A method, device, equipment and storage medium for determining the measurement error of an instrument transformer

Abstract

The present invention relates to a method, device, equipment and storage medium for determining the measurement error of an instrument transformer, belonging to the technical field of instrument transformers. The method includes: predicting the measurement error of the target-time instrument transformer through a first model to obtain a first measurement error; the first model is constructed based on the change trend information of the first actual measurement error of the instrument transformer within the historical time; predicting the residual of the first measurement error through a second model to obtain a target residual; the second model is constructed based on the error fluctuation information of the first actual measurement error and the predicted residual of the first model; determining the target measurement error of the instrument transformer according to the first measurement error and the target residual. The present invention predicts the target measurement error based on the internal influencing factors of the measurement error, such as the error change trend and the error fluctuation information, which can make the predicted target measurement error more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of transformers, and particularly to a method, device, equipment and storage medium for determining the measurement error of a transformer. Background Art

[0002] The prediction of the measurement error of a transformer is an important link in the power system, which is related to the stable operation of the power system, the effective transmission of energy, and the reliability of user power consumption. As a key device in the power system, the transformer is mainly used for measurement and protection, and its accuracy is directly related to the safe and economic operation of the power system.

[0003] With the continuous expansion of the scale and the increasing complexity of the power system, the prediction and management of the measurement error of transformers become particularly important. The error of the transformer may be caused by various factors, such as equipment aging, environmental factors, manufacturing processes, etc. If these errors are not predicted and corrected in time, it may lead to inaccurate measurement, thus affecting the normal operation of the power system. The traditional methods for predicting the measurement error of transformers mainly rely on regular detection and calibration. This method is not only costly but also difficult to reflect the error state of the transformer in real time. Therefore, in recent years, the data-driven methods for predicting the measurement error of transformers have gradually attracted attention. These methods use a large amount of operation data and advanced machine learning algorithms to achieve fast and accurate prediction of the transformer error. However, these methods usually only consider the influence of external environmental factors, such as temperature, humidity and other factors on the measurement error of the transformer. Summary of the Invention

[0004] In view of this, it is necessary to provide a method, device, equipment and storage medium for determining the measurement error of a transformer to solve the problem that the existing technology only considers the influence of external environmental factors on the measurement error of the transformer.

[0005] To solve the above problems, in a first aspect, the present invention provides a method for determining the measurement error of a transformer, including:

[0006] Predicting the measurement error of the target-time transformer through a first model to obtain a first measurement error; the first model is constructed based on the change trend information of the first actual measurement error of the transformer within the historical time;

[0007] Predicting the residual of the first measurement error through a second model to obtain a target residual; the second model is constructed based on the error fluctuation information of the first actual measurement error and the prediction residual of the first model;

[0008] Determining the target measurement error of the transformer according to the first measurement error and the target residual.

[0009] Optionally, each date within the historical time period corresponds to a first actual measurement error and multiple second actual measurement errors; the first actual measurement error corresponding to each date within the historical time period is the average value of the multiple second actual measurement errors corresponding to each date; and, the change trend information of the first actual measurement error includes: the first actual measurement error corresponding to the target date, and the average value of the first actual measurement error within multiple time periods with the target date as the end date; the target date includes at least one.

[0010] Optionally, the error fluctuation information of the first actual measurement error includes at least one of the highest error fluctuation, lowest error fluctuation, error fluctuation, relative error fluctuation magnitude, and relative error fluctuation position corresponding to each date within the historical time period;

[0011] The error fluctuation information is calculated through the following formula:

[0012] , where represents the highest error fluctuation on the t th day; represents the maximum actual measurement error on the t th day, determined based on the multiple second actual measurement errors within the date corresponding to the t th day, represents the first actual measurement error corresponding to the t th day;

[0013] , where represents the lowest error fluctuation on the t th day; represents the minimum actual measurement error on the t th day, determined based on the multiple second actual measurement errors within the date corresponding to the t th day;

[0014] , where represents the error fluctuation on the t th day, represents the first actual measurement error corresponding to the th day;

[0015] , where represents the relative error fluctuation magnitude on the t th day;

[0016] , where represents the relative error fluctuation position on the t th day.

