Power system data restoration method and system based on data completion and error correction

Abstract

The invention provides a power system data restoration method and system based on data completion and error correction, and the method comprises the steps: collecting power system data in real time, and obtaining the real-time power system data; detecting the power system data to obtain missing data summarized by the power system data, and complementing the missing data by using a missing data complementing method based on a DTW distance to obtain the complemented power system data; performing error correction on the complemented power system data by utilizing a mode based on data simulation complementation and actual data comparison to obtain corrected power system data; and after the correction of the power system data is completed, acquiring operation data complemented by missing data in real time, detecting the acquisition quality of the power system data according to the operation data complemented by the missing data, and adjusting the acquisition frequency of the power system data. The system comprises modules corresponding to the steps of the method.

Description

TECHNICAL FIELD

[0001] The application provides a power system data repair method and system based on data completion and error correction, and belongs to the technical field of power system data processing. BACKGROUND

[0002] Power system data is the core basis for realizing power network monitoring, dispatching optimization, fault early warning and equipment maintenance, and its integrity and accuracy directly affect the safe and stable operation of the power system. With the advancement of smart grid construction, the scale of power system data collection is continuously expanding, covering multi-dimensional information such as voltage, current, power and equipment state, and the collection frequency is gradually increasing to meet the real-time monitoring needs. However, in the actual collection process, due to factors such as collection equipment failure, communication link interruption, electromagnetic interference and insufficient sensor precision, power system data inevitably has missing and error problems. For example, the data transmission of distribution terminals in remote areas may be interrupted due to weak communication signals, resulting in data missing; environmental factors such as high temperature and humidity may cause measurement instruments to drift, introducing error data. In the prior art, the completion method for missing data mostly uses linear interpolation, mean filling or a single machine learning model (such as BP neural network), which is difficult to capture the time sequence correlation and periodicity characteristics of power system data, especially in the case of long sequence missing, the completion accuracy is low. For error correction, traditional methods mostly rely on fixed threshold judgment or simple filtering algorithm, without fully combining the inherent error characteristics of power measurement instruments (such as nonlinear error, temperature drift error, etc.), resulting in unsatisfactory correction effect. In addition, the existing technology usually lacks a dynamic evaluation mechanism for data collection quality, and cannot adaptively adjust the collection frequency according to the running state of missing data completion (such as missing ratio, completion time consumption, etc.), which may cause resource waste due to too high collection frequency, or exacerbate data missing problems due to too low frequency. Therefore, there is an urgent need for a power system data repair method that can comprehensively solve the problems of low missing data completion accuracy, insufficient error correction and dynamic adjustment of collection frequency, to improve the reliability and effectiveness of power system data. SUMMARY

[0003] The application provides a power system data repair method and system based on data completion and error correction to solve the technical problems existing in the prior art, and the technical solutions adopted are as follows: The power system data repair method based on data completion and error correction comprises: real-time collection of power system data to obtain real-time power system data; The real-time power system data is detected, missing data of power system data summary is obtained, and the missing data is completed by using a missing data completion method based on DTW distance, to obtain completed power system data; The completed power system data is error corrected by using a data simulation completion and actual data comparison method, to obtain corrected power system data; Real-time collection of missing data completed operation data, detection of power system data collection quality according to the missing data completed operation data, and adjustment of power system data collection frequency.

[0004] Further, real-time collection of power system data to obtain real-time power system data, comprising: Real-time collection of power system data by using a power big data edge collection architecture, wherein the power system data includes power process data, environmental data and field state data; The power system data is denoised to obtain real-time denoised power system data.

[0005] Further, the power big data edge collection architecture includes a plurality of edge nodes and an edge gateway module; wherein the plurality of edge nodes are data connected with the edge gateway module; the edge gateway module is data connected with a missing data completion module; The edge node is used for real-time collection of power system data, and the power system data is sent to the edge gateway module; The edge gateway module is used for format conversion and encryption of the received power system data, and the format converted and encrypted power system data is sent to the missing data completion module.

[0006] Further, the missing data completion method based on DTW distance includes: The similarity of the candidate sequence set S1 and the target sequence corresponding to the missing data is determined by using a preset initial distance threshold D0 to determine an optimal similar fragment set S; Each historical sequence contained in the optimal similar fragment set S is dynamically weighted by using a real-time working condition adaptive coefficient; The missing data is classified, and the missing data completion is performed according to the classification result.

[0007] Further, the similarity of the candidate sequence set S1 and the target sequence corresponding to the missing data is determined by using a preset initial distance threshold D0 to determine an optimal similar fragment set S, comprising: The preset initial distance threshold D0 is called, wherein the preset initial distance threshold D0 takes a value of 60% of the average value of the DTW distance between all historical sequences and the target sequence corresponding to the missing data; The DTW distance between each historical sequence and the target sequence is compared with a preset initial distance threshold D0, and historical sequences with a DTW distance lower than the preset initial distance threshold D0 are selected as a candidate sequence set S1; The cosine similarity T between each historical sequence contained in the candidate sequence set S1 and the target sequence is obtained by using a first-order difference method; A preset similarity threshold is retrieved, wherein the preset similarity threshold has a value of 0.7; The cosine similarity T between each historical sequence contained in the candidate sequence set S1 and the target sequence is compared with the preset similarity threshold, and historical sequences with a cosine similarity T not lower than the preset similarity threshold are selected to form a sequence component set S2; The mutation point positions between the target sequence and the historical sequences contained in the sequence component set S2 are extracted by using a wavelet transform method, and the matching number of the mutation point positions between the target sequence and the historical sequences contained in the sequence component set S2 is obtained; The matching number of the mutation point positions is processed by ratio with the total number of mutation points possessed by the historical sequences contained in the sequence component set S2, and a mutation point matching rate between the target sequence and the historical sequences contained in the sequence component set S2 is obtained; Historical sequences with a mutation point matching rate exceeding a preset mutation point matching rate threshold are selected to generate an optimal similar fragment set S, wherein the preset mutation point matching rate threshold is 0.6.

[0008] Further, each historical sequence contained in the optimal similar fragment set S is dynamically weighted by using a real-time working condition adaptation coefficient, including: Each historical sequence contained in the optimal similar fragment set S is retrieved as a reference historical sequence; The DTW distance of each reference historical sequence is retrieved, and the reciprocal of the DTW distance of the reference historical sequence is taken as a basic weight value of each reference historical sequence; The power system working condition parameters at the corresponding time of each reference historical sequence and the power system working condition parameters corresponding to the target sequence are retrieved, and the Euclidean distance between the power system working condition parameters at the corresponding time of each reference historical sequence and the power system working condition parameters corresponding to the target sequence is obtained; The Euclidean distance between the power system working condition parameters at the corresponding time of each reference historical sequence and the power system working condition parameters corresponding to the target sequence is combined with the basic weight value of each reference historical sequence to obtain a dynamic weight of each reference historical sequence.

