Methods, apparatus, equipment, and storage media for feature reconstruction of virtual power plant transaction data

Abstract

This application relates to a method, apparatus, device, and storage medium for feature reconstruction of virtual power plant transaction data. The method includes: acquiring virtual power plant transaction data to be detected; mapping the virtual power plant transaction data to generate an initial standardized transaction dataset; the virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies; performing anomaly classification on the initial standardized transaction dataset to obtain anomaly classification results, and determining anomaly handling operations based on the anomaly classification results; reconstructing and verifying the data based on the anomaly classification results; and generating a target standardized transaction dataset if the reconstructed transaction data passes verification. By classifying anomalies and matching specific alarm levels and handling procedures to different levels of anomalies, precise control of anomaly risks is achieved. Implementing differentiated data reconstruction operations based on anomaly classification results significantly improves the utilization value and reliability of virtual power plant transaction data.

Description

Technical Field

[0001] This application relates to the field of electricity market trading technology, and in particular to a method, apparatus, equipment and storage medium for reconstructing the characteristics of virtual power plant trading data. Background Technology

[0002] As the market-oriented reform of the current power market continues to deepen, virtual power plants, as an important vehicle for integrating distributed energy and improving the efficiency of power resource allocation, are increasingly participating in and influencing the power trading system, becoming a core component of the diversified participants in the power market. The power trading process of virtual power plants covers the entire lifecycle and multiple stages, including medium- and long-term contracts, short-term plan breakdown, spot market declaration and clearing, real-time settlement, and deviation assessment. The transaction data generated at each stage exhibits significant characteristics of multi-level, multi-time granularity, and multi-collection frequency, and there are close temporal correlations and hierarchical constraints between transaction data at different levels and time periods. The accuracy, completeness, and timeliness of transaction data directly determine the rationality of virtual power plant market declarations, the precision of settlement execution, and the effectiveness of risk management. As electricity market trading rules continue to improve, the frequency, complexity, and data volume of virtual power plant transactions are increasing simultaneously. Data is prone to various anomalies during collection, transmission, and processing due to factors such as equipment failure, transmission interference, and manual operation. If abnormal data cannot be detected and repaired in a timely manner, it will lead to problems such as deviations in trading plans, settlement errors, and even market violations. Therefore, conducting efficient anomaly detection and feature reconstruction of virtual power plant transaction data has become the key to ensuring the stable participation of virtual power plants in electricity market transactions and is also an urgent technical requirement for the digital development of the electricity market.

[0003] Currently, the industry mostly uses common data processing technologies for anomaly detection and reconstruction of power trading data. The core processing logic is as follows: perform simple standardized format conversion on the original power trading data, use a single numerical indicator or basic time series analysis method to determine data anomalies, use a unified manual review and handling method for detected anomalies, complete the reconstruction of abnormal data through a single numerical repair method, and perform only simple numerical rationality verification after reconstruction.

[0004] However, the above data processing methods use a single numerical indicator to determine anomalies and a single numerical repair method to reconstruct abnormal data. The accuracy and rationality of the reconstructed data cannot be effectively guaranteed, and the reconstructed data cannot directly support subsequent trading decisions. Summary of the Invention

[0005] Based on this, it is necessary to provide a feature reconstruction method, apparatus, equipment, and storage medium for virtual power plant transaction data that can ensure the reconstructed data meets the requirements of inter-level temporal dependency and numerical rationality, addressing the aforementioned technical problems.

[0006] Firstly, this application provides a method for reconstructing the features of virtual power plant transaction data, including:

[0007] The process involves acquiring virtual power plant transaction data to be tested, mapping the virtual power plant transaction data, and generating an initial standardized transaction dataset. The virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies.

[0008] Anomaly classification is performed on the initial standardized transaction dataset to obtain anomaly classification results, and anomaly handling operations are determined based on the anomaly classification results.

[0009] Based on the anomaly classification results, the data is reconstructed and verified. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated.

[0010] In one embodiment, data reconstruction and verification are performed based on the anomaly classification results, including:

[0011] The association strength of transaction data at each level in the initial standardized transaction dataset is determined according to the preset time-series association algorithm, and a time-series association matrix is ​​constructed based on the association strength.

[0012] Based on the anomaly classification results and the temporal correlation matrix, the initial standardized transaction dataset is reconstructed to obtain the reconstructed transaction data.

[0013] The reconstructed transaction data is validated based on the temporal correlation matrix to obtain the validation result; the validation result is used to characterize whether the reconstructed transaction data passes the validation.

[0014] In one embodiment, constructing a time-series correlation matrix based on correlation strength includes:

[0015] Construct cross-level association matrices and multiple intra-level association matrices based on the association strength;

[0016] For each level of the association matrix, determine the mean of the association matrix within the level.

[0017] Using the cross-level correlation matrix as a benchmark, the matrix mean of the intra-level correlation matrix is ​​superimposed with the corresponding elements of the cross-level correlation matrix to obtain the time-series correlation matrix.

[0018] In one embodiment, the initial standardized transaction dataset is reconstructed based on the anomaly classification results and the temporal correlation matrix to obtain reconstructed transaction data, including:

[0019] In the case of a fatal anomaly in the anomaly classification result, the transaction data at each level is reconstructed according to the preset reconstruction optimization algorithm and constraints to obtain the reconstructed transaction data; the constraints include end-to-end constraints and total power constraints.

[0020] When the anomaly classification result is severe anomaly, the association level of the abnormal data in the time series correlation matrix is ​​determined. The abnormal data is then reconstructed based on the normal transaction data of the association level and the preset reconstruction optimization algorithm to obtain the reconstructed transaction data.