[0017] Optionally, the second model is constructed as follows:

[0018] Obtain the environmental information in the historical time; the environmental information includes at least one of temperature information, humidity information, and secondary load information of the current transformer;

[0019] Construct a first mapping relationship between the environmental information and the prediction residual based on the first sample set;

[0020] The first sample set is represented by the following formula:

[0021] , where represents the first sample set, respectively represent the environmental information corresponding to the (t + 1)-th day to the (t + n)-th day in the historical time; respectively represent the prediction residuals obtained after the first model predicts the measurement errors of the current transformer for the (t + 1)-th day to the (t + n)-th day in the historical time;

[0022] The first mapping relationship is represented by the following formula:

[0023] , where represents the first residual, represents the feature representation of the environmental information, represents the first bias, represents the first hyperplane weight vector, and are determined based on the first sample set;

[0024] Construct a second mapping relationship between the error fluctuation information and the prediction residual based on the second sample set;

[0025] The second sample set is represented by the following formula:

[0026] , where represents the second sample set, respectively represent the error fluctuation information corresponding to the t-th day to the (t + n - 1)-th day in the historical time;

[0027] The second mapping relationship is represented by the following formula:

[0028] , where represents the second residual, represents the feature representation of the error fluctuation information, represents the second bias, represents the second hyperplane weight vector; and determined based on the second sample set;

[0029] Construct a second model according to the first mapping relationship and the second mapping relationship.

[0030] Optionally, the predicting the residual of the first measurement error through the second model to obtain a target residual includes:

[0031] Determine the first residual of the first measurement error through the first mapping relationship;

[0032] Determine the second residual of the first measurement error through the second mapping relationship;

[0033] Determine the target residual according to the first residual and the second residual.

[0034] Optionally, the determining the target residual according to the first residual and the second residual includes:

[0035] Determine the first weight of the first residual and the second weight of the second residual;

[0036] Take the sum of the product of the first residual and the first weight and the product of the second residual and the second weight as the target residual.

[0037] Optionally, the determining the first weight of the first residual and the second weight of the second residual includes:

[0038] Determine the first sub-prediction residual corresponding to each prediction residual time according to the first mapping relationship;

[0039] Determine the first credibility and the first incredibility of the first mapping relationship according to the prediction residual and the first sub-prediction residual;

[0040] Determine the first weight of the first residual according to the first credibility and the first incredibility;

[0041] Determine the second sub-prediction residual corresponding to each prediction residual time according to the second mapping relationship;

[0042] Determine the second credibility and the second incredibility of the second mapping relationship according to the prediction residual and the second sub-prediction residual;

[0043] Determine the second weight of the second residual according to the second credibility and the second incredibility.

[0044] In a second aspect, the present invention further provides a device for determining the measurement error of an instrument transformer, including:

[0045] The first measurement error prediction module is used to predict the measurement error of the target time transformer through the first model to obtain the first measurement error; the first model is constructed based on the change trend information of the first actual measurement error of the transformer within the historical time;

[0046] The target residual prediction module is used to predict the residual of the first measurement error through the second model to obtain the target residual; the second model is constructed based on the error fluctuation information of the first actual measurement error and the predicted residual of the first model;

[0047] The second measurement error determination module is used to determine the target measurement error of the transformer according to the first measurement error and the target residual.

[0048] In a third aspect, the present invention also provides an electronic device, including a memory and a processor. Among them, the memory is used to store a program; the processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the transformer measurement error determination method described in any one of the above.

[0049] In a fourth aspect, the present invention also provides a computer-readable storage medium for storing a computer-readable program, and when the program or instruction is executed by a processor, it can implement the steps in the transformer measurement error determination method described in any one of the above.

[0050] The beneficial effects of the present invention are:

[0051] The measurement error of the target time transformer is predicted through the first model to obtain the first measurement error; the first model is constructed based on the change trend information of the first actual measurement error of the transformer within the historical time. The first model predicts the first measurement error of the transformer at a future time (target time) based on the change trend information of the first actual measurement error.

[0052] The residual of the first measurement error is predicted through the second model to obtain the target residual; the second model is constructed based on the error fluctuation information of the first actual measurement error and the predicted residual of the first model. The second model studies the influence of the error fluctuation information of the first actual measurement error on the predicted residual when the first model makes a prediction, and predicts the residual of the first measurement error through the relationship between the error fluctuation information of the first actual measurement error and the predicted residual of the first model to obtain the target residual. Finally, the first measurement error is corrected according to the target residual to obtain the target measurement error, that is, the finally determined measurement error of the transformer.

[0053] The present invention proposes a new prediction idea, mainly considering the internal influencing factors of the mutual inductor, such as the influence of the error change trend and error fluctuation information on the measurement error of the mutual inductor, which can make the predicted target measurement error more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a schematic flowchart of an embodiment of the method for determining the measurement error of the mutual inductor provided by the present invention;

[0055] Figure 2 It is a schematic flowchart of an embodiment of the second model construction method provided by the present invention;

[0056] Figure 3 It is a schematic diagram of a prediction effect provided by the present invention;

[0057] Figure 4 It is a schematic structural diagram of an embodiment of the device for determining the measurement error of the mutual inductor provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.

