Power transmission and transformation equipment data filling method and system based on time sequence prediction

Abstract

The invention relates to the technical field of power systems and digitization thereof, in particular to a power transmission and transformation equipment data filling method and system based on time sequence prediction, missing value inference is carried out through a least square support vector machine and a dynamic Bayesian network, and missing data in a complete time sequence curve is replaced with an existing data rule; strong robustness is shown when long-time data missing or sudden abnormity is processed, and model distortion caused by data blank can be effectively avoided; through continuous prediction of a time sequence and filling of blank data, the digital twin model maintains real-time and accurate reflection of the operation state of the power transmission and transformation equipment; therefore, more reliable support is provided for fault diagnosis, state monitoring and predictive maintenance of the equipment, so that the operation efficiency and safety of the power transmission and transformation equipment are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems and digitalization thereof, and in particular to a method and system for filling data of power transmission and transformation equipment based on time series prediction. Background Art

[0002] At present, the digital twin technology of power transmission and transformation equipment is a comprehensive application of information technologies such as physical perception, numerical calculation, and simulation. By establishing virtual entities corresponding to physical entities, the digital description of physical entities is realized, and the monitoring, diagnosis, and prediction of physical entities are completed with the help of data analysis, modeling and other technologies. The application of this technology in power transmission and transformation equipment can establish a connection between physical and information space and realize the interactive mapping between the entity and the twin of power transmission and transformation equipment. However, the application of digital twin technology in power transmission and transformation equipment is still in the theoretical research stage, but it provides a new technical path for the full life cycle management of power transmission and transformation equipment. The directions for improvement include standard formulation, model fusion and data mining.

[0003] The application of existing time series prediction digital twin technology in power transmission and transformation equipment faces some technical problems, including insufficient understanding of the physical laws of defects and faults, obvious differences in the quality of perceived data leading to the inability to effectively integrate data, the accuracy of existing power transmission and transformation equipment status evaluation models needs to be improved and the model's scalability is poor, and the existing level of intelligence is unable to achieve the function of auxiliary judgment. Summary of the invention

[0004] In view of the above problems existing in the prior art, the present invention is proposed.

[0005] Therefore, the technical problem to be solved by the present invention is: to solve the problem of data gaps that may occur due to sensor failure, data transmission interruption or data loss during the construction of digital twins of power transmission and transformation equipment. The method based on time series prediction can fill these data gaps, and effectively prevent equipment failures and predict maintenance needs by analyzing the trend of equipment performance indicators changing over time, periodic feature recognition and anomaly detection. This method ensures that the digital twin model can accurately reflect the state of the physical entity. To ensure the continuity and integrity of the digital twin model data.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a method for filling data of power transmission and transformation equipment based on time series prediction, comprising: obtaining a multivariate time series by collecting and analyzing historical operation data of the power transmission and transformation equipment; performing missing value detection in the multivariate time series to determine the data area to be filled; filling the missing data; inferring the missing values, and optimizing the parameters of the filling algorithm according to the verification results.

[0007] As a preferred solution of the method for filling data of power transmission and transformation equipment based on time series prediction described in the present invention, the multivariate time series includes observations arriving in time series, which are expressed as:

[0008] x i (t), (i=1, 2,..., m; t=1, 2,..., n)

[0009] Here, i indexes the measurement value at time point t[1]; when i=1, it is a univariate time series, and when i≥2, it is a multivariate time series.

[0010] As a preferred solution of the method for filling data of power transmission and transformation equipment based on time series prediction described in the present invention, the missing value detection includes obtaining an item from an incomplete subset and comparing it with the item from a complete subset; replacing the missing value with the corresponding value in the complete item, calculating the dynamic time warping distance of the two time series curves, finding the minimum distance between the entire complete subset, and replacing the missing value with the corresponding value in the optimal sequence curve.

[0011] As a preferred solution of the data filling method for power transmission and transformation equipment based on time series prediction described in the present invention, wherein: the determination of the filling data area includes a multivariate filling method, if there is missing data in the i-th variable dimension of sequence S1, a similar sequence of S1 is searched in the multivariate time series data set, and if the search result is S1, S2, ..., S n , then it is regarded as the training set D={(x k ,y k )}; x k ∈R m-1 represents the input vector, which consists of (m-1) variables from dimensions S2, S3, …, S except the i-th variable. n The data composition of the time point; k ∈R represents the output vector, which is composed of S1, S2, …, S n , which consists of data from each time point of the i-th variable dimension.