[0021] If the anomaly classification result is a general anomaly, determine the normal time period data within the transaction level to which the anomaly data belongs, and reconstruct the anomaly data based on the normal time period data to obtain the reconstructed transaction data.

[0022] If the anomaly classification result is minor, determine the historical data of the same period of the anomaly, and reconstruct the anomaly based on the historical data of the same period to obtain the reconstructed transaction data.

[0023] In one embodiment, the reconstructed transaction data is verified based on the temporal correlation matrix to obtain the verification result, including:

[0024] Substitute the reconstructed transaction data into the time-series correlation matrix to determine the correlation strength and deviation between the reconstructed transaction data and the transaction data at the correlation level;

[0025] The correlation strength is compared with the preset comprehensive time series correlation strength, and the deviation is compared with the preset deviation threshold to obtain the verification result.

[0026] In one embodiment, the virtual power plant transaction data includes annual data, monthly data, day-ahead data, intraday data, real-time data, and deviation assessment data; the virtual power plant transaction data is mapped to generate an initial standardized transaction dataset, including:

[0027] Virtual power plant transaction data is uniformly mapped to the same time-series baseline according to time series.

[0028] Based on the forward decomposition link, the annual data is decomposed layer by layer according to the time and planning logic of power trading to obtain initial transaction data at different levels.

[0029] Based on the time-series feedback loop, and using real-time data and deviation assessment data as benchmarks, the initial transaction data is corrected layer by layer to obtain the feedback-received initial transaction data, and the initial standardized transaction dataset is determined based on the feedback-received initial transaction data.

[0030] Secondly, this application also provides a feature reconstruction device for virtual power plant transaction data, comprising:

[0031] The mapping module is used to acquire the virtual power plant transaction data to be detected, map the virtual power plant transaction data, and generate an initial standardized transaction dataset; the virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies;

[0032] The determination module is used to classify anomalies in the initial standardized transaction dataset, obtain the anomaly classification results, and determine the anomaly handling operations based on the anomaly classification results.

[0033] The reconstruction module is used to reconstruct and verify data based on the anomaly classification results. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated.

[0034] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0035] The process involves acquiring virtual power plant transaction data to be tested, mapping the virtual power plant transaction data, and generating an initial standardized transaction dataset. The virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies.

[0036] Anomaly classification is performed on the initial standardized transaction dataset to obtain anomaly classification results, and anomaly handling operations are determined based on the anomaly classification results.

[0037] Based on the anomaly classification results, the data is reconstructed and verified. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated.

[0038] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0039] The process involves acquiring virtual power plant transaction data to be tested, mapping the virtual power plant transaction data, and generating an initial standardized transaction dataset. The virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies.

[0040] Anomaly classification is performed on the initial standardized transaction dataset to obtain anomaly classification results, and anomaly handling operations are determined based on the anomaly classification results.

[0041] Based on the anomaly classification results, the data is reconstructed and verified. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated.

[0042] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0043] The process involves acquiring virtual power plant transaction data to be tested, mapping the virtual power plant transaction data, and generating an initial standardized transaction dataset. The virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies.

[0044] Anomaly classification is performed on the initial standardized transaction dataset to obtain anomaly classification results, and anomaly handling operations are determined based on the anomaly classification results.

[0045] Based on the anomaly classification results, the data is reconstructed and verified. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated.

[0046] The aforementioned method, apparatus, equipment, and storage medium for reconstructing virtual power plant transaction data acquire the virtual power plant transaction data to be detected, map the virtual power plant transaction data, and generate an initial standardized transaction dataset. The virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies. Anomaly classification is performed on the initial standardized transaction dataset to obtain anomaly classification results, and anomaly handling operations are determined based on the anomaly classification results. Data reconstruction and verification are performed based on the anomaly classification results. If the reconstructed transaction data passes verification, a target standardized transaction dataset is generated. Based on the time scale, impact range, and other characteristics of the data, refined anomaly classification is achieved, matching specific alarm levels and handling procedures for different levels of anomalies, realizing precise control of anomaly risks. This approach breaks through the limitations of traditional single-dimensional analysis, effectively avoiding the problems of missed or false anomaly detection. At the same time, the graded handling mechanism achieves differentiated and efficient anomaly response, quickly blocking the cross-level transmission of anomalies, significantly improving the comprehensiveness and accuracy of virtual power plant transaction data anomaly detection, and ensuring the orderly progress of the transaction process. By matching targeted reconstruction strategies to anomaly classification results, differentiated data reconstruction operations are implemented based on the anomaly level and impact scope. This ensures the reconstruction process aligns with the hierarchical characteristics and temporal patterns of virtual power plant transaction data, balancing reconstruction accuracy with data temporal consistency. Simultaneously, the reconstructed data is substituted into a temporal correlation matrix for end-to-end verification. Iterative optimization ensures the reconstructed data meets the requirements for inter-level temporal dependencies and numerical rationality, resolving the issues of traditional reconstruction methods' limited granularity and poor adaptability to the constraints of the entire transaction process. The final standardized transaction data output meets the usage requirements of the entire virtual power plant transaction process, providing high-quality data support for subsequent transaction decisions, settlement assessments, and other stages, significantly enhancing the utilization value and reliability of virtual power plant transaction data. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A structural block diagram of a computer device for a feature reconstruction method of virtual power plant transaction data in one embodiment;

[0049] Figure 2 This is a flowchart illustrating a method for reconstructing the features of virtual power plant transaction data in one embodiment;

[0050] Figure 3 This is a flowchart illustrating the data reconstruction and verification process based on anomaly classification results in one embodiment.