[0009] Further, the missing data is classified, and the missing data is completed according to the classification result, including: Retrieve the number of data collection points contained in the target sequence corresponding to the missing data; The number of data collection points contained in the target sequence is compared with a preset point count threshold. Missing data with more than the preset point count threshold is marked as a long missing segment, and missing data with less than the preset point count threshold is marked as a short missing segment; wherein, the preset point count threshold is 3. The missing data corresponding to the missing data in the long missing data segment and the short missing data segment are obtained by using the long missing data completion model and the short missing data completion model respectively, and the missing data in the long missing data segment and the short missing data segment are replaced by the completed data respectively.

[0010] Furthermore, error correction is performed on the completed power system data using a method based on data simulation completion and comparison with actual data to obtain corrected power system data, including: The time series Δt = (Δt1, Δt2, ..., Δt) is completed by real-time data retrieval. m ) Valid power system data of the same type as the missing data; where Δt1, Δt2, ..., Δt m These represent the 1 to m data collection times contained in the time series; The missing data completion method based on DTW distance is used to simulate and complete the data according to the data collection time included in the time series Δt, and the simulated and completed data corresponding to time Δt after data completion is obtained. Based on the data completion, the simulated completed data and time series Δt = (Δt1, Δt2, ..., Δt) corresponding to time Δt after data completion are given. m The corresponding error correction item for the actual collected valid power system data; The error correction term is used to correct the error in the completed data corresponding to the missing data, and the error-corrected completed data is obtained. The error-corrected completed data is the corrected power system data.

[0011] Furthermore, after the power system data has been corrected, operational data to fill in missing data is collected in real time, and the frequency of power system data collection is adjusted, including: After the power system data is corrected, the operation data during the missing data completion process is collected in real time. The operation data during the missing data completion process includes the proportion of missing data, the number of data collection points contained in each missing data segment, and the completion time of each missing data segment. The proportion of missing data, the number of data collection points contained in each missing data segment, and the time for completing each missing data segment are normalized to obtain the normalized proportion of missing data, the number of data collection points contained in each missing data segment, and the time for completing each missing data segment. acquire the data acquisition frequency adjustment coefficient by using the missing data proportion after normalization processing, the number of data acquisition points contained in each missing data segment, and the missing data completion duration of each missing data segment; adjust the acquisition frequency of the power system data by using the data acquisition frequency adjustment coefficient, and acquire the adjusted acquisition frequency.

[0012] Further, the acquisition quality of the power system data is detected according to the operation data completed by the missing data, including: acquire the missing data proportion and the number of data acquisition points contained in each missing data segment from the operation data in the missing data completion process; compare the missing data proportion with a preset proportion threshold value; when the missing data proportion exceeds the preset proportion threshold value, compare the number of data acquisition points contained in each missing data segment with a preset point number threshold value; according to the comparison result between the number of data acquisition points contained in each missing data segment and the preset point number threshold value, screen out the missing data segment whose data acquisition point number exceeds the preset point number threshold value as a target missing data segment; acquire the data quality parameter by using the data amount corresponding to the target missing data segment and the total data amount of the missing data in combination with the missing data proportion; a power system data repair system based on data completion and error correction, the power system data repair system comprising: a data acquisition module, configured to acquire power system data in real time, and acquire real-time power system data; a missing data completion module, configured to detect the power system data, acquire missing data of the power system data, and complete the missing data by using a missing data completion method based on DTW distance, and acquire the power system data after completion; an error data correction module, configured to correct errors of the power system data after completion by using a data simulation completion and actual data comparison method, and acquire the power system data after correction; a data acquisition quality monitoring module, configured to acquire operation data of missing data completion in real time after the power system data is corrected, detect the acquisition quality of the power system data according to the operation data of missing data completion, and adjust the acquisition frequency of the power system data.

[0013] The present application has the following advantages: The application provides a power system data repairing method and system based on data completion and error correction, and improves the accuracy of missing data completion. Through a completion method based on DTW distance, the time sequence correlation and periodicity characteristics of the power system data are effectively captured, and the completion effect is better than that of traditional methods, especially in the case of long sequence missing. The pertinence and accuracy of error correction are enhanced. By using a power measurement instrument error model, the inherent error characteristics of the instrument are fully combined for correction, and the problem of unsatisfactory correction effect of traditional methods is improved. Dynamic adjustment of data acquisition frequency is realized. According to the running data of missing data completion, the acquisition frequency of the power system data is adjusted in real time, the problems of resource waste caused by too high acquisition frequency and aggravated data missing caused by too low acquisition frequency are avoided, and the rationality and efficiency of data acquisition are improved. The integrity, accuracy and reliability of the power system data are improved as a whole, and high-quality data support is provided for power network monitoring, dispatching optimization, fault warning and equipment maintenance. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 a flowchart of the method of the application; Figure 2 a system block diagram of the system of the application. DETAILED DESCRIPTION

[0015] The preferred embodiments of the application are described below in conjunction with the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the application, and are not used to limit the application.

[0016] The application embodiment provides a power system data repairing method based on data completion and error correction, as shown in Figure 1 The power system data repairing method comprises the following steps: Real-time acquisition of power system data, real-time acquisition of power system data; Detecting the real-time power system data, obtaining missing data of power system data summary, and completing the missing data by using a missing data completion method based on DTW distance, and obtaining completed power system data; Error correction of the completed power system data by using a data simulation completion and actual data comparison method, and obtaining corrected power system data; Real-time acquisition of running data of missing data completion, detection of acquisition quality of power system data according to the running data of missing data completion, and adjustment of acquisition frequency of power system data.

[0017] The working principle of the technical solution is that: real-time acquisition of power system data provides original data basis for subsequent processing. The acquired power system data is detected to identify missing data, and then a missing data completion method based on DTW distance is used for completion. This method uses DTW distance to accurately select similar historical data segments with missing data, and combines dynamic weight distribution and segment completion strategies to complete missing data of different lengths to obtain completed power system data. The completed power system data is corrected. The errors in the data are accurately identified and corrected to obtain the corrected power system data. Real-time acquisition of operation data in the missing data completion process, including the proportion of missing data, the number of data acquisition points contained in each missing data, and the duration of each missing data completion, based on these data to obtain a data acquisition frequency adjustment coefficient, and then adjust the acquisition frequency of the power system data to realize dynamic optimization of the acquisition frequency.