[0059] In the description of the embodiments of the present invention, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0060] The descriptions such as "first" and "second" involved in the embodiments of the present invention are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Therefore, the technical features defined with "first" and "second" may explicitly or implicitly include at least one of such features.

[0061] Referring to "embodiment" in this article means that the specific features, structures or characteristics described in connection with the embodiment may be included in at least one embodiment of the present invention. The phrase appears in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0062] Before presenting the embodiments, the following terms are explained first:

[0063] Current transformer: An electrical device mainly used in a power system to convert voltage or current from one level to another, measure the current or voltage in a circuit system, protect the circuit, regulate the current or voltage, etc.

[0064] Secondary load of the current transformer: Refers to the total impedance of the electrical equipment or circuit connected to the secondary side of the current transformer. Connecting to the secondary side of the current transformer means connecting through the secondary winding of the current transformer.

[0065] Residual: The difference between the predicted value and the true value of the model.

[0066] The present invention provides a method, device, equipment and storage medium for determining the measurement error of a current transformer, which will be described separately below.

[0067] Refer to Figure 1 , which shows a schematic flowchart of an embodiment of the method for determining the measurement error of a current transformer provided by the present invention. The method includes:

[0068] Step S101, predicting the measurement error of the current transformer at the target time through a first model to obtain a first measurement error; the first model is constructed based on the change trend information of the first actual measurement error of the current transformer within the historical time.

[0069] Step S102, predicting the residual of the first measurement error through a second model to obtain a target residual; the second model is constructed based on the error fluctuation information of the first actual measurement error and the predicted residual of the first model.

[0070] Step S103, determining the target measurement error of the current transformer according to the first measurement error and the target residual.

[0071] The first model can be a measurement error prediction model of the current transformer. The first model can be a neural network model, such as a Bidirectional Long Short-Term Memory (BiLSTM) model. The first model can be trained based on the change trend information of the first actual measurement error of the current transformer within the historical time.

[0072] The first actual measurement error can be the actual measurement error of the current transformer measured within the historical time. For example, the historical time can be from the 0th day to the tth day; a first actual measurement error can be measured for each day from the 0th day to the tth day.

[0073] In one example, each date (each day) within the historical time can correspond to a first actual measurement error and can also correspond to multiple second actual measurement errors; the first actual measurement error corresponding to each date can be the average of the multiple second actual measurement errors corresponding to each date.

[0074] For example, for the t-th day, the true value of the target parameter and the measured value of the target parameter by the current transformer can be obtained once every hour. Since there are 24 hours in a day, a total of 24 groups of true values and measured values are obtained. Based on these 24 groups of true values and measured values, 24 second actual measurement errors can be calculated. The second actual measurement error can be the ratio error. Specifically, the calculation method of the second actual measurement error is shown in the following formula:

[0075] Second actual measurement error = (true value of the target parameter - measured value of the target parameter by the current transformer) / true value of the target parameter; where the target parameter includes current or voltage in the power system.

[0076] Then, by calculating the average value of these 24 second actual measurement errors, a first actual measurement error can be obtained, and this first actual measurement error is the first actual measurement error corresponding to the t-th day.

[0077] There are multiple first actual measurement errors. Based on each first actual measurement error, the change trend information of the first actual measurement error can be analyzed. The first actual measurement errors sorted in the order of measurement time can itself characterize its change trend information.

[0078] Alternatively, the change trend information of the first actual measurement error can also include: the first actual measurement error corresponding to the target date, and the average value of the first actual measurement errors within multiple time periods with the target date as the end date; the target date includes at least one. That is, the change trend information of the first actual measurement error is characterized by the first actual measurement error corresponding to the target date and the average value of the first actual measurement errors within multiple time periods with the target date as the end date.

[0079] For example, when the target date is the t-th day, the change trend information of the first actual measurement error can be characterized by the first actual measurement error on the t-th day, the average value of the first actual measurement errors from t - 4 to t days, the average value of the first actual measurement errors from t - 9 to t days, and the average value of the first actual measurement errors from t - 19 to t days.

[0080] In this example, the change trend information can be expressed by the following formula:

[0081] ; represents the average value of the first actual measurement errors from t - 4 to t days, represents the first actual measurement error on the i-th day;

[0082] ; represents the average value of the first actual measurement errors from t - 9 to t days;

[0083] ; represents the average value of the first actual measurement error from t - 19 to t days.