[0012] As a preferred solution of the method for filling data of power transmission and transformation equipment based on time series prediction described in the present invention, wherein: the filling of missing data includes training LSSVM using a training set, and the function x in the original space k →y k Converts to:

[0013]

[0014] in, represents the minimization of ω, b and e, ω represents the weight vector, b represents the bias term, e represents the error term, J(ω, e) represents the objective function, ωT ω represents the regularization term, where ω T represents the transpose of the weight vector ω, ω T ω represents the sum of squares of the weight vector, represents the sum of squares of the error terms, represents the weight vector, y represents the regularization parameter, represents the kernel space mapping function; e k ∈R represents the error and b represents the deviation.

[0015] As a preferred solution of the data filling method for power transmission and transformation equipment based on time series prediction described in the present invention, wherein: the inference of missing values ​​includes using a dynamic Bayesian network to infer missing values. If two curves x and Y, x i A set of points representing a curve x, y j represents a set of points on the curve Y, then the Euclidean distance d(X, Y) is expressed as:

[0016]

[0017] The calculation of dynamic time warping distance is expressed as: if the time series curves x and Y, x = <x1,x2,...,x i , ..., x n >,Y= <y1,y2,...,y j , ..., y m >, n represents the length of x, m represents the length of Y; the minimum dynamic time warping distance is expressed as:

[0018]

[0019] Among them, d k Represents the distance between elements; the DTW distance is calculated by dynamic programming of the cumulative matrix, expressed as:

[0020]

[0021] Among them, r(i, j) represents the cumulative distance, d(x i ,y j ) represents the distance between the i-th element of time series X and the j-th element of time series Y, p represents the penalty term, and min represents the function of taking the minimum value.

[0022] As a preferred solution of the method for filling data of power transmission and transformation equipment based on time series prediction described in the present invention, the parameters of the optimization filling algorithm include solving the optimization problem by constructing the Lagrangian equation, which is expressed as:

[0023]

[0024] Among them, L(ω, b, e, α) represents the Lagrangian function, α represents the Lagrangian multiplier, α k ∈R represents the Lagrange multiplier, k represents the data point, J(ω, e) represents the original objective function, Expresses the difference between predicted and actual values;

[0025] If L to ω, b, e k , α k The partial derivative of is equal to 0, and ω, e k Elimination, expressed as:

[0026]

[0027] in, represents a v×v matrix, y=[y1,y2,...,y N ] T represents the regularization function, α=[α1,α2,…,α N ] T represents the Lagrange multiplier vector, 1 v =[1, 1, ..., 1] T represents a column vector of dimension v;

[0028] According to the Mercer condition, there exists a mapping function The relationship between it and the kernel function K(·,·) is expressed as:

[0029]

[0030] The estimation function of the least squares support vector machine is expressed as:

[0031]

[0032] Another object of the present invention is to provide a data filling system for power transmission and transformation equipment based on time series prediction, which obtains characteristic values ​​of data judgment from past time series through modules such as data collection, missing value detection, data filling and filling verification, and through time series prediction methods. According to the curve similarity principle, the more similar the corresponding features in two curves are, the more similar the two lines are. By using the correlation between variables in multivariate time series, the incomplete time series curve is compared with all complete curves in the data set, and then the missing data in the complete time series curve is replaced by the missing value inference of the least squares support vector machine and the dynamic Bayesian network.

[0033] In order to solve the above technical problems, the present invention provides the following technical solutions: a data filling system for power transmission and transformation equipment based on time series prediction, comprising: a data collection module, a missing detection module, a data filling module and a filling verification module;

[0034] The data collection module obtains a multivariate time series by collecting and analyzing historical operation data of power transmission and transformation equipment;

[0035] The missing value detection module performs missing value detection on the multivariate time series and determines the filled data area;

[0036] The data filling module fills in the missing data;

[0037] The filling verification module verifies the filled data, evaluates the error between the filled data and the actual data, and optimizes the parameters of the filling algorithm according to the verification results.

[0038] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the method for filling data of power transmission and transformation equipment based on time series prediction as described above are implemented.