[0018] The effect of the above technical solution is: the accuracy of missing data completion is improved. Through the completion method based on DTW distance, the time sequence correlation and periodicity characteristics of the power system data are effectively captured, especially in the long sequence missing scenario, the completion effect is better than that of the traditional method, and the accuracy and accuracy of error correction are enhanced. The error model of the power measuring instrument is used to correct the inherent error characteristics of the instrument, and the problem of unsatisfactory correction effect of the traditional method is improved. The dynamic adjustment of the data acquisition frequency is realized. According to the operation data of the missing data completion, the acquisition frequency is adjusted in real time, which avoids the problems of resource waste caused by too high acquisition frequency and aggravated data missing caused by too low frequency, and improves the rationality and efficiency of data acquisition. The integrity, accuracy and reliability of the power system data are improved as a whole, which provides high-quality data support for power network monitoring, dispatching optimization, fault warning and equipment maintenance.

[0019] In one embodiment of the present application, real-time acquisition of power system data obtains real-time power system data, including: Real-time acquisition of power system data is realized by using a power big data edge acquisition architecture, wherein the power system data includes power process data, environmental data and field state data; The power system data is denoised to obtain real-time denoised power system data.

[0020] The power big data edge acquisition architecture includes a plurality of edge nodes and an edge gateway module; wherein the plurality of edge nodes are connected with the edge gateway module; the edge gateway module is connected with a missing data completion module; The edge node is used for real-time acquisition of power system data, and the power system data is sent to the edge gateway module; The edge gateway module is configured to format and encrypt the received power system data, and send the formatted and encrypted power system data to the missing data completion module.

[0021] The working principle of the above technical solution is that multiple edge nodes collect power system data in real time, the power system data covers power process data, environmental data and field state data, and the collected data is sent to an edge gateway module. After receiving the data from the edge nodes, the edge gateway module formats and encrypts the data to unify the data format and ensure the security of the data, and then sends the processed power system data to a missing data completion module. At the same time, the power system data is denoised to obtain real-time denoised power system data.

[0022] The effect of the above technical solution is that the embodiment realizes real-time collection of multiple types of power system data. With the help of multiple edge nodes, multi-dimensional data such as power processes, environments and field states can be fully captured to ensure the comprehensiveness and real-time nature of data collection. The efficiency and security of data processing are improved. The format conversion function of the edge gateway module unifies data from different sources, which is convenient for subsequent processing. Encryption processing ensures the security of data transmission and prevents data leakage. Through denoising, noise interference in the data is effectively reduced, providing more reliable basic data for subsequent missing data completion and other data processing steps.

[0023] In one embodiment of the present application, a missing data completion method based on DTW distance includes: An optimal similar segment set S is determined by using a preset initial distance threshold D0 and the similarity of the candidate sequence set S1 and the target sequence corresponding to the missing data. Each historical sequence contained in the optimal similar segment set S is dynamically weighted by using a real-time working condition adaptation coefficient. The missing data is classified, and the missing data is completed according to the classification result.

[0024] The working principle of the above technical solution is that: first, taking the preset initial distance threshold D0 as the screening condition, and combining the similarity of each sequence in the candidate sequence set S1 and the target sequence corresponding to the missing data, the required sequences are screened out from the candidate sequences to form the optimal similar fragment set S, providing a reliable historical data basis for subsequent completion. Then, for each historical sequence contained in the optimal similar fragment set S, a real-time working condition adaptation coefficient is introduced, and the weight of each historical sequence is dynamically adjusted according to the matching degree of the real-time working condition (such as system load, voltage deviation, etc.) of the current power system and the working condition at the corresponding time of the historical sequence, so that the weight distribution is more in line with the actual operating state. Finally, the missing data is classified according to the preset classification standard (such as the length and distribution characteristics of the missing data), and then according to different classification results, the corresponding completion strategy (such as weighted average of adjacent points for short missing segments, interpolation and weighting for long missing segments, etc.) is used to complete the missing data, so as to improve the pertinence and accuracy of the completion.

[0025] The effect of the above technical solution is that the accuracy of similar fragment screening is improved. Through double screening of the initial distance threshold and the similarity, it is ensured that the historical sequences in the optimal similar fragment set S have high similarity with the target sequence, providing high-quality data reference for completion. The adaptability of weight distribution is enhanced, and the dynamic weight configuration based on the real-time working condition adaptation coefficient enables the weight of each historical sequence to be adjusted with the working condition, avoiding the problem that the completion result is inconsistent with the actual working condition caused by fixed weight. The accuracy of missing data completion is improved, and through classification of missing data and use of corresponding completion strategy, targeted processing of different types of missing data is realized, effectively improving the poor effect of single completion method in complex missing scenario. The reliability of missing data completion is improved as a whole, providing a strong guarantee for the integrity of power system data, and helping the smooth development of subsequent data processing and analysis work.

[0026] In an embodiment of the present application, the preset initial distance threshold D0 and the similarity of the candidate sequence set S1 and the target sequence corresponding to the missing data are used to determine the optimal similar fragment set S, which includes: The preset initial distance threshold D0 is retrieved, wherein the preset initial distance threshold D0 is 60% of the average value of the DTW distance between all historical sequences and the target sequence corresponding to the missing data; The DTW distance between each historical sequence and the target sequence is compared with the preset initial distance threshold D0, and the historical sequence with a DTW distance lower than the preset initial distance threshold D0 is selected as the candidate sequence set S1; The cosine similarity T between each historical sequence contained in the candidate sequence set S1 and the target sequence is obtained by using the first-order difference method; retrieve a preset similarity threshold, wherein the preset similarity threshold is 0.7; The cosine similarity T between each historical sequence included in the candidate sequence set S1 and the target sequence is compared with a preset similarity threshold, and historical sequences with a cosine similarity T not lower than the preset similarity threshold are selected to form a sequence composition set S2. The mutation point positions between the target sequence and the historical sequences included in the sequence composition set S2 are extracted by wavelet transform, and the matching number of the mutation point positions between the target sequence and the historical sequences included in the sequence composition set S2 is obtained. The matching number of the mutation point positions is processed by ratio to the total number of mutation points possessed by the historical sequences included in the sequence composition set S2, and a mutation point matching rate between the target sequence and the historical sequences included in the sequence composition set S2 is obtained. The historical sequences with a mutation point matching rate exceeding a preset mutation point matching rate threshold are selected to generate an optimal similar segment set S, wherein the preset mutation point matching rate threshold is 0.6.