[0084] The target date can also include the (t + 1)-th day, the (t + 2)-th day, etc., and multiple time periods with the target date as the end date can also be set according to requirements.

[0085] This embodiment does not specifically limit the method for characterizing the change trend information of the first actual measurement error.

[0086] The second model can be a residual prediction model. Specifically, the second model can be a Support Vector Regression (SVR) model. The second model can be constructed based on the error fluctuation information of the first actual measurement error and the prediction residuals of the first model.

[0087] Based on each first actual measurement error, the error fluctuation information of the first actual measurement error can be analyzed. The error fluctuation information of the first actual measurement error can be represented in various ways such as the skewness, kurtosis, range, standard deviation, and variance of the first actual measurement error.

[0088] In one example, the error fluctuation information can include at least one of the highest error fluctuation, lowest error fluctuation, error fluctuation, relative size of error fluctuation, and relative position of error fluctuation corresponding to each date within the historical time; the error fluctuation information is calculated by the following formula:

[0089] , where represents the highest error fluctuation on the t -th day; represents the largest actual measurement error on the t -th day, determined according to multiple second actual measurement errors within the date corresponding to the t -th day, represents the first actual measurement error corresponding to the t -th day;

[0090] , where represents the lowest error fluctuation on the t -th day; represents the smallest actual measurement error on the t -th day, determined according to multiple second actual measurement errors within the date corresponding to the t -th day;

[0091] , where represents the tThe error fluctuation of the day, represents the first actual measurement error corresponding to the th day;

[0092] , where represents the relative magnitude of the error fluctuation on the t th day;

[0093] , where represents the relative position of the error fluctuation on the t th day.

[0094] This embodiment does not specifically limit the method for characterizing the error fluctuation information of the first actual measurement error.

[0095] After the first model is constructed based on the change trend information of the first actual measurement error of the mutual inductor within the historical time, the first model can predict the measurement error of the mutual inductor within the historical time to obtain at least one second measurement error. Each predicted second measurement error corresponds to a first actual measurement error; by calculating the difference between the second measurement error and the first actual measurement error corresponding to the second measurement error, the prediction residual corresponding to the second measurement error can be obtained. The second model is then constructed based on the error fluctuation information of the first actual measurement error and the prediction residual of the first model, that is, the second model is used to study the influence of the error fluctuation information on the prediction accuracy (prediction residual) of the first model and determine the mapping relationship between the error fluctuation information and the prediction residual of the first model.

[0096] For example, the first model can predict the measurement error of the mutual inductor on the (t + 1)-th day based on the following change trend information to obtain the second measurement error corresponding to the (t + 1)-th day . Then, combining the first actual measurement error corresponding to the (t + 1)-th day, the prediction residual on the (t + 1)-th day can be obtained. The change trend information is represented by the following matrix:

[0097] ,

[0098] where is the first actual measurement error on the t-th day, the t-th day is one day in the first historical time, L is the number of days, and represents the first actual measurement error of the L days before the t-th day.

[0099] The second model can then be used to study the influence of the error fluctuation information corresponding to the t-th day on the prediction residual on the (t + 1)-th day.

[0100] Predict the measurement error of the target time transformer through the first model to obtain the first measurement error, and predict the residual of the first measurement error through the second model to obtain the target residual. Then, the sum of the first measurement error and the target residual can be used as the target measurement error, that is, the finally determined measurement error of the transformer.

[0101] In the present invention, the first model predicts the first measurement error of the transformer at a certain future time (target time) based on the change trend information of the first actual measurement error. The second model predicts the residual of the first measurement error based on the mapping relationship between the error fluctuation information of the first actual measurement error and the prediction residual of the first model to obtain the target residual. Finally, the first measurement error is corrected according to the target residual to obtain the target measurement error. The present invention mainly considers the internal influencing factors of the transformer, such as the influence of the error change trend and error fluctuation information on the measurement error of the transformer, which can make the predicted target measurement error more accurate.

[0102] In one embodiment, the prediction residual of the first model can be obtained in the following manner:

[0103] (1) Obtain the historical error sequence of the transformer .

[0104] (2) Based on the historical error sequence and the change trend information, construct the matrix E:

[0105]

[0106] (3) Considering the different dimensions of the data, perform preprocessing such as normalization on the data in the matrix E to obtain the matrix .

[0107] (4) Use the CNN-BiLSTM model to perform short-term prediction on the matrix .

[0108] Perform multi-dimensional feature extraction through the Convolutional Neural Network (CNN): ; where .