[0039] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the method for filling data of power transmission and transformation equipment based on time series prediction as described above are implemented.

[0040] The beneficial effects of the present invention are as follows: the gaps in the data of power transmission and transformation equipment are filled by a method based on time series prediction, ensuring the continuity and integrity of the digital twin model data. The use of LSSVM and dynamic Bayesian network for missing value inference enhances the robustness of the model in dealing with long-term data missing or sudden anomalies, and avoids model distortion caused by data gaps. Through the analysis of the changing trends of equipment performance indicators and anomaly detection, the digital twin model can more accurately reflect the state of the physical entity, provide reliable support for equipment fault diagnosis and predictive maintenance, and optimize the operating efficiency and safety of power transmission and transformation equipment. This method provides technical support for equipment status monitoring, fault diagnosis and predictive maintenance, and promotes a new path for the full life cycle management of power transmission and transformation equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0042] Figure 1 A flowchart of a method for filling data of power transmission and transformation equipment based on time series prediction provided by one embodiment of the present invention.

[0043] Figure 2 A flowchart of a missing algorithm for a method for filling data of power transmission and transformation equipment based on time series prediction provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0045] Example 1, reference Figure 1-Figure 2 , is an embodiment of the present invention, which provides a method for filling data of power transmission and transformation equipment based on time series prediction, comprising:

[0046] S1: By collecting and analyzing the historical operation data of power transmission and transformation equipment, a multivariate time series is obtained.

[0047] It should be noted that if Figure 1 As shown in S1, the time series includes observations that arrive in time sequence, expressed as:

[0048] x i (t), (i=1,2,…,m; t=1,2,…,n)

[0049] Where i indexes the measurement value at time point t[1]; when i = 1, it is a univariate time series, and when i ≥ 2, it is a multivariate time series

[0050] Furthermore, due to various interference factors, data loss in multivariate time series is inevitable, which will affect data processing and analysis. Therefore, a method to fill missing data in multivariate time series is needed.

[0051] Furthermore, the multivariate autoregressive model is a classic time series analysis method, and the method based on functional relationships is used to fill in missing data; the data set can be regarded as an information system, and the data in a set of data usually have the same change pattern; the algorithm can mine reasonable values ​​by analyzing the inherent principles of the data set; LSSVM introduces the least squares linear system into the vector machine, which transforms the quadratic programming problem into a linear equation, which helps LSSVM effectively improve the calculation speed and generalization ability.

[0052] S2: Perform missing value detection in multivariate time series and determine the area of ​​filled data.

[0053] It should be noted that if Figure 1As shown in S2, an item is taken from the incomplete subset and compared with the item from the complete subset; the missing value is replaced with the corresponding value in the complete item, and the dynamic time warping distance of the two time series curves is calculated. In this way, the minimum distance between the entire complete subset is found, and the missing value is replaced with the corresponding value in the best series curve; the algorithm flow is as shown in the flow Figure 2 shown.

[0054] Furthermore, there is correlation between the variables in the multivariate time series, so the missing data in the variable dimension can be filled with the data of other related variables; without loss of generality, in the present invention, it is assumed that there is missing data in the i-th variable dimension of sequence S1.

[0055] Furthermore, based on the search for similar sequences of S1 in the multivariate time series dataset, if the search results are S1, S2, …, S n , then it is regarded as the training set D={(x k ,y k )}; x k ∈R m-1 represents the input vector, which consists of (m-1) variables from dimensions S2, S3, …, S except the i-th variable. n The data composition of the time point; k ∈R represents the output vector, which is composed of S1, S2, …, S n , which consists of data from each time point of the i-th variable dimension.

[0056] S3: Fill in missing data.

[0057] It should be noted that if Figure 1 As shown in S3, filling the missing data includes training the LSSVM using the training set, and the function x in the original space k →y k Converts to:

[0058]

[0059] in, represents the minimization of ω, b and e, ω represents the weight vector, b represents the bias term, e represents the error term, J(ω,e) represents the objective function, which is used to measure the performance of the model and needs to be minimized, ω T ω represents the regularization term, where ω T represents the transpose of the weight vector ω, ω T ω represents the sum of squares of the weight vector, Represents the sum of squares of error terms, which is used to control model complexity and avoid overfitting. represents the weight vector, y represents the regularization parameter, represents the kernel space mapping function; e k ∈R represents the error and b represents the deviation.