[0027] The working principle of the above technical solution is that a preset initial distance threshold D0 (with a value of 60% of the average value of the DTW distances of all historical sequences and the target sequence) is retrieved, the DTW distances of the historical sequences and the target sequence are compared with D0, and historical sequences with a DTW distance lower than D0 are selected to form a candidate sequence set S 1。 The cosine similarity T of each historical sequence in the candidate sequence set S1 and the target sequence is calculated by using a first-order difference method, a preset similarity threshold (0.7) is retrieved, historical sequences with a cosine similarity T not lower than the threshold are selected to form a sequence composition set S2. The mutation point positions of the target sequence and the historical sequences in the sequence composition set S2 are extracted by wavelet transform, the matching number of the mutation point positions is calculated, and the matching number is processed by ratio to the total number of mutation points of the historical sequences to obtain a mutation point matching rate. A preset mutation point matching rate threshold (0.6) is retrieved, historical sequences with a mutation point matching rate exceeding the threshold are selected to generate an optimal similar segment set S.

[0028] The effect of the above technical solution is that the embodiment improves the level and precision of similar segment screening. Through multi-dimensional screening of DTW distance, cosine similarity, and mutation point matching rate, the sequences that do not match are gradually eliminated, and it is ensured that the sequences in the optimal similar segment set S are highly consistent with the target sequence in overall distance, change trend, and mutation characteristics. The objectivity and pertinence of the screening standard are enhanced. Each threshold is set based on data characteristics (for example, D0 is based on the average value of the DTW distance, and the similarity threshold and the mutation point matching rate threshold are based on experience and data characteristics), which avoids the randomness of subjective setting and makes the screening result more in line with actual needs. High-quality basic data is provided for subsequent missing data completion. The accurate determination of the optimal similar segment set S reduces the interference of invalid historical sequences on the completion result, lays a reliable foundation for improving the completion precision, and improves the efficiency of similar segment screening as a whole. Through multi-round progressive screening, the candidate range is gradually narrowed while ensuring the screening quality, and the calculation amount of subsequent processing is reduced.

[0029] At the same time, the embodiment realizes progressive application of multi-dimensional thresholds, first screens the basic candidate sequence by using the initial distance threshold (60% of the average value of the DTW distance) based on historical data statistics, further purifies it by using the cosine similarity (threshold 0.7), and finally realizes deep screening by combining the mutation point matching rate (threshold 0.6), which effectively avoids the limitations of single-index judgment, greatly improves the precision and pertinence of similar segment screening, and makes the optimal similar segment set screened more highly consistent with the target sequence in overall trend, local characteristics, and mutation mode, thereby reducing the interference of irrelevant sequences. At the same time, the three indexes of DTW distance (measuring overall sequence similarity), first-order difference cosine similarity (reflecting the consistency of sequence change rate), and mutation point matching rate (capturing the degree of agreement of key turning points) are fused to comprehensively depict the sequence similarity from global trend to local details and from smooth segment to mutation segment, overcome the one-sidedness caused by relying only on a single similarity measure in the prior art, and especially have stronger adaptability to sequences containing mutation characteristics, thereby enhancing the comprehensiveness and robustness of similarity judgment. In addition, the initial distance threshold is dynamically determined based on the average value of the DTW distance of the historical sequence and the target sequence, and the similarity threshold and the mutation point matching rate threshold are set based on sequence feature analysis, which avoids the randomness of subjectively setting the threshold. This threshold system based on data statistics and feature rules makes the screening standard better adapt to sequence characteristics in different scenarios, thereby improving the adaptability of the scheme to diversified sequence data. The high precision and comprehensiveness of the optimal similar segment set further provide a high-quality data basis for subsequent tasks based on similar segments (such as missing data completion, trend prediction, etc.), reduce the processing errors caused by improper selection of similar segments, and thereby improve the accuracy and reliability of the overall data processing result.

[0030] In one embodiment of the present application, real-time operating condition adaptation coefficients are used to dynamically configure the weights of each historical sequence included in the optimal similar segment set S, including: Each historical sequence included in the optimal similar segment set S is retrieved as a reference historical sequence; The DTW distance of each reference historical sequence is retrieved, and the inverse of the DTW distance of the reference historical sequence is taken as the basic weight value of each reference historical sequence; The power system operating condition parameters at the corresponding time of each reference historical sequence and the power system operating condition parameters corresponding to the target sequence are retrieved, and the Euclidean distance between the power system operating condition parameters at the corresponding time of each reference historical sequence and the power system operating condition parameters corresponding to the target sequence is obtained; The Euclidean distance between the power system operating condition parameters at the corresponding time of each reference historical sequence and the power system operating condition parameters corresponding to the target sequence is used in combination with the basic weight value of each reference historical sequence to obtain the dynamic weight of each reference historical sequence.

[0031] The dynamic weight of each reference historical sequence is obtained by the following formula: The dynamic weight of the i-th reference historical sequence is represented by w i The basic weight value of the i-th reference historical sequence is represented by L i The Euclidean distance between the power system operating condition parameters of the i-th reference historical sequence and the power system operating condition parameters corresponding to the target sequence is represented by L h The total sum of the Euclidean distances corresponding to all reference historical sequences is represented by L

[0032] The working principle of the above technical solution is as follows: first, the historical sequences in the optimal similar segment set S are selected as reference sequences, and the inverse of the DTW distance is taken as the basic weight to preliminarily distinguish the values of the sequences (since the DTW distance reflects the similarity of the sequences, the smaller the distance, the more similar, and taking the inverse makes the basic weight of the sequence with high similarity larger). Extract the power operating condition parameters (such as voltage, current, load, etc.) corresponding to the reference sequence and the target sequence at the corresponding time, calculate the Euclidean distance between them, and quantify the operating condition difference - the smaller the Euclidean distance, the closer the operating condition. Through the formula wherein L h is the total sum of the Euclidean distances of all reference sequences, and the basic weight is combined with the operating condition difference. Then, the basic weight is adjusted by the proportion of the Euclidean distance to make the reference sequence closer to the target sequence, the dynamic weight is strengthened, and the "similarity + operating condition adaptation" two-dimensional weighting is realized.

[0033] ​The effect of the above technical solution is to improve the level and precision of similar segment screening. Through multi-dimensional screening of DTW distance, cosine similarity, and mutation point matching rate, the sequences that do not match are gradually eliminated, ensuring that the sequences in the optimal similar segment set S are highly consistent with the target sequence in terms of overall distance, change trend, and mutation characteristics. The objectivity and pertinence of the screening criteria are enhanced, as each threshold is set based on data characteristics, avoiding the randomness of subjective setting and making the screening results more in line with actual needs. The accurate determination of the optimal similar segment set S provides high-quality basic data for subsequent missing data completion, reduces the interference of invalid historical sequences on the completion results, and lays a reliable foundation for improving the completion precision. The overall efficiency of similar segment screening is improved, as the candidate range is gradually narrowed through multiple progressive screenings, ensuring the screening quality while reducing the computational load of subsequent processing.

