dataset based on the maximum mutual information coefficient; Calculate the Pearson correlation coefficients among the electrical parameter features in the selected parameter-state dataset; If the Pearson correlation coefficient between the electrical parameter features in the filtering parameter-state dataset is high, then the filtering parameter-state dataset is further filtered according to the maximum mutual information coefficient to obtain the final parameter-state dataset. The parameters under the normal grounding state and the open circuit state are traversed and cross-validated by the grid search method to obtain the hyperparameters of the normal grounding state and the open circuit state. Based on the hyperparameters and the final parameter-state dataset, a preset random forest model is used to predict whether there is a wire breakage fault in the transformer core clamping components.

2. The method according to claim 1, characterized in that, The step of performing an order-up process on the electrical parameter features in the original parameter-state dataset to obtain an ordered parameter-state dataset includes, prior to: The original parameter-state dataset is preprocessed.

3. The method according to claim 1, characterized in that, The process of using a grid search method to traverse and cross-validate the parameters under the normal grounding state and the open circuit state yields the following hyperparameters for the normal grounding state and the open circuit state: By using a grid search method, the parameters under the normal grounding state and the open circuit state are traversed to obtain all combinations of parameters; Cross-validation of all the parameters was performed to obtain the hyperparameters for the normal grounding state and the open circuit state.

4. The method according to claim 1, characterized in that, The step of predicting whether there is a broken wire fault in the transformer core clamping component based on the hyperparameters and the final parameter-state dataset using a preset random forest model also includes: Divide the original parameter-state dataset into a training set and a test set; The learning performance of the preset random forest model is evaluated based on the training and test sets.

5. A transformer core clamp breakage early warning device, characterized in that, The device includes: The original parameter-state dataset acquisition module is used to acquire the parameters of the transformer core clamps under normal grounding and open-circuit states, and obtain the original parameter-state dataset. The upgraded parameter-state dataset acquisition module is used to perform upgraded processing on the electrical parameter features in the original parameter-state dataset to obtain the upgraded parameter-state dataset. The parameter-state dataset acquisition module is used to acquire target values ​​representing the normal grounding state and the open circuit state of the preset electrical parameter features; calculate the maximum mutual information coefficient between the upgraded parameter-state dataset and the electrical parameter features representing the normal grounding state and the open circuit state based on the upgraded parameter-state dataset and the target values; and filter the upgraded parameter-state dataset based on the maximum mutual information coefficient to obtain the filtered parameter-state dataset. The final parameter-state dataset acquisition module is used to calculate the Pearson correlation coefficient between electrical parameter features in the filtered parameter-state dataset; if the Pearson correlation coefficient between electrical parameter features in the filtered parameter-state dataset is high, the filtered parameter-state dataset is further filtered based on the maximum mutual information coefficient to obtain the final parameter-state dataset. The hyperparameter acquisition module is used to traverse and cross-validate the parameters under the normal grounding state and the open circuit state using a grid search method to obtain the hyperparameters under the normal grounding state and the open circuit state. The open circuit fault prediction module is used to predict whether there is an open circuit fault in the transformer core clamping component based on the hyperparameters and the final parameter-state dataset and a preset random forest model.

6. The apparatus according to claim 5, characterized in that, The hyperparameter acquisition module is also used for: The parameters under the normal grounding state and the open circuit state are traversed using a grid search method to obtain all parameter combinations; the combinations of all parameters are then cross-validated to obtain the hyperparameters for the normal grounding state and the open circuit state.

7. The apparatus according to claim 5, characterized in that, The device is also used for: Divide the original parameter-state dataset into a training set and a test set; The learning performance of the preset random forest model is evaluated based on the training and test sets.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

Images