[0051] Figure 4 This is a flowchart illustrating the process of constructing a temporal correlation matrix based on correlation strength in one embodiment;

[0052] Figure 5 This is a flowchart illustrating the process of verifying reconstructed transaction data based on a temporal correlation matrix in one embodiment.

[0053] Figure 6 This is a schematic diagram illustrating the process of mapping virtual power plant transaction data in one embodiment;

[0054] Figure 7 This is a structural block diagram of a feature reconstruction device for virtual power plant transaction data in one embodiment. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0057] As the market-oriented reform of the current power market continues to deepen, virtual power plants, as an important vehicle for integrating distributed energy and improving the efficiency of power resource allocation, are increasingly participating in and influencing the power trading system, becoming a core component of the diversified participants in the power market. The power trading process of virtual power plants covers the entire lifecycle and multiple stages, including medium- and long-term contracts, short-term plan breakdown, spot market declaration and clearing, real-time settlement, and deviation assessment. The transaction data generated at each stage exhibits significant characteristics of multi-level, multi-time granularity, and multi-collection frequency, and there are close temporal correlations and hierarchical constraints between transaction data at different levels and time periods. The accuracy, completeness, and timeliness of transaction data directly determine the rationality of virtual power plant market declarations, the precision of settlement execution, and the effectiveness of risk management. As electricity market trading rules continue to improve, the frequency, complexity, and data volume of virtual power plant transactions are increasing simultaneously. Data is prone to various anomalies during collection, transmission, and processing due to factors such as equipment failure, transmission interference, and manual operation. If abnormal data cannot be detected and repaired in a timely manner, it will lead to problems such as deviations in trading plans, settlement errors, and even market violations. Therefore, conducting efficient anomaly detection and feature reconstruction of virtual power plant transaction data has become the key to ensuring the stable participation of virtual power plants in electricity market transactions and is also an urgent technical requirement for the digital development of the electricity market.

[0058] Currently, the industry mostly uses common data processing technologies for anomaly detection and reconstruction of power trading data. The core processing logic is as follows: perform simple standardized format conversion on the original power trading data, use a single numerical indicator or basic time series analysis method to determine data anomalies, use a unified manual review and handling method for detected anomalies, complete the reconstruction of abnormal data through a single numerical repair method, and perform only simple numerical rationality verification after reconstruction.

[0059] However, the above data processing methods use a single numerical indicator to determine anomalies and a single numerical repair method to reconstruct abnormal data. The accuracy and rationality of the reconstructed data cannot be effectively guaranteed, and the reconstructed data cannot directly support subsequent trading decisions.

[0060] In view of the above-mentioned technical problems, this application provides a feature reconstruction method for virtual power plant transaction data that can ensure that the reconstructed data meets the requirements of inter-level temporal dependency and numerical rationality. The following embodiments will specifically illustrate the feature reconstruction method for virtual power plant transaction data.

[0061] The feature reconstruction method for virtual power plant transaction data provided in this application embodiment can be applied to, for example... Figure 1 The computer device shown may be a server, and its internal structure diagram may be as follows. Figure 1As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores annual medium- and long-term contract trading data, monthly trading plan breakdown data, day-ahead electricity market declaration data, intraday spot market clearing data, real-time settlement data, and deviation assessment data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for reconstructing the characteristics of virtual power plant trading data.

[0062] Those skilled in the art will understand that Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0063] In one exemplary embodiment, such as Figure 2 As shown, a method for feature reconstruction of virtual power plant transaction data is provided. This embodiment illustrates the application of this method to computer equipment. In this embodiment, the method includes:

[0064] S201: Obtain the virtual power plant transaction data to be detected, map the virtual power plant transaction data, and generate an initial standardized transaction dataset.

[0065] Virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies; virtual power plant transaction data includes annual medium and long-term contract transaction data, monthly transaction plan breakdown data, day-ahead electricity market declaration data, intraday spot market clearing data, real-time settlement data, and deviation assessment data.

[0066] In the embodiments of this application, the computer device can collect annual medium- and long-term contract transaction data on an annual cycle, monthly transaction plan breakdown data on a monthly cycle, daily electricity market declaration data during 96 time periods, intraday spot market clearing data every 15 minutes, real-time settlement data every 5 minutes, and deviation assessment data between the daily and monthly cycles to obtain the virtual power plant transaction data to be tested. In some embodiments, unique transaction link identifiers and time sequence identifiers can be assigned to the collected data at various time granularities to ensure data traceability and that the collected data is compatible with subsequent time sequence reference axis mapping requirements.

[0067] In some embodiments, the computer device can uniformly map transaction data with different time granularities and different collection frequencies to the same time-series baseline axis, and complete the step-by-step decomposition of data at each level starting from the annual medium- and long-term contract transaction data based on the forward decomposition link, and complete the reverse feedback of data at each level starting from the real-time settlement data and deviation assessment data based on the time-series feedback link, to obtain the initial standardized transaction dataset.

[0068] S202, perform anomaly classification on the initial standardized transaction dataset, obtain anomaly classification results, and determine anomaly handling operations based on the anomaly classification results.

[0069] The anomaly classification results can be any of the following types: critical anomaly, severe anomaly, general anomaly, and minor anomaly. Each type of anomaly classification result has corresponding anomaly handling operations. The anomaly classification results are generated in a structured form, including a unique anomaly number, anomaly occurrence level, anomaly occurrence timestamp, original anomaly data value, anomaly level, a list of related levels, multi-dimensional time-series correlation coefficient values, and anomaly judgment criteria. The results support filtering and exporting by transaction stage, anomaly level, and time period.