[0034] At the same time, the traditional weight acquisition method has limitations when determining the weight of historical sequences, as it only relies on single-dimensional information. For example, simply using Euclidean distance can measure the numerical differences between sequences in some dimensions, but when dealing with time series, it is easily affected by distortion in the time axis direction, making it difficult to effectively identify potential patterns and unable to fully consider the overall shape of the sequence and time misalignment. If only the inverse of the DTW distance is used as the weight, it can better handle the stretching and shifting of time series and achieve matching of sequences with similar shapes but different time sequences, but it ignores the impact of real-time working condition parameter differences on the weight.

[0035] This embodiment combines the two, effectively solving the above problems. On the one hand, it effectively improves the comprehensiveness and precision of weight determination. By considering the Euclidean distance between the reference historical sequence and the target sequence in the power system working condition parameters, the real-time working condition factor is taken into account in weight consideration. Different power system working conditions have an important impact on sequence similarity and subsequent applications, and combining the inverse of the DTW distance as the basic weight takes into account the overall similarity of the sequence shape. From the two important aspects of sequence shape and real-time working condition, the weight is determined comprehensively, which greatly improves the precision of the weight compared to a single method.

[0036] On the other hand, it enhances the adaptability of the model to dynamic changing environments. The power system working condition is in a dynamic state, and this scheme can adjust the weight of the historical sequence based on the real-time working condition parameter changes. When the working condition changes, the Euclidean distance changes to prompt the dynamic weight to change accordingly, ensuring that the historical sequence can participate in subsequent data processing with reasonable weight under different working conditions. For example, in different working condition scenarios such as power system load mutation and voltage fluctuation, the analysis and prediction tasks based on dynamic weight can maintain high accuracy, overcoming the defect of traditional fixed weight methods that are difficult to adapt to dynamic changes in working conditions, and improving the reliability and practicality of the entire technical solution in complex and variable power system environments.

[0037] One embodiment of the present application classifies missing data and completes the missing data according to the classification results, comprising: retrieve the number of data acquisition points contained in the target sequence corresponding to the missing data; compare the number of data acquisition points contained in the target sequence with a preset point threshold, mark the missing data whose number of data acquisition points exceeds the preset point threshold as a long missing segment, and mark the missing data whose number of data acquisition points does not exceed the preset point threshold as a short missing segment; wherein the preset point threshold is 3; respectively use a long missing data completion model and a short missing data completion model to obtain the completion data corresponding to the missing data of the long missing segment and the short missing segment, and respectively replace the missing data of the long missing segment and the short missing segment with the completion data.

[0038] wherein the long missing data completion model has the following structure: wherein X ct represents the completion data corresponding to the long missing data; n represents the total number of reference historical sequences contained in the optimal similar segment set S; represents the dynamic weight of the i-th reference historical sequence; interp(Y i represents the data value obtained by linear interpolation of the t-th acquisition point position corresponding to the short missing data in the i-th reference historical sequence; the short missing data completion model has the following structure: wherein X dt represents the completion data corresponding to the short missing data; n represents the total number of reference historical sequences contained in the optimal similar segment set S; represents the dynamic weight of the i-th reference historical sequence; Y it represents the historical data value of the t-th acquisition point position corresponding to the short missing data in the i-th reference historical sequence.

[0039] The working principle of the above technical solution is as follows: the number of data acquisition points contained in the target sequence corresponding to the missing data is retrieved as the classification basis. The number of data acquisition points in the target sequence is compared with a preset point threshold (value 3), if it exceeds the threshold, the corresponding missing data is marked as a long missing segment; if it does not exceed, it is marked as a short missing segment. For long missing segments and short missing segments, long missing data completion model and short missing data completion model are used for processing respectively to obtain their corresponding completion data, and the original missing data is replaced with the completion data to complete the completion of the missing data.

[0040] The technical scheme has the effects that: taking the number of data collection points as a quantitative standard, long missing sections and short missing sections are clearly distinguished in combination with a preset threshold, the classification logic is clear, and a foundation is laid for subsequent targeted completion. The adaptability of missing data completion is improved, a special completion model is used for missing sections of different lengths, the problem of insufficient adaptation of a single model to different missing types is avoided, the completion method is more in line with the characteristics of various missing data, the accuracy of missing data completion is improved, the differential design of long and short missing section completion models can better meet the completion needs of missing data of different lengths, reduces completion errors, and ensures the reliability of the completed data. At the same time, the logic and efficiency of the completion process are enhanced, the classification processing makes the completion process more organized, avoids resource waste caused by indiscriminate processing, and improves the overall efficiency of missing data completion.

[0041] At the same time, by classifying the missing data into long and short missing sections, an adaptive completion model can be used for different length missing sections according to their characteristics, improving the pertinence of completion. Secondly, the long missing data completion model uses linear interpolation combined with dynamic weights, which can effectively solve the completion problem caused by the large amount of missing data and incomplete information in long missing sections, making the completed data more in line with the sequence trend. The short missing data completion model directly uses the historical data of the corresponding points of the reference historical sequence combined with dynamic weights, ensuring the completion efficiency while using the similarity of historical data to ensure the completion accuracy. Furthermore, the introduction of dynamic weights fully considers the differences between the reference historical sequence and the target sequence in terms of sequence similarity and real-time working conditions, allowing the completion process to dynamically adjust the contribution of each reference sequence according to different situations, further improving the accuracy and reliability of missing data completion, and overall achieving efficient and accurate completion of different types of missing data.

[0042] In an embodiment of the present application, the completed power system data is error-corrected based on data simulation completion and actual data comparison to obtain corrected power system data, comprising: real-time data completion time series Δt = (Δt1, Δt2, …, Δt m ) and effective power system data of the same type as the missing data; wherein Δt1, Δt2, …, Δt m represent 1 to m data collection times included in the time series, respectively; using a missing data completion method based on DTW distance to simulate and complete data according to the data collection times included in the time series Δt, to obtain simulated completion data corresponding to the Δt time after data completion; according to the simulated completion data corresponding to the Δt time after data completion and the time series Δt = (Δt1, Δt2, …, Δt mcorresponding to the actual collected effective power system data; error correction of the completed data corresponding to the missing data by using the error correction term, to obtain the completed data after error correction, wherein the completed data after error correction is the corrected power system data.

[0043] The structure of the error correction term is as follows: Wherein, S represents the error correction term; m represents the total number of data collection time points contained in the time series Δt; X si and X mi represent the actual collected effective power system data and the simulated completed data corresponding to the i-th data collection time point contained in the time series Δt; Meanwhile, the completed data after error correction is obtained by the following formula: Wherein, X wi represents the completed data after error correction corresponding to the i-th missing data; X ti represents the completed data corresponding to the i-th missing data; S represents the error correction term.