[0109] The BiLSTM model passes through the gated unit (forget gate , input gate , output gate ) to read and modify the information of the memory unit according to the formula:

[0110]

[0111] Where , and The weight matrices for the forget gate, input gate, and output gate respectively; , , are the corresponding bias terms, is the weight matrix of the input cell state, is the bias term of the input cell state, is the Sigmoid activation function, which transforms the output into the interval [0, 1]; is the hyperbolic tangent activation function, which transforms the elements of the output into the interval [-1, 1].

[0112] Then, the output of the hidden layer is:

[0113]

[0114] Compared with the standard unidirectional LSTM, BiLSTM can obtain the correlation of historical and current information simultaneously, improving the prediction ability. The complete BiLSTM hidden elements provide the output concatenated vectors of the forward and backward processes, as follows:

[0115]

[0116] Input to get the output .

[0117] , where represents the predicted measurement error, and are the corresponding weight and bias terms of the model output respectively.

[0118] (5) Obtain the prediction residual.

[0119]

[0120] Among them, is the prediction residual on the (t + 1)-th day, is the second measurement error on the (t + 1)-th day, is the first actual measurement error on the (t + 1)-th day.

[0121] And so on, the first model is based on the historical error sequence , predicts to get , and then calculates to get .

[0122] (6) Construct the residual set .

[0123] Refer to Figure 2, showing a schematic flowchart of an embodiment of the second model construction method provided by the present invention. The second model can be constructed in the following manner:

[0124] Step S201, obtain environmental information in historical time; the environmental information includes at least one of temperature information, humidity information, and secondary load information of the mutual inductor.

[0125] Step S202, construct a first mapping relationship between environmental information and prediction residuals based on the first sample set.

[0126] Among them, the first sample set can be represented by the following formula:

[0127] , where represents the first sample set, respectively represent the environmental information corresponding to the (t + 1)-th day to the (t + n)-th day in historical time; respectively represent the prediction residuals obtained after the first model predicts the measurement errors of the mutual inductor for the (t + 1)-th day to the (t + n)-th day in historical time.

[0128] In one example, the environmental information corresponding to the (t + 1)-th day can be represented as: , represents the temperature data of the (t + 1)-th day, represents the humidity data of the (t + 1)-th day, represents the secondary load data of the mutual inductor on the (t + 1)-th day; the environmental information corresponding to the (t + 2)-th day to the (t + n)-th day ~ And so on.

[0129] The first mapping relationship can be represented by the following formula:

[0130] , where represents the first residual, represents the feature representation of the environmental information, represents the first bias, represents the first hyperplane weight vector, and are determined based on the first sample set.

[0131] In one example, and can be determined by minimizing the objective function in the first regression function, and the objective function in the first regression function is:

[0132] ;

[0133] ;

[0134] Among them, C1 represents the penalty factor in the first regression function, represents the slack variable in the first regression function, stipulates the error requirement of the first regression function, j = t + 1, t + 2, …, t + n, representing the number of samples.

[0135] Step S203, construct a second mapping relationship between the error fluctuation information and the prediction residual based on the second sample set.

[0136] The second sample set can be expressed by the following formula:

[0137] , where represents the second sample set, respectively represent the error fluctuation information corresponding to the t-th day to the (t + n - 1)-th day in the historical time.

[0138] In one example, the error fluctuation information corresponding to the t-th day can be expressed as: , the error fluctuation information corresponding to the (t + 1)-th day to the (t + n - 1)-th day ~ And so on.

[0139] The second mapping relationship is expressed by the following formula:

[0140] , where represents the second residual, represents the feature representation of the error fluctuation information, represents the second bias, represents the second hyperplane weight vector; and are determined based on the second sample set.

[0141] In one example, and can be determined by minimizing the objective function in the second regression function, and the objective function in the second regression function is:

[0142]

[0143]

[0144] Among them, C2 represents the penalty factor in the second regression function, represents the slack variable in the second regression function, stipulates the error requirement of the second regression function, j = t + 1, t + 2, …, t + n, representing the number of samples.

[0145] Step S204: Construct a second model according to the first mapping relationship and the second mapping relationship.

[0146] In this embodiment, on the one hand, a first mapping relationship between environmental information and prediction residuals is constructed, and the first mapping relationship characterizes the influence of environmental information on the prediction residuals of the first model. On the other hand, a second mapping relationship between error fluctuation information and prediction residuals is constructed, and the second mapping relationship characterizes the influence of error fluctuation information on the prediction residuals of the first model. Finally, a second model is constructed according to the first mapping relationship and the second mapping relationship, and the second model can comprehensively evaluate the influence of environmental information and error fluctuation information on the prediction residuals of the first model.