[0070] In the embodiments of this application, the computer device can check each transaction data in the initial standardized transaction dataset one by one. The check items may include, but are not limited to, data being empty, electricity consumption being less than a first preset threshold or electricity price being less than a second preset threshold, timestamps being discontinuous or formatted incorrectly, violation of basic logic within the hierarchy, deviation from the historical range of the same period, etc. Abnormal data is determined through the check items, and the transaction hierarchy of the abnormal data is determined. Based on the mapping relationship between the transaction hierarchy and the abnormal classification, the abnormal classification result corresponding to the abnormal data is determined.

[0071] In some embodiments, the mapping relationship between transaction level and anomaly classification can be as follows: a fatal anomaly is an anomaly data appearing at the annual medium- and long-term contract level or the monthly transaction plan decomposition level, and the anomaly data is associated with 3 or more transaction levels; a severe anomaly is an anomaly data appearing at the day-ahead electricity market reporting level, and the anomaly data is associated with 2 transaction levels; a general anomaly is an anomaly data appearing at the intraday spot market clearing level or the real-time settlement level, and the anomaly data is associated with only 1 transaction level; a minor anomaly is an anomaly data that is a single sampling point data at the real-time settlement level or the deviation assessment level, and there is no cross-level or cross-time period association.

[0072] In some embodiments, the computer device, based on the mapping relationship between the determined anomaly classification result and the anomaly handling operation, determines the corresponding anomaly handling operation and executes the anomaly handling operation. The mapping relationship between the anomaly classification result and the anomaly handling operation can be as follows: a fatal anomaly triggers a red level 1 alarm, pushing alarm information to the transaction manager, risk control department manager, and transaction execution personnel through multiple channels, immediately suspending all declaration and execution operations for the corresponding transaction period, initiating a manual emergency review process, and resuming the transaction after the review is completed and approved by the risk control department; a serious anomaly triggers an orange level 2 alarm, pushing alarm information to the transaction execution personnel and transaction supervisor, initiating a mandatory review mechanism for the transaction plan, requiring the transaction execution personnel to complete data verification and plan adjustment within 2 hours, and submitting the adjusted plan to the market trading platform after review; a general anomaly triggers a yellow level 3 alarm, pushing alarm information to the data operation and maintenance personnel, initiating an automatic data review process, requiring the data operation and maintenance personnel to complete anomaly confirmation, marking, and archiving within 1 working day; a minor anomaly triggers a blue level 4 alarm, automatically recording the anomaly information without manual intervention, directly pushing it to the subsequent reconstruction process, and generating a minor anomaly statistical report daily.

[0073] S203. Based on the anomaly classification results, data reconstruction and verification are performed. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated.

[0074] The anomaly classification results correspond uniquely to the data reconstruction methods. Specifically, fatal anomalies are reconstructed using full-link time-series reconstruction, severe anomalies using multi-node joint reconstruction, general anomalies using single-level adjacent time-segment joint reconstruction, and minor anomalies using single-node numerical repair.

[0075] In the embodiments of this application, the computer device can select the corresponding data reconstruction method to reconstruct the abnormal data based on the anomaly classification result. The reconstruction process can be assisted by an adaptive multi-granularity reconstruction algorithm to obtain the reconstructed transaction data. In some embodiments, the reconstructed transaction data can be verified by a pre-established temporal correlation matrix. This temporal correlation matrix can be a matrix covering the entire transaction process, taking into account cross-level and intra-level temporal dependencies, and long- and short-term cycle information, constructed from quantitative indicators of hierarchical data dependencies. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated. If the reconstructed transaction data fails the verification, the weights of the adaptive multi-granularity reconstruction algorithm can be adjusted, and the abnormal data can be reconstructed again using the adjusted reconstruction algorithm to generate new reconstructed data. The new reconstructed data is then verified again based on the temporal correlation matrix. This process is repeated until the current reconstructed data passes the verification, generating the target standardized transaction dataset.

[0076] The aforementioned feature reconstruction method for virtual power plant transaction data acquires the virtual power plant transaction data to be detected, maps the data, and generates an initial standardized transaction dataset. This virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies. Anomaly classification is performed on the initial standardized transaction dataset to obtain anomaly classification results, and anomaly handling operations are determined based on these results. Data reconstruction and verification are performed based on the anomaly classification results. If the reconstructed transaction data passes verification, a target standardized transaction dataset is generated. This method achieves refined anomaly classification based on data characteristics such as time scale and impact scope, matching specific alarm levels and handling procedures for different levels of anomalies, thus achieving precise control of anomaly risks. This approach overcomes the limitations of traditional single-dimensional analysis, effectively avoiding missed or false anomaly detection. Simultaneously, the graded handling mechanism achieves differentiated and efficient anomaly response, quickly blocking the cross-level transmission of anomalies, significantly improving the comprehensiveness and accuracy of virtual power plant transaction data anomaly detection, and ensuring the orderly progress of the transaction process. By matching targeted reconstruction strategies to anomaly classification results, differentiated data reconstruction operations are implemented based on the anomaly level and impact scope. This ensures the reconstruction process aligns with the hierarchical characteristics and temporal patterns of virtual power plant transaction data, balancing reconstruction accuracy with data temporal consistency. Simultaneously, the reconstructed data is substituted into a temporal correlation matrix for end-to-end verification. Iterative optimization ensures the reconstructed data meets the requirements for inter-level temporal dependencies and numerical rationality, resolving the issues of traditional reconstruction methods' limited granularity and poor adaptability to the constraints of the entire transaction process. The final standardized transaction data output meets the usage requirements of the entire virtual power plant transaction process, providing high-quality data support for subsequent transaction decisions, settlement assessments, and other stages, significantly enhancing the utilization value and reliability of virtual power plant transaction data.