[0044] The working principle of the above technical solution is as follows: real-time retrieval of effective power system data with the same type of missing data in the time series Δt=(Δt1, Δt2, …, Δt m )after data completion, wherein Δt1 to Δt m are 1 to m data collection time points in the time series. Using the missing data completion method based on DTW distance, data simulation completion is performed according to each data collection time point in the time series Δt, to obtain simulated completed data corresponding to the Δt time point after data completion. According to the simulated completed data corresponding to the Δt time point after data completion and the actual collected effective power system data corresponding to the time series Δt, the error correction term is calculated, which reflects the deviation between the simulated completed data and the actual effective data. The error correction term is used to correct the completed data corresponding to the missing data, and the completed data after correction is the corrected power system data.

[0045] The effect of the above technical solution is that: by comparing the simulated completed data with the actual effective data to obtain the error correction term, the correction process has a reliable basis, and the deviation existing in the completed data can be corrected. The reliability of the completed data is improved, the completed data is adjusted by the error correction term, the error possibly introduced in the completion process is effectively reduced, and the final corrected data is closer to the real power system data. The rationality of data correction is enhanced, since the effective power system data and the missing data are of the same type and are simulated based on the same completion method, the applicability of the error correction term is ensured, and the correction logic is more in line with the actual data characteristics. High-quality data support is provided for subsequent application of the power system data, and the corrected power system data is more accurate, which can better meet the requirements of power system monitoring, dispatching and other work on data quality.

[0046] Meanwhile, the embodiment corrects the error based on the comparison between the simulated completion and the actual data, which can effectively improve the accuracy of the completed power system data. The same type of effective data is real-time retrieved and simulated to complete, providing a reliable reference benchmark for error correction; the error correction term is calculated to accurately quantify the deviation between the simulated completed data and the actual effective data; and the completion result of the missing data is adjusted using the error correction term, which can correct the deviation caused by model assumptions, data similarity and other factors in the completion process, so that the final corrected power system data is closer to the real situation, enhancing the reliability and usability of the data. Meanwhile, as the real-time working condition of the power system changes, the comparison between the simulated completion and the actual data dynamically updates the error correction term, so that the correction process can adapt to the dynamic changes of the system, without the need for frequent manual adjustment of the correction rules, enhancing the applicability of the scheme in complex and variable power scenarios. Through multiple interactions between simulated completion and actual data, the influence of noise interference and occasional abnormal data points in the data can be effectively resisted, and even if there is local data fluctuation, the error correction mechanism can ensure the stability and reliability of the overall data.

[0047] In an embodiment of the present application, after the power system data is corrected, the running data in the missing data completion process is collected in real time, including: After the power system data is corrected, the running data in the missing data completion process is collected in real time, wherein the running data in the missing data completion process includes the missing data proportion, the number of data acquisition points contained in each missing data, and the completion time length of each missing data; The missing data proportion, the number of data acquisition points contained in each missing data, and the completion time length of each missing data are normalized to obtain the normalized missing data proportion, the number of data acquisition points contained in each missing data, and the completion time length of each missing data; The missing data proportion after normalization processing, the data acquisition point number contained in each missing data segment, and the missing data completion duration of each missing data segment are used to obtain a data acquisition frequency adjustment coefficient; The data acquisition frequency adjustment coefficient is used to adjust the acquisition frequency of the power system data, and an adjusted acquisition frequency is obtained.

[0048] The adjusted acquisition frequency is obtained by the following formula: Wherein, f t represents the adjusted acquisition frequency; f0 represents the acquisition frequency before adjustment; B represents the missing data proportion after normalization processing; k represents the total number of segments of the missing data that have been completed; C i represents the data acquisition point number contained in the i-th missing data segment after normalization processing; T ei represents the missing data completion duration corresponding to the i-th missing data segment after normalization processing.

[0049] The working principle of the above technical solution is that the running data in the missing data completion process is collected in real time, including the missing data proportion, the data acquisition point number contained in each missing data segment, and the missing data completion duration, to provide original data for subsequent analysis. The collected missing data proportion, data acquisition point number contained in each missing data segment, and missing data completion duration are normalized to eliminate the influence of different data dimensions, so that each data is in the same order of magnitude, facilitating comprehensive analysis. Based on the missing data proportion after normalization processing, the data acquisition point number contained in each missing data segment, and the missing data completion duration of each missing data segment, a data acquisition frequency adjustment coefficient is calculated and obtained, which comprehensively reflects the influence degree of the running data on the acquisition frequency. The current acquisition frequency of the power system data is adjusted by using the data acquisition frequency adjustment coefficient, and an adjusted acquisition frequency is obtained, so as to realize dynamic optimization of the acquisition frequency.

[0050] The technical scheme has the effects that: the quality of power system data acquisition is effectively detected, the operation data with missing data completed through real-time acquisition and analysis can timely master the situation of missing data in the data acquisition process and the completion difficulty, and provides a quantitative basis for evaluating the acquisition quality; the dynamic adjustment of the power system data acquisition frequency is realized, the adjustment coefficient based on the operation data can make the acquisition frequency flexibly change according to the actual data acquisition condition, and the disadvantages caused by the fixed frequency are avoided; the adaptability and rationality of data acquisition are improved, the normalization processing ensures the fairness of different operation data in analysis, the calculation of the adjustment coefficient is more scientific, and then the adjustment of the acquisition frequency is more in line with the actual demand; the overall efficiency of the power system data acquisition is improved, the acquisition frequency is reasonably adjusted, the data quality is ensured, unnecessary resource consumption is reduced, and the economy and efficiency of data acquisition are improved.

[0051] In one embodiment of the application, the acquisition quality of power system data is detected according to the operation data with missing data completed, including: The missing data proportion and the number of data acquisition points contained in each missing data are retrieved from the operation data in the missing data completion process; The missing data proportion is compared with a preset proportion threshold value; When the missing data proportion exceeds the preset proportion threshold value, the number of data acquisition points contained in each missing data is compared with a preset point threshold value; According to the comparison result between the number of data acquisition points contained in each missing data and the preset point threshold value, the missing data segment whose number of data acquisition points exceeds the preset point threshold value is screened out as a target missing data segment; The data quality parameter is obtained by combining the missing data proportion with the total data amount of the missing data and the data amount corresponding to the target missing data segment; The data quality parameter is compared with a preset data quality parameter threshold value, and when the data quality parameter exceeds the preset data quality parameter threshold value, it is determined that the data acquisition quality is abnormal, and acquisition quality abnormality alarm is performed.