[0147] In this embodiment, step S102 may specifically include: determining a first residual of the first measurement error through the first mapping relationship; determining a second residual of the first measurement error through the second mapping relationship; and determining a target residual according to the first residual and the second residual.

[0148] In one embodiment, the step of determining the target residual according to the first residual and the second residual may specifically include: determining a first weight of the first residual and a second weight of the second residual; and taking the sum of the product of the first residual and the first weight and the product of the second residual and the second weight as the target residual.

[0149] In one embodiment, the step of determining the first weight of the first residual and the second weight of the second residual may specifically include: determining a first sub-prediction residual corresponding to each prediction residual according to the first mapping relationship; determining a first credibility and a first unbelievability of the first mapping relationship according to each prediction residual and the first sub-prediction residual; and determining the first weight of the first residual according to the first credibility and the first unbelievability. And it includes: determining a second sub-prediction residual corresponding to each prediction residual according to the second mapping relationship; determining a second credibility and a second unbelievability of the second mapping relationship according to each prediction residual and the second sub-prediction residual; and determining the second weight of the second residual according to the second credibility and the second unbelievability. In this embodiment, the prediction residual may be the model prediction residual when the first model determined according to steps (1) to (6) in the above embodiment predicts the measurement error of the current transformer from the (t + 1)-th day to the (t + n)-th day, and the corresponding first sub-prediction residual can be understood as: the model prediction residual when the first model determined according to the first mapping relationship predicts the measurement error of the current transformer from the (t + 1)-th day to the (t + n)-th day;

[0150] In this embodiment, it may be the model prediction residual when the first model determined according to steps (1) to (6) in the above embodiment predicts the measurement error of the current transformer from the (t + 1)-th day to the (t + n)-th day, and the corresponding first sub-prediction residual can be understood as: the model prediction residual when the first model determined according to the first mapping relationship predicts the measurement error of the current transformer from the (t + 1)-th day to the (t + n)-th day; corresponding to the first sub-prediction residual can be understood as: the model prediction residual when the first model determined according to the first mapping relationship predicts the measurement error of the current transformer from the (t + 1)-th day to the (t + n)-th day: The corresponding second sub-prediction residual can be understood as the model prediction residual when the first model determines according to the second mapping relationship to predict the measurement error of the mutual inductor from the (t + 1)-th day to the (t + n)-th day.

[0151] According to each prediction residual and each first sub-prediction residual, the first credibility and the first incredibility of the first mapping relationship can be determined. According to each prediction residual and each second sub-prediction residual, the second credibility and the second incredibility of the second mapping relationship can be determined. Then, based on the credibility and incredibility of the first mapping relationship and the credibility and incredibility of the second mapping relationship, weight allocation is performed on the first relationship and the second mapping relationship, and the result fusion of the two mapping relationships is completed.

[0152] In this embodiment, the first weight and the second weight can be specifically determined in the following manner:

[0153] (1) Determine the credibility of the first mapping relationship and the credibility of the second mapping relationship.

[0154] The first sub-prediction residual and the prediction residual The distance between them:

[0155] , where is the first sub-prediction residual of the j-th sample, is the prediction residual of the j-th sample.

[0156] The first sub-prediction residual The similarity of:

[0157] The first sub-prediction residual The credibility of (the credibility of the first mapping relationship):

[0158] Similarly, the credibility of the second mapping relationship , is the second sub-prediction residual.

[0159] (2) Normalize the two credibilities respectively.

[0160]

[0161]

[0162] (3) Determine the incredibility of the first mapping relationship and the incredibility of the second mapping relationship.

[0163]

[0164] , is the second sub-prediction residual of the j-th sample.

[0165] (4) Normalize the two uncredibilities respectively.

[0166]

[0167]

[0168] (5) Determine the first weight and the second weight according to the two credibilities and the two uncredibilities after the above normalization processing.

[0169] , , is the first weight.

[0170] , , is the second weight.

[0171] After determining the first weight and the second weight, the residual of the measurement error predicted by the first model can be predicted by the second model.