[0077] In one exemplary embodiment, such as Figure 3 As shown, data reconstruction and validation are performed based on the anomaly classification results, including:

[0078] S301, determine the association strength of each level of transaction data in the initial standardized transaction dataset according to the preset time-series association algorithm, and construct a time-series association matrix based on the association strength.

[0079] The preset temporal correlation algorithm can be represented by the following relation (1):

[0080] (1);

[0081] In the formula, The comprehensive temporal correlation coefficient between the transaction data of level i and level j; These are the forward constraint weight coefficients; The basic forward constraint strength of level i to level j; The time scale weight for the i-th level; The weight of the influence range of the ij-th level; Let be the backfeedback strength of level j to level i; The data reliability weight for the j-th level; The bias propagation weight is the weight of the ji-th level.

[0082] In the embodiments of this application, for each level of transaction data, the computer device can substitute the transaction data of that level with the correlation strength data of all levels into a preset time-series correlation algorithm for calculation, to obtain the correlation strength between the transaction data of that level and the transaction data of all levels. A time-series correlation matrix is ​​then established according to the correlation strength.

[0083] S302, Reconstruct the initial standardized transaction dataset based on the anomaly classification results to obtain the reconstructed transaction data.

[0084] In the embodiments of this application, after determining the anomaly classification result, the computer device selects the corresponding data reconstruction method to reconstruct the abnormal data in the initial standardized transaction dataset through the mapping relationship between the anomaly classification result and the data reconstruction method, thereby obtaining the reconstructed transaction data. In some embodiments, when the anomaly classification result is a fatal anomaly, full-link time-series reconstruction is used to reconstruct the abnormal data in the initial standardized transaction dataset, resulting in the reconstructed transaction data. When the anomaly classification result is a severe anomaly, multi-node joint reconstruction is used to reconstruct the abnormal data in the initial standardized transaction dataset, resulting in the reconstructed transaction data. When the anomaly classification result is a general anomaly, single-level adjacent time-segment joint reconstruction is used to reconstruct the abnormal data in the initial standardized transaction dataset, resulting in the reconstructed transaction data. When the anomaly classification result is a minor anomaly, single-node numerical repair is used to reconstruct the abnormal data in the initial standardized transaction dataset, resulting in the reconstructed transaction data.

[0085] S303, the reconstructed transaction data is verified based on the time-series correlation matrix to obtain the verification result.

[0086] The verification result is used to characterize whether the reconstructed transaction data passes the verification, including whether it passes or fails. The time-series correlation matrix can be a matrix that covers the entire transaction process, takes into account cross-level and intra-level time-series dependencies, and long- and short-term period information, and is constructed from quantitative indicators of hierarchical data dependencies.

[0087] In the embodiments of this application, a computer device can verify the reconstructed transaction data using a pre-established temporal correlation matrix. If the verification result is satisfactory, a target standardized transaction dataset is generated. If the reconstructed transaction data fails verification, the weights of the adaptive multi-granularity reconstruction algorithm can be adjusted. The weight adjustment can be set according to the actual scenario requirements. The adjusted reconstruction algorithm is then used to continue reconstructing the abnormal data, generating new reconstructed data. The new reconstructed data is then verified again based on the temporal correlation matrix. This process is repeated until the current reconstructed data passes verification, generating the target standardized transaction dataset.

[0088] The above method constructs a complete temporal correlation matrix covering both cross-level and intra-level data through a multi-dimensional temporal correlation coefficient algorithm, quantifying the temporal dependency strength of transaction data from multiple dimensions, and providing anomaly detection with correlation evidence throughout the entire process.

[0089] In one exemplary embodiment, such as Figure 4 As shown, a time-series correlation matrix is ​​constructed based on the correlation strength, including:

[0090] S401, construct cross-level association matrices and multiple intra-level association matrices based on the association strength.

[0091] The cross-level correlation matrix can be a 6×6 matrix, with rows and columns corresponding to the six levels of virtual power plant transactions: annual medium- and long-term contract level, monthly transaction plan decomposition level, day-ahead electricity market declaration level, intraday spot market clearing level, real-time settlement level, and deviation assessment level. The matrix elements represent the correlation strength calculated by a preset time-series correlation algorithm, used to quantify the time-series dependency strength between transaction data at different levels. The intra-level correlation matrix is ​​constructed based on the time granularity of each level and can include day-ahead declaration level sub-matrices and real-time settlement level sub-matrices. The day-ahead declaration level sub-matrice can have a dimension of 96×96, and the real-time settlement level sub-matrice can have a dimension of 288×288. The sub-matrices contain the time-series correlation coefficients of transaction data from different time periods within the same level, reflecting the time-series dependency relationships between time periods within a single level.

[0092] In the embodiments of this application, after the computer device determines the correlation strength of transaction data at each level in the initial standardized transaction dataset according to the preset time-series correlation algorithm, it can fill the correlation strength into the correlation matrix according to the layout of each correlation matrix to generate cross-level correlation matrix and multiple intra-level correlation matrices.

[0093] S402, for each level of the intra-level association matrix, determine the matrix mean of the intra-level association matrix.

[0094] In the embodiments of this application, for each level of the intra-level association matrix, the computer device can calculate the matrix mean of the intra-level association matrix according to the mean() function in MATLAB; it can also calculate the matrix mean of the intra-level association matrix according to the sum() and numel() functions in MATLAB; or it can calculate the matrix mean of the intra-level association matrix according to for loops and while loops.

[0095] S403, using the cross-level correlation matrix as a benchmark, superimposes the matrix mean of the intra-level correlation matrix with the corresponding elements of the cross-level correlation matrix to obtain the time-series correlation matrix.