[0052] The data quality parameter is obtained by the following formula: Wherein, U represents the data quality parameter; g represents the number of target missing data segments; G j represents the data amount corresponding to the jth target missing data segment; G represents the total data amount of the missing data; C j represents the number of data acquisition points contained in the jth target missing data segment after normalization processing; B represents the missing data proportion after normalization processing.

[0053] The working principle of the technical solution is as follows: two key indicators are extracted from the missing data completion process, that is, the missing data proportion (reflecting the overall missing degree) and the number of data acquisition points contained in each missing data segment (reflecting the scale of single missing data). By using progressive screening logic, the missing data proportion is compared with the preset threshold value first, and only when the proportion is out of limit, the number of acquisition points of each missing data segment is compared with the threshold value, and the segment with the number of acquisition points exceeding the threshold value is selected as the "target missing data segment" (focusing on the concentrated and large-scale missing data which has more significant impact on data quality). Based on the number of target missing segments, the data amount corresponding to each segment, the total amount of missing data, and in combination with the normalized single missing acquisition point number and missing data proportion, the data quality parameter U (comprehensively reflecting the severity, distribution characteristics and overall impact of missing data) is calculated through a specific formula. The data quality parameter U is compared with the preset threshold value, and if U is out of limit, it is determined that the data acquisition quality is abnormal, and the alarm mechanism is triggered.

[0054] The effect of the technical solution is as follows: through the progressive screening of "missing proportion -> single acquisition point number", the target missing segment with large scale and strong impact is preferentially focused on, and false judgment caused by slight or scattered missing data is avoided, so that the detection is more focused on the core problem. The data quality parameter U integrates the number of target missing segments, data amount, normalized acquisition point number and missing proportion, and comprehensively reflects the "scale-distribution-overall impact" of missing data, which can more objectively reflect the real state of data acquisition quality compared with a single indicator. Through the threshold triggering mechanism, automatic judgment and alarm of data acquisition quality anomaly are realized, which can quickly identify the problem and start intervention, reducing the impact of data quality problems on power system analysis and decision-making. Based on the operation data of the missing data completion process, the detection is carried out, the existing data resources are fully utilized, no additional acquisition burden is needed, and the detection accuracy and implementation efficiency are considered.

[0055] The embodiment of the present application proposes a power system data repair system based on data completion and error correction, as shown in Figure 2 The power system data repair system comprises: A data acquisition module is configured to acquire real-time power system data and obtain real-time power system data. A missing data completion module is configured to detect the power system data, obtain missing data of the power system data summary, and complete the missing data by using a missing data completion method based on DTW distance, and obtain completed power system data. An error data correction module is configured to correct errors of the completed power system data by using a data simulation completion and actual data comparison method, and obtain corrected power system data. The data acquisition quality monitoring module is used for acquiring running data of missing data completion in real time after the power system data is corrected, detecting the acquisition quality of the power system data according to the running data of missing data completion, and adjusting the acquisition frequency of the power system data.

[0056] The working principle of the technical solution is that the power system data is acquired in real time to provide original data basis for subsequent processing. The acquired power system data is detected to identify missing data, and then a missing data completion method based on DTW distance is used for completion. This method uses DTW distance to accurately select historical data segments similar to the sequence where the missing data is located, and combines dynamic weight distribution and segment completion strategies to complete missing data of different lengths in a targeted manner to obtain completed power system data. The completed power system data is corrected. The errors in the data are accurately identified and corrected to obtain corrected power system data. Running data in the missing data completion process is acquired in real time, including the proportion of missing data, the number of data acquisition points contained in each missing data, and the duration of each missing data completion. Based on these data, a data acquisition frequency adjustment coefficient is obtained, and the acquisition frequency of the power system data is adjusted to realize dynamic optimization of the acquisition frequency.

[0057] The effect of the technical solution is that the accuracy of missing data completion is improved. Through the completion method based on DTW distance, the time sequence correlation and periodicity characteristics of the power system data are effectively captured, especially in the long sequence missing scenario, the completion effect is better than that of the traditional method. The accuracy and accuracy of error correction are enhanced. The error model of the power measuring instrument is used to correct the inherent error characteristics of the instrument, and the problem of unsatisfactory correction effect of the traditional method is improved. Dynamic adjustment of the data acquisition frequency is realized. The acquisition frequency is adjusted in real time according to the running data of missing data completion, which avoids the problems of resource waste caused by too high acquisition frequency and aggravated data missing caused by too low acquisition frequency, and improves the rationality and efficiency of data acquisition. The integrity, accuracy and reliability of the power system data are improved as a whole, which provides high-quality data support for power network monitoring, dispatching optimization, fault warning and equipment maintenance.

[0058] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.

Claims

1. A power system data restoration method based on data completion and error correction, characterized in that, The power system data repairing method comprises: Real-time acquisition of power system data to obtain real-time power system data; Detection of the real-time power system data to obtain missing data of power system data summary, and completion of the missing data by using a missing data completion method based on DTW distance to obtain completed power system data; Error correction of the completed power system data by using a data simulation completion and actual data comparison method to obtain corrected power system data; After the power system data is corrected, real-time acquisition of operation data for missing data completion, detection of the acquisition quality of the power system data according to the operation data for missing data completion, and adjustment of the acquisition frequency of the power system data.

2. The method for power system data restoration based on data completion and error correction according to claim 1, characterized in that, Real-time acquisition of power system data to obtain real-time power system data, comprising: Real-time acquisition of power system data by using a power big data edge acquisition architecture, wherein the power system data comprises power process data, environmental data and field state data; De-noising of the power system data to obtain real-time de-noised power system data; The power big data edge acquisition architecture comprises a plurality of edge nodes and an edge gateway module; the plurality of edge nodes are connected with the edge gateway module; the edge gateway module is connected with a missing data completion module; The plurality of edge nodes are configured to acquire power system data in real time and send the power system data to the edge gateway module; The edge gateway module is configured to format and encrypt the received power system data and send the formatted and encrypted power system data to the missing data completion module.

3. The method for power system data restoration based on data completion and error correction according to claim 1, wherein, The missing data completion method based on DTW distance comprises: Determination of an optimal similar segment set S by using a preset initial distance threshold D0 and a similarity of a candidate sequence set S1 to a target sequence corresponding to missing data; Dynamic weight configuration of each historical sequence contained in the optimal similar segment set S by using a real-time working condition adaptation coefficient; Classification of the missing data and completion of the missing data according to the classification results.