[0172] For example, for the (t + 1)-th day, the residual corresponding to the (t + 1)-th day is:

[0173]

[0174] The finally determined measurement error of the mutual inductor on the (t + 1)-th day is:

[0175]

[0176] Table 1 Comparison table of actual measurement error and model measurement error

[0177]

[0178] Referring to Table 1, the comparison result of the actual measurement error of the mutual inductor obtained by testing the present invention and the measurement error predicted by the model is shown. The measurement error predicted by the model in Table 1 refers to the target measurement error obtained by integrating the prediction results of the first model and the second model. The difference between the two refers to the difference between the actual measurement error of the mutual inductor and the measurement error predicted by the model. According to Table 1, it can be seen that the measurement error predicted based on the present invention has a small deviation from the actual measurement error of the mutual inductor.

[0179] Referring to Figure 3 , a schematic diagram of the prediction effect provided by the present invention is shown. Figure 3 In which the vertical coordinate is the error value and the horizontal coordinate is the sample serial number, Figure 3The measurement error predicted by the intermediate model refers to the target measurement error obtained by integrating the prediction results of the first model and the second model. According to Figure 3 it can be seen that the measurement error predicted based on the present invention has a small deviation from the actual measurement error of the mutual inductor.

[0180] Referring to Figure 4 , a structural schematic diagram of an embodiment of the mutual inductor measurement error determination device provided by the present invention is shown. The device 300 includes:

[0181] A first measurement error prediction module 301, configured to predict the measurement error of the target-time mutual inductor through a first model to obtain a first measurement error; the first model is constructed based on the change trend information of the first actual measurement error of the mutual inductor within the historical time;

[0182] A target residual prediction module 302, configured to predict the residual of the first measurement error through a second model to obtain a target residual; the second model is constructed based on the error fluctuation information of the first actual measurement error and the prediction residual of the first model;

[0183] A second measurement error determination module 303, configured to determine the target measurement error of the mutual inductor according to the first measurement error and the target residual.

[0184] It should be noted that: the implementation principle or implementation process of the above modules can refer to the embodiments of the foregoing mutual inductor measurement error determination method, and will not be elaborated here one by one.

[0185] In one embodiment, the present invention further provides an electronic device, including a memory and a processor. A computer program is stored on the memory, and when the computer program is executed by the processor, the steps of any one of the above-mentioned mutual inductor measurement error determination methods are implemented.

[0186] In one embodiment, the present invention further provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by the processor, the steps of any one of the above-mentioned mutual inductor measurement error determination methods are implemented.

[0187] Those skilled in the art can understand that all or part of the processes of implementing the above embodiment methods can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory, etc.

[0188] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for determining the measurement error of an instrument transformer, characterized in that, Including: Predicting the measurement error of the target time transformer through a first model to obtain a first measurement error; the first model is constructed based on the change trend information of the first actual measurement error of the transformer within a historical time; Predicting the residual of the first measurement error through a second model to obtain a target residual; The second model is constructed based on the error fluctuation information of the first actual measurement error and the predicted residual of the first model; Determining the target measurement error of the transformer according to the first measurement error and the target residual; Wherein, the second model is constructed in the following manner: Obtaining the environmental information in the historical time; the environmental information includes at least one of temperature information, humidity information, and secondary load information of the transformer; Constructing a first mapping relationship between the environmental information and the predicted residual based on a first sample set; The first sample set is represented by the following formula: , where represents the first sample set, respectively represent the environmental information corresponding to the (t + 1)-th day to the (t + n)-th day in the historical time; respectively represent the prediction residuals obtained after the first model predicts the measurement errors of the mutual inductor for the (t + 1)-th day to the (t + n)-th day in the historical time; The first mapping relationship is represented by the following formula: , where represents the first residual, represents the feature representation of the environmental information, represents the first bias, represents the first hyperplane weight vector, and is determined based on the first sample set; Constructing a second mapping relationship between the error fluctuation information and the predicted residual based on a second sample set; The second sample set is represented by the following formula: , where represents the second sample set, respectively represent the error fluctuation information corresponding to the t-th day to the (t + n - 1)-th day in the historical time; The second mapping relationship is represented by the following formula: , where represents the second residual, represents the feature representation of the error fluctuation information, represents the second offset, represents the second hyperplane weight vector; and is determined based on the second sample set; Constructing a second model according to the first mapping relationship and the second mapping relationship; The predicting the residual of the first measurement error through the second model to obtain a target residual includes: Determining a first residual of the first measurement error through the first mapping relationship; Determining a second residual of the first measurement error through the second mapping relationship; Determining a target residual according to the first residual and the second residual.

2. The method for determining the measurement error of the mutual inductor according to claim 1, characterized in that, Each date within the historical time corresponds to a first actual measurement error and multiple second actual measurement errors; the first actual measurement error corresponding to each date within the historical time is the average value of the multiple second actual measurement errors corresponding to each date; And, the change trend information of the first actual measurement error includes: the first actual measurement error corresponding to the target date, and the average value of the first actual measurement error within multiple time periods with the target date as the end date; The target date includes at least one.