[0060] S4: Infer the missing values ​​and optimize the parameters of the filling algorithm based on the verification results.

[0061] It should be noted that if Figure 1 As shown in S4, the missing value inference includes using a dynamic Bayesian network to infer the missing value. If two curves X and Y, x i A set of points representing the curve X, y j represents a set of points on the curve Y, then the Euclidean distance d(X, Y) is expressed as:

[0062]

[0063] Dynamic time wrapping is applicable to complex curves in similarity measurement. In dynamic time warping distance (DTW), it needs to find the nearest map path, allow one-to-many correspondence in curve pairs, allow curves to have different lengths, and insert new points when necessary to ensure orderliness; the calculation of dynamic time warping distance is expressed as: if the time series curves X and Y, X = <x1,x2,...,x i , ..., x n >,Y= <y1,y2,...,y j , ..., y m >, n represents the length of X, m represents the length of Y; construct an n×m matrix m, and the two curves have a point-to-point relationship. The matrix can ensure that the package path grows at any time; it can be seen from the present invention that too many paths can meet the needs of the conditions, and how to choose the smallest path is the key; the minimum dynamic time warping distance is expressed as:

[0064]

[0065] Among them, d k It means d(x i ,y j ) is the element-to-element distance, which is the distance between elements. Different distance measurements can be selected, such as the Euclidean distance. The DTW distance is calculated by dynamic programming of the accumulation matrix, expressed as:

[0066]

[0067] Among them, r(i, j) represents the cumulative distance, d(x i ,y j) represents the distance or difference measure between the i-th element of time series X and the j-th element of time series Y. This distance can be the Euclidean distance, the absolute value difference, or any other suitable measure. p represents the penalty term, which is used to handle the insertion or deletion operations in the time series. In the standard DTW algorithm, this penalty term is 0, but in some variants, a positive penalty value may be added to control the degree of curvature of the time series. min represents the minimum function, which is used to select the smallest one from the three cumulative distances to determine the cumulative distance of the current element pair.

[0068] Furthermore, optimizing the parameters of the filling algorithm involves solving the optimization problem by constructing the Lagrangian equation, which is expressed as:

[0069]

[0070] Among them, L(ω, b, e, α) represents the Lagrangian function, which is an extension of the original optimization problem and contains α, which represents the Lagrangian multiplier. k ∈R represents the Lagrange multiplier corresponding to the constraint condition for each data point k, k represents the data point, J(ω, e) represents the original objective function that needs to be minimized, Represents the difference between the predicted value and the actual value; after multiplying by the Lagrangian multiplier α, it forms part of the Lagrangian function;

[0071] If L to ω, b, e k , α k The partial derivative of is equal to 0, and ω, e k Elimination, expressed as:

[0072]

[0073] in, Represents a v×v matrix, usually representing the similarity or kernel matrix between data points, y=[y1,y2,…,y N ] T represents the regularization function, which is used to control the trade-off between model complexity and fitting error, α = [α1, α2, ..., α N ] T represents the Lagrange multiplier vector corresponding to each data point, 1 v =[1, 1, ..., 1] T represents a v-dimensional column vector, all elements of which are 1;

[0074] According to Mercer's condition, any semi-positive definite function can be used as a kernel function, and there exists a mapping function The relationship between and kernel function K(·, ·) is expressed as:

[0075]

[0076] The estimation function of the least squares support vector machine is expressed as:

[0077]

[0078] a and b can be obtained from the Lagrange multiplier matrix. Finally, the present invention uses the trained LSSVM model to fill in the missing data of S1; at each time point of S1, the present invention uses the data of other (m-1) variable dimensions as input values, and the output is the filled data of S1.

[0079] Furthermore, in the LSSVM model, a kernel function is used to project samples into a high-dimensional space, converting nonlinear problems into linear problems; since the radial basis function has good performance and has only one shape parameter, the present invention uses RBF as the kernel function; the LSSVM model contains two parameters: a regularization parameter y and a shape parameter σ, y represents the weight of the fitting error, the larger y is, the closer the fitting function is to the sampling point; σ affects the smoothness of the fitting curve, the larger σ is, the smoother the fitting curve is, but the fitting effect may not be good; in order to obtain a better filling effect, the parameters should be reasonably selected in the process of using the present invention to ensure that the fitting curve is smooth and the fitting error is small.