4. The method for power system data restoration based on data completion and error correction according to claim 3, characterized in that, Determination of an optimal similar segment set S by using a preset initial distance threshold D0 and a similarity of a candidate sequence set S1 to a target sequence corresponding to missing data, comprising: Retrieval of a preset initial distance threshold D0, wherein the preset initial distance threshold D0 takes a value of 60% of an average value of DTW distances between all historical sequences and the target sequence corresponding to the missing data; Comparison of a DTW distance between each historical sequence and the target sequence with the preset initial distance threshold D0, and screening of a historical sequence with a DTW distance lower than the preset initial distance threshold D0 as a candidate sequence set S1; Obtainment of a cosine similarity T between each historical sequence contained in the candidate sequence set S1 and the target sequence by using a first-order difference method; Retrieval of a preset similarity threshold, wherein the preset similarity threshold takes a value of 0.7; The cosine similarity T between each historical sequence contained in the candidate sequence set S1 and the target sequence is compared with a preset similarity threshold, and historical sequences with a cosine similarity T not lower than the preset similarity threshold are screened to form a sequence composition set S2; The mutation point positions between the target sequence and the historical sequences contained in the sequence composition set S2 are extracted by wavelet transform, and the matching number of the mutation point positions between the target sequence and the historical sequences contained in the sequence composition set S2 is obtained; The matching number of the mutation point positions is processed by ratio with the total number of mutation points possessed by the historical sequences contained in the sequence composition set S2, and the mutation point matching rate between the target sequence and the historical sequences contained in the sequence composition set S2 is obtained; The historical sequences with a mutation point matching rate exceeding a preset mutation point matching rate threshold are screened to generate an optimal similar fragment set S, wherein the preset mutation point matching rate threshold is 0.

6.

5. The method for power system data restoration based on data completion and error correction according to claim 3, wherein, The real-time working condition adaptation coefficients are used to dynamically configure the weights of the historical sequences contained in the optimal similar fragment set S, including: Each historical sequence contained in the optimal similar fragment set S is called as a reference historical sequence; The DTW distance of each reference historical sequence is called, and the reciprocal of the DTW distance of the reference historical sequence is taken as the basic weight value of each reference historical sequence; The power system working condition parameters at the corresponding time of each reference historical sequence and the power system working condition parameters corresponding to the target sequence are called, and the Euclidean distance between the power system working condition parameters at the corresponding time of each reference historical sequence and the power system working condition parameters corresponding to the target sequence is obtained; The Euclidean distance between the power system working condition parameters at the corresponding time of each reference historical sequence and the power system working condition parameters corresponding to the target sequence is used to obtain the dynamic weight of each reference historical sequence in combination with the basic weight value of each reference historical sequence.

6. The method for power system data restoration based on data completion and error correction according to claim 3, wherein, The missing data is classified, and the missing data is completed according to the classification result, including: The data acquisition point number contained in the target sequence corresponding to the missing data is called; The data acquisition point number contained in the target sequence is compared with a preset point number threshold, the missing data with a data acquisition point number exceeding the preset point number threshold is marked as a long missing segment, and the missing data with a data acquisition point number not exceeding the preset point number threshold is marked as a short missing segment; wherein the preset point number threshold is 3; The long missing segment and the short missing segment are obtained by using the long missing data completion model and the short missing data completion model respectively, and the missing data of the long missing segment and the short missing segment is replaced by using the completion data respectively.

7. The method for power system data restoration based on data completion and error correction according to claim 1, wherein, The error of the completed power system data is corrected by using a data simulation completion and actual data comparison method to obtain corrected power system data, including: The real-time data call the time series Δt=(Δt1, Δt2, …, Δt m ) after the data completion, and the valid power system data corresponding to the missing data of the same type; wherein Δt1, Δt2, …, Δt m respectively represent 1 to m data collection time points contained in the time series; The missing data is completed by using a DTW distance-based missing data completion method according to the data acquisition time contained in the time sequence Δt to obtain simulation completion data corresponding to the Δt time after data completion; According to the data completion, the simulation completion data corresponding to the time sequence Δt= (Δt1, Δt2, …, Δt m ) corresponding to the actual acquisition error correction term of the effective power system data The error correction term is used to correct the completed data corresponding to the missing data, and error-corrected completed data is obtained, wherein the error-corrected completed data is the corrected power system data.

8. The method for power system data restoration based on data completion and error correction of claim 1, wherein, After the power system data is corrected, the operation data in the process of missing data completion is collected in real time, and the collection frequency of the power system data is adjusted, including: After the power system data is corrected, the operation data in the process of missing data completion is collected in real time, wherein the operation data in the process of missing data completion includes a missing data proportion, a number of data collection points contained in each missing data segment, and a missing data completion duration of each missing data segment; The missing data proportion, the number of data collection points contained in each missing data segment, and the missing data completion duration of each missing data segment are normalized to obtain a normalized missing data proportion, a normalized number of data collection points contained in each missing data segment, and a normalized missing data completion duration of each missing data segment; The normalized missing data proportion, the normalized number of data collection points contained in each missing data segment, and the normalized missing data completion duration of each missing data segment are used to obtain a data collection frequency adjustment coefficient; The data collection frequency adjustment coefficient is used to adjust the collection frequency of the power system data to obtain an adjusted collection frequency.

9. The method for power system data restoration based on data completion and error correction of claim 1, wherein, The collection quality of the power system data is detected according to the operation data in the process of missing data completion, including: The missing data proportion and the number of data collection points contained in each missing data segment are retrieved from the operation data in the process of missing data completion; The missing data proportion is compared with a preset proportion threshold value; When the missing data proportion exceeds the preset proportion threshold value, the number of data collection points contained in each missing data segment is compared with a preset point number threshold value; According to a comparison result between the number of data collection points contained in each missing data segment and the preset point number threshold value, a missing data segment whose number of data collection points exceeds the preset point number threshold value is screened out as a target missing data segment; A data quality parameter is obtained by combining the missing data proportion with a total data amount of the missing data and a data amount corresponding to the target missing data segment; The data quality parameter is compared with a preset data quality parameter threshold value, and when the data quality parameter exceeds the preset data quality parameter threshold value, it is determined that the data collection quality is abnormal, and a collection quality abnormality alarm is given.

10. A power system data restoration system based on data completion and error correction, characterized by, The power system data repair system includes: A data collection module is configured to collect power system data in real time to obtain real-time power system data. A missing data completion module is configured to detect the power system data, obtain missing data of the power system data, and complete the missing data by using a missing data completion method based on a DTW distance to obtain completed power system data. An error data correction module is configured to correct the completed power system data by using a data simulation completion and actual data comparison method to obtain corrected power system data. The data acquisition quality monitoring module is used for collecting the operation data with missing data completed in real time, detecting the acquisition quality of the power system data according to the operation data with missing data completed, and adjusting the acquisition frequency of the power system data.

Images