[0096] In the embodiments of this application, the computer device determines the diagonal elements of the cross-level association matrix, that is, elements with the same row and column transaction level. It matches the transaction level of the intra-level association matrix with the transaction level of the diagonal elements, and adds the matrix mean of the intra-level association matrix to the successfully matched diagonal elements, that is, adds the matrix mean to the corresponding diagonal elements. After the above calculation operation is completed for all diagonal elements, the time-series association matrix is ​​obtained.

[0097] The above method constructs a complete temporal correlation matrix covering both cross-level and intra-level data through a multi-dimensional temporal correlation coefficient algorithm, quantifying the temporal dependency strength of transaction data from multiple dimensions, and providing anomaly detection with correlation evidence throughout the entire process.

[0098] In an exemplary embodiment, the initial standardized transaction dataset is reconstructed based on the anomaly classification results and the temporal correlation matrix to obtain reconstructed transaction data, including:

[0099] In the case of a fatal anomaly in the anomaly classification result, the transaction data at each level is reconstructed according to the preset reconstruction optimization algorithm and constraints to obtain the reconstructed transaction data.

[0100] The constraints include end-to-end constraints and total electricity constraints. End-to-end constraints refer to the entire time-series link of a virtual power plant transaction from the annual contract to real-time settlement as a whole, with the cross-level dependencies defined by the time-series correlation matrix and the total annual contract electricity volume as hard boundaries. End-to-end constraints can include cross-level time-series dependency constraints and time-series order constraints, while total electricity constraints can be total balance constraints, such as the sum of the electricity volumes at each level being consistent with the total annual contract electricity volume.

[0101] In some embodiments, the preset reconstruction optimization algorithm can be represented by the following relation (2):

[0102] (2);

[0103] In the formula, X rec For the reconstructed virtual power plant transaction data; X ori This refers to the preprocessed raw normal transaction data; Reconstruct weights for anomaly classification; R represents the reconstruction error term; λ is the timing consistency penalty coefficient; ij X represents the correlation strength between the i-th and j-th level transaction data calculated using a multi-dimensional temporal correlation coefficient algorithm. rec,i For the reconstructed transaction data of level i; X rec,j This refers to the transaction data at the j-th level after reconstruction.

[0104] In the embodiments of this application, when the anomaly classification result is a fatal anomaly, the computer device can use the full-link time-series reconstruction method to substitute the abnormal data in the initial standardized transaction dataset into a preset reconstruction optimization algorithm for calculation. The calculation process must meet the constraints to reconstruct the transaction data at each level and obtain the reconstructed transaction data.

[0105] In an exemplary embodiment, when the anomaly classification result is severe anomaly, the association level of the abnormal data in the time series correlation matrix is ​​determined, and the abnormal data is reconstructed based on the normal transaction data of the association level and a preset reconstruction optimization algorithm to obtain the reconstructed transaction data.

[0106] In the embodiments of this application, when the anomaly classification result is a severe anomaly, the computer device adopts a multi-node joint reconstruction method. First, it determines the association level of the abnormal data, that is, the association level is the transaction level to which the abnormal data belongs and its subordinate transaction levels. Normal transaction data is extracted from the association level. The normal transaction data and the abnormal data are simultaneously substituted into a preset reconstruction optimization algorithm for calculation. The calculation process must meet constraints to reconstruct the abnormal data, obtaining the reconstructed transaction data.

[0107] In an exemplary embodiment, when the anomaly classification result is a general anomaly, the normal time period data within the transaction level to which the anomaly data belongs is determined, and the anomaly data is reconstructed based on the normal time period data to obtain the reconstructed transaction data.

[0108] In the embodiments of this application, when the anomaly classification result is a general anomaly, the computer device adopts a single-level adjacent time period joint reconstruction method to determine the normal time period data adjacent to the abnormal data within the transaction level to which the abnormal data belongs, and reconstructs the abnormal data in accordance with the changing trend of the normal time period data to obtain the reconstructed transaction data.

[0109] In an exemplary embodiment, if the anomaly classification result is a minor anomaly, historical data from the same period of the anomaly is determined, and the anomaly is reconstructed based on the historical data to obtain the reconstructed transaction data.

[0110] In the embodiments of this application, when the anomaly classification result is a minor anomaly, the computer device adopts a single-node numerical repair method, that is, to determine the median of the same point and time type of the historical data of the anomaly, and repair the anomaly data by means of the median. The repair value must be within a reasonable range to obtain the reconstructed transaction data.

[0111] In one exemplary embodiment, such as Figure 5 As shown, the reconstructed transaction data is validated based on the time-series correlation matrix, and the validation results are as follows:

[0112] S501, substitute the reconstructed transaction data into the time series correlation matrix to determine the correlation strength and deviation between the reconstructed transaction data and the transaction data at the correlation level.

[0113] In the embodiments of this application, after the computer device calculates the reconstructed transaction data, it calculates the reconstructed transaction data based on a preset time-series correlation algorithm to obtain the correlation strength between the reconstructed transaction data and the transaction data at the correlation level. The matrix elements at corresponding positions in the time-series correlation matrix are then replaced using this correlation strength, and the deviation between the reconstructed electricity volume of the transaction level to which the reconstructed transaction data belongs and the preset electricity volume is calculated. Specifically, when the transaction level to which the reconstructed transaction data belongs is the annual medium-to-long-term contract level or the monthly transaction plan decomposition level, the deviation between the annual reconstructed electricity volume, the monthly reconstructed electricity volume, and the total annual contractually agreed electricity volume is calculated. When the transaction level to which the reconstructed transaction data belongs is the day-ahead declaration level, the deviation between the day-ahead declared reconstructed electricity volume and the monthly decomposed daily electricity volume is calculated. When the transaction level to which the reconstructed transaction data belongs is the intraday clearing level, the deviation between the intraday clearing reconstructed electricity volume and the day-ahead declaration period electricity volume is calculated. When the transaction level to which the reconstructed transaction data belongs is the real-time settlement level, the deviation between the real-time settlement reconstructed electricity volume and the intraday clearing period electricity volume is calculated.