[0080] The above is a schematic scheme of a method for filling data of power transmission and transformation equipment based on time series prediction in this embodiment. It should be noted that the technical scheme of the system of the method for filling data of power transmission and transformation equipment based on time series prediction and the technical scheme of the method for filling data of power transmission and transformation equipment based on time series prediction mentioned above belong to the same concept. For the details not described in detail in the technical scheme of the system for filling data of power transmission and transformation equipment based on time series prediction in this embodiment, please refer to the description of the technical scheme of the method for filling data of power transmission and transformation equipment based on time series prediction mentioned above.

[0081] Embodiment 2 is an embodiment of the present invention, which provides a method for filling data of power transmission and transformation equipment based on time series prediction. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0082] 680 patient data were collected in this test. First, we deleted some elements from the data set to artificially form an incomplete data set. The data set was divided into a complete data subset (Sc) and an incomplete data subset (Si), and then the average distance, pattern distance and our method were tested. Our experimental environment is dell760, core2 dual-core e8500CPU, 2GB memory. The software uses MATLAB V7.5. We programmed and did some experiments, using three methods to fill in missing values, and then compared the filled data with the actual values, compared the filling effects of different methods, and used the average relative error e as the basis for comparison. The error calculation formula is as follows:

[0083]

[0084] e represents the normalized error, which is an indicator of the difference between the predicted value and the actual value. Where n is the number of missing data, S t ,r t Represent the tth filled data and the actual data respectively. It can be seen from the table that the method proposed in the invention shows the best filling result. In order to further verify the robustness of the algorithm, some values ​​are deleted from the previous section under the condition of a data set given in the power grid system to make it an incomplete data set. After each filling, it is compared with the real data in the present invention. The algorithm is effective when the size of the incomplete data set does not exceed the complete data set.

[0085] The results are shown in Table 1 below:

[0086] Table 1 Data results

[0087] Least Squares Method Traditional Bayesian Methods Patented method All data 100 100 100 The right data 36 32 68 Correct proportions 36％ 32％ 68％ deviation 47.40％ 52.32％ 25.64％

[0088] Embodiment 3 is an embodiment of the present invention, which provides a data filling system for power transmission and transformation equipment based on time series prediction, including: a data collection module, a missing detection module, a data filling module and a filling verification module;

[0089] The data collection module obtains a multivariate time series by collecting and analyzing historical operation data of power transmission and transformation equipment;

[0090] The missing value detection module performs missing value detection on the multivariate time series and determines the filled data area;

[0091] The data filling module fills in the missing data;

[0092] The filling verification module verifies the filled data, evaluates the error between the filled data and the actual data, and optimizes the parameters of the filling algorithm according to the verification results.

[0093] This embodiment further provides a computing device, which is applicable to the case of a method for filling data of power transmission and transformation equipment based on time series prediction, and includes:

[0094] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the data filling method for power transmission and transformation equipment based on time series prediction as proposed in the above embodiment.

[0095] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for filling data of power transmission and transformation equipment based on time series prediction as proposed in the above embodiment is implemented.

[0096] The storage medium proposed in this embodiment and the power transmission and transformation equipment data filling method based on time series prediction proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0097] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0098] Logic and / or steps otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch instructions from and execute instructions on, an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" may be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0099] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0100] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for filling data of power transmission and transformation equipment based on time series prediction, characterized in that: include: By collecting and analyzing the historical operation data of power transmission and transformation equipment, multivariate time series are obtained; Perform missing value detection on multivariate time series and determine the area of ​​filled data; Fill in missing data; Inference is performed on missing values ​​and the parameters of the filling algorithm are optimized based on the validation results.

2. The method for filling data of power transmission and transformation equipment based on time series prediction according to claim 1, characterized in that: The multivariate time series consists of observations that arrive in time sequence, expressed as: x i (t),(i=1,2,…,m;t=1,2,…,n) Here, i indexes the measurement value at time point t[1]; when i=1, it is a univariate time series, and when i≥2, it is a multivariate time series.