3. The method for determining the measurement error of the mutual inductor according to claim 2, wherein The error fluctuation information of the first actual measurement error includes at least one of the highest error fluctuation, the lowest error fluctuation, the error fluctuation, the relative size of the error fluctuation, and the relative position of the error fluctuation corresponding to each date within the historical time; The error fluctuation information is calculated through the following formula: , where represents the maximum error fluctuation on the t th day; represents the maximum actual measurement error on the t th day, which is determined according to multiple said second actual measurement errors within the date corresponding to the t th day; represents the first actual measurement error corresponding to the t th day. , where represents the minimum error fluctuation on the t th day; represents the minimum actual measurement error on the t th day, determined according to multiple said second actual measurement errors within the date corresponding to the t th day. , where represents the error fluctuation on the t th day, represents the first actual measurement error corresponding to the th day; , where represents the relative magnitude of the error fluctuation on the t th day; , where represents the relative position of the error fluctuation on the t th day.

4. The method for determining the measurement error of the mutual inductor according to claim 1, characterized in that, The determining the target residual according to the first residual and the second residual includes: Determining a first weight of the first residual and a second weight of the second residual; Taking the sum of the product of the first residual and the first weight and the product of the second residual and the second weight as the target residual.

5. The method for determining the measurement error of the mutual inductor according to claim 4, wherein The determining the first weight of the first residual and the second weight of the second residual includes: Determining a first sub-predicted residual corresponding to the time of each predicted residual according to the first mapping relationship; Determining a first credibility and a first incredibility of the first mapping relationship according to the predicted residual and the first sub-predicted residual; Determine a first weight of the first residual according to the first credibility and the first incredibility; Determine a second sub-prediction residual corresponding to each prediction residual time according to the second mapping relationship; Determine a second credibility and a second incredibility of the second mapping relationship according to the prediction residual and the second sub-prediction residual; Determine a second weight of the second residual according to the second credibility and the second incredibility.

6. A device for determining the measurement error of an instrument transformer, characterized in that, It includes: A first measurement error prediction module, configured to predict a measurement error of a target time transformer through a first model to obtain a first measurement error; the first model is constructed based on change trend information of the first actual measurement error of the transformer within a historical time; A target residual prediction module, configured to predict a residual of the first measurement error through a second model to obtain a target residual; The second model is constructed based on error fluctuation information of the first actual measurement error and a prediction residual of the first model; A second measurement error determination module, configured to determine a target measurement error of the transformer according to the first measurement error and the target residual; Wherein, the second model is constructed in the following manner: Obtain environmental information in the historical time; the environmental information includes at least one of temperature information, humidity information, and secondary load information of the transformer; Construct a first mapping relationship between the environmental information and the prediction residual based on a first sample set; The first sample set is represented by the following formula: , where represents the first sample set, respectively represent the environmental information corresponding to the (t + 1)-th day to the (t + n)-th day in historical time; respectively represent the prediction residuals obtained after the first model predicts the measurement errors of the mutual inductor for the (t + 1)-th day to the (t + n)-th day in historical time; The first mapping relationship is represented by the following formula: , where represents the first residual represents the feature representation of the environmental information represents the first offset represents the first hyperplane weight vector and are determined based on the first sample set Construct a second mapping relationship between the error fluctuation information and the prediction residual based on a second sample set; The second sample set is represented by the following formula: , where represents the second sample set, respectively represent the error fluctuation information corresponding to the t-th day to the (t + n - 1)-th day in the historical time; The second mapping relationship is represented by the following formula: , where represents the second residual, represents the feature representation of the error fluctuation information, represents the second bias, represents the second hyperplane weight vector; and is determined based on the second sample set; Construct a second model according to the first mapping relationship and the second mapping relationship; The target residual prediction module is specifically configured to: Determine a first residual of the first measurement error through the first mapping relationship; Determine a second residual of the first measurement error through the second mapping relationship; Determine a target residual according to the first residual and the second residual.

7. An electronic device, characterized in that, It includes a memory and a processor, wherein, The memory is used to store a program; The processor is coupled to the memory and is configured to execute the program stored in the memory to implement the steps in the method for determining the measurement error of the transformer according to any one of claims 1 to 5 above.

8. A computer-readable storage medium, characterized in that, For storing a computer-readable program, when the program is executed by a processor, it can implement the steps in the method for determining the measurement error of the transformer according to any one of claims 1 to 5 above.

Images