[0114] S502, compare the correlation strength with the preset comprehensive temporal correlation strength, and compare the deviation with the preset deviation threshold to obtain the verification result.

[0115] The preset comprehensive time-series correlation strength can be 0.6. The preset deviation thresholds can include a deviation threshold of ±1% between the annual and monthly reconstructed electricity volume and the total contractually agreed amount, a deviation threshold of 2% between the daily reconstructed electricity volume declared before the date and the daily electricity volume of the monthly breakdown, a deviation threshold of ±5% between the reconstructed electricity volume cleared within the day and the electricity volume of the declared period before the date, and a deviation of ±8% between the reconstructed electricity volume settled in real time and the electricity volume of the cleared period within the day.

[0116] In the embodiments of this application, the computer device compares the correlation strength with a preset comprehensive time-series correlation strength and compares the deviation with a preset deviation threshold. If all correlation strengths in the time-series correlation matrix are not lower than 0.6, and the deviation between the annual and monthly reconstructed electricity volume and the contractually agreed total does not exceed ±1%, the deviation between the daily reconstructed electricity volume declared before the current day and the daily decomposed electricity volume does not exceed ±2%, the deviation between the daily cleared reconstructed electricity volume and the electricity volume declared before the current day does not exceed ±5%, and the deviation between the real-time settled reconstructed electricity volume and the electricity volume during the daily cleared period does not exceed ±8%, the verification result is determined to be a pass. If any correlation strength does not meet the condition of the preset comprehensive time-series correlation strength or any deviation does not meet the condition of the preset deviation threshold, the verification result is determined to be a fail.

[0117] In one exemplary embodiment, such as Figure 6As shown, the virtual power plant transaction data includes annual data, monthly data, day-ahead data, intraday data, real-time data, and deviation assessment data. The virtual power plant transaction data is mapped to generate an initial standardized transaction dataset, including:

[0118] S601 maps virtual power plant transaction data to the same time-series reference axis in a unified manner according to time sequence.

[0119] In the embodiments of this application, the computer device parses the virtual power plant transaction data to obtain the time sequence of the virtual power plant transaction data, and maps the virtual power plant transaction data to the same time sequence reference axis in the order from front to back according to the time sequence.

[0120] S602, based on the forward decomposition link, decomposes the annual data layer by layer downward according to the time and planning logic of power trading to obtain the initial trading data at different levels.

[0121] In the embodiments of this application, when the computer device processes virtual power plant transaction data based on the forward decomposition link, it starts with annual data (i.e., annual medium- and long-term contract data), decomposes the annual contract electricity volume into each month according to the monthly proportion agreed in the contract or the monthly plan specified by the user, and obtains the monthly contract target electricity volume and the initial value of the monthly plan; then, the monthly electricity volume is further decomposed into 96 time periods per day according to working days / non-working days and peak, flat and valley periods, to obtain the daily 96 time period declaration basic plan curve; then, the daily declaration basic plan curve is corrected with the actual declaration data to form the daily declaration time sequence; the daily declaration time sequence is compared with the intraday actual clearing time period by time, the deviation is recorded, and decomposed into 15-minute granularity to obtain the intraday clearing sequence, and a constraint relationship with the daily data is established; the 15-minute granular clearing data is linearly or interpolated to 5-minute granularity, aligned with the real-time settlement data, and a real-time settlement time sequence is generated. Among them, the monthly contract target electricity volume and the initial value of the monthly plan, the day-ahead declaration time sequence, the intraday clearing sequence, and the real-time settlement time sequence are initial transaction data at different levels.

[0122] S603, based on the time-series feedback link, uses real-time data and deviation assessment data as a benchmark to correct the initial transaction data layer by layer to obtain the feedback-received initial transaction data, and determines the initial standardized transaction dataset based on the feedback-received initial transaction data.

[0123] In the embodiments of this application, when the computer device processes virtual power plant transaction data based on the time-series feedback link, it first aggregates 5-minute real-time settlement data into 15-minute granularity, calculates the deviation rate with the intraday clearing data in the forward decomposition link, and obtains a 15-minute granularity deviation sequence. Based on the deviation assessment data, it reverse-engineers which time periods / dates have irregular deviations in intraday or day-ahead declaration data, and marks or corrects the intraday clearing or day-ahead declaration data for the corresponding time periods, obtaining the corrected intraday clearing sequence and day-ahead declaration sequence to be corrected. It calculates the difference between the corrected intraday clearing sequence and the original day-ahead declaration in the same 15-minute time period. If the difference exceeds the assessment threshold, it reverse-adjusts the reasonable confidence interval of the day-ahead declaration, obtaining the corrected day-ahead declaration constraints. It compares the actual monthly total with the original monthly plan, adjusts the decomposition coefficient or monthly plan decomposition logic for subsequent months, and obtains the adjusted monthly plan decomposition rules. Finally, it calculates the deviation between the total annual contract electricity and the actual total executed electricity, updates the annual contract decomposition ratio for the following year or contract adjustment period, obtains the annual contract correction curve and decomposition strategy, and integrates the information from different stages to obtain the initial standardized transaction dataset.