3. The method for filling data of power transmission and transformation equipment based on time series prediction according to claim 2, characterized in that: The missing value detection includes obtaining an item from the incomplete subset and comparing it with the item from the complete subset; replacing the missing value with the corresponding value in the complete item, calculating the dynamic time warping distance of the two time series curves, finding the minimum distance between the entire complete subset, and replacing the missing value with the corresponding value in the best series curve.

4. The method for filling data of power transmission and transformation equipment based on time series prediction according to claim 3, characterized in that: The method of determining the filled data region includes a multivariate filling method. If there is missing data in the i-th variable dimension of sequence S1, a similar sequence of S1 is searched in the multivariate time series data set. If the search result is S1, S2, ..., S n , then it is regarded as the training set D={(x k ,y k )}; x k ∈R m-1 represents the input vector, which consists of (m-1) variables from dimensions S2, S3, …, S except the i-th variable. n The data composition of the time point; k ∈R represents the output vector, which is composed of S1, S2, …, S n , which consists of data from each time point of the i-th variable dimension.

5. The method for filling data of power transmission and transformation equipment based on time series prediction according to claim 4, characterized in that: The filling of missing data includes training LSSVM using a training set, and the function x in the original space k →y k Converts to: in, represents the minimization of ω, b and e, ω represents the weight vector, b represents the bias term, e represents the error term, J(ω,e) represents the objective function, ω T ω represents the regularization term, where ω T represents the transpose of the weight vector ω, ω T ω represents the sum of squares of the weight vector, represents the sum of squares of the error terms, represents the weight vector, y represents the regularization parameter, represents the kernel space mapping function; e k ∈R represents the error and b represents the deviation.

6. The method for filling data of power transmission and transformation equipment based on time series prediction according to claim 5, characterized in that: The missing value inference includes using a dynamic Bayesian network to infer the missing value. If two curves X and Y, x i A set of points representing the curve X, y j represents a set of points on the curve Y, then the Euclidean distance d(X,Y) is expressed as: The calculation of dynamic time warping distance is expressed as: if the time series curves X and Y, X = <x1,x2,…,x i ,…,x n >,Y= <y1,y2,…,y j ,…,y m >, n represents the length of X, m represents the length of Y; the minimum dynamic time warping distance is expressed as: Among them, d k Represents the distance between elements; the DTW distance is calculated by dynamic programming of the cumulative matrix, expressed as: Among them, r(i,j) represents the cumulative distance, d(x i ,y j ) represents the distance between the i-th element of time series X and the j-th element of time series Y, p represents the penalty term, and min represents the function of taking the minimum value.

7. The method for filling data of power transmission and transformation equipment based on time series prediction according to claim 6, characterized in that: The parameters of the optimization filling algorithm include solving the optimization problem by constructing the Lagrangian equation, expressed as: Among them, L(ω,b,e,α) represents the Lagrangian function, α represents the Lagrangian multiplier, α k ∈R represents the Lagrange multiplier, k represents the data point, J(ω,e) represents the original objective function, Expresses the difference between predicted and actual values; If L to ω,b,e k ,α k The partial derivative of is equal to 0, and ω,e k Elimination, expressed as: in, represents a v×v matrix, y=[y1,y2,…,y N ] T represents the regularization function, α=[α1,α2,…,α N ] T represents the Lagrange multiplier vector, 1 v =[1,1,…,1] T represents a column vector of dimension v; According to the Mercer condition, there exists a mapping function The relationship between it and the kernel function K(·,·) is expressed as: The estimation function of the least squares support vector machine is expressed as:

8. A system for filling data of power transmission and transformation equipment based on time series prediction according to any one of claims 1 to 7, characterized in that: include: Data collection module, missing detection module, data filling module and filling verification module; The data collection module obtains a multivariate time series by collecting and analyzing historical operation data of power transmission and transformation equipment; The missing value detection module performs missing value detection on the multivariate time series and determines the filled data area; The data filling module fills in the missing data; The filling verification module verifies the filled data, evaluates the error between the filled data and the actual data, and optimizes the parameters of the filling algorithm according to the verification results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for filling data of power transmission and transformation equipment based on time series prediction according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for filling data of power transmission and transformation equipment based on time series prediction according to any one of claims 1 to 7 are implemented.

Images