[0124] The aforementioned method maps virtual power plant transaction data to the same time-series baseline, addressing the industry pain point of difficulty in collaborative analysis of multi-granularity data. It establishes forward decomposition and time-series feedback links. The forward decomposition link starts with annual medium- and long-term contract transaction data to complete the hierarchical decomposition of data at each level. The time-series feedback link starts with real-time settlement data and deviation assessment data to complete the reverse feedback of data at each level. These two links are interconnected and precisely correspond to the time-series baseline, clearly defining the decomposition constraints and reverse verification relationships between data at each level. Based on this dual-link approach, preprocessing operations such as cleaning, deduplication, and completion are performed on the original data to remove invalid interference information. The resulting standardized transaction dataset can directly support subsequent time-series correlation matrix construction and anomaly detection without requiring secondary data processing, significantly improving overall data processing efficiency.

[0125] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0126] Based on the same inventive concept, this application also provides a feature reconstruction apparatus for virtual power plant transaction data to implement the feature reconstruction method for virtual power plant transaction data described above. The solution provided by this apparatus is similar to the implementation described in the above method. Therefore, the specific limitations of one or more embodiments of the feature reconstruction apparatus for virtual power plant transaction data provided below can be found in the limitations of the feature reconstruction method for virtual power plant transaction data described above, and will not be repeated here.

[0127] In one exemplary embodiment, such as Figure 7 As shown, a feature reconstruction device for virtual power plant transaction data is provided, including a mapping module 71, a determination module 72, and a reconstruction module 73, wherein:

[0128] The mapping module 71 is used to acquire the virtual power plant transaction data to be detected, map the virtual power plant transaction data, and generate an initial standardized transaction dataset; the virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies;

[0129] The determination module 72 is used to perform anomaly classification judgment on the initial standardized transaction dataset, obtain anomaly classification result, and determine anomaly handling operation based on the anomaly classification result;

[0130] The reconstruction module 73 is used to reconstruct and verify the data based on the anomaly classification results, and generate a target standardized transaction dataset if the reconstructed transaction data passes the verification.

[0131] In an exemplary embodiment, the above-mentioned reconstruction module 73 is specifically used for:

[0132] The correlation strength of each level of transaction data in the initial standardized transaction dataset is determined according to a preset time-series correlation algorithm, and a time-series correlation matrix is ​​constructed based on the correlation strength.

[0133] Based on the anomaly classification results and the temporal correlation matrix, the initial standardized transaction dataset is reconstructed to obtain the reconstructed transaction data.

[0134] The reconstructed transaction data is verified based on the time-series correlation matrix to obtain a verification result; wherein, the verification result is used to characterize whether the reconstructed transaction data passes the verification.

[0135] In an exemplary embodiment, the above-mentioned reconstruction module 73 is specifically used for:

[0136] Based on the correlation strength, construct cross-level correlation matrices and multiple intra-level correlation matrices respectively;

[0137] For each level of the association matrix, determine the mean of the association matrix within that level;

[0138] Using the cross-level correlation matrix as a benchmark, the mean of the intra-level correlation matrix is ​​superimposed with the corresponding elements of the cross-level correlation matrix to obtain the time-series correlation matrix.

[0139] In an exemplary embodiment, the above-mentioned reconstruction module 73 is specifically used for:

[0140] In the case where the anomaly classification result is a fatal anomaly, the transaction data at each level is reconstructed according to the preset reconstruction optimization algorithm and constraints to obtain the reconstructed transaction data; the constraints include end-to-end constraints and total power constraints.

[0141] If the anomaly classification result is a severe anomaly, the association level of the abnormal data in the time series correlation matrix is ​​determined, and the abnormal data is reconstructed based on the normal transaction data of the association level and the preset reconstruction optimization algorithm to obtain the reconstructed transaction data.

[0142] If the anomaly classification result is a general anomaly, determine the normal time period data within the transaction level to which the anomaly data belongs, and reconstruct the anomaly data based on the normal time period data to obtain the reconstructed transaction data.

[0143] If the anomaly classification result is a minor anomaly, the historical data of the same period of the anomaly is determined, and the anomaly is reconstructed based on the historical data of the same period of the anomaly to obtain the reconstructed transaction data.

[0144] In an exemplary embodiment, the above-mentioned reconstruction module 73 is specifically used for:

[0145] Substitute the reconstructed transaction data into the time-series correlation matrix to determine the correlation strength and deviation between the reconstructed transaction data and the transaction data at the correlation level;

[0146] The verification result is obtained by comparing the correlation strength with the preset comprehensive temporal correlation strength and comparing the deviation with the preset deviation threshold.

[0147] In an exemplary embodiment, the mapping module 71 described above is specifically used for:

[0148] The virtual power plant transaction data is uniformly mapped to the same time-series reference axis according to the time sequence.

[0149] Based on the forward decomposition link, the annual data is decomposed layer by layer according to the time and planning logic of power trading to obtain initial transaction data at different levels.

[0150] Based on the time-series feedback chain, and using real-time data and deviation assessment data as benchmarks, the initial transaction data is corrected layer by layer upwards to obtain the feedback-received initial transaction data, and the initial standardized transaction dataset is determined based on the feedback-received initial transaction data.

[0151] Each module in the aforementioned virtual power plant transaction data feature reconstruction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0152] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the steps in the above-described method embodiments.

[0153] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0154] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0155] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0156] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0157] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for reconstructing features of virtual power plant transaction data, characterized in that, The method includes: The virtual power plant transaction data to be detected is acquired, and the virtual power plant transaction data is mapped to generate an initial standardized transaction dataset; the virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies. Anomaly classification is performed on the initial standardized transaction dataset to obtain anomaly classification results, and anomaly handling operations are determined based on the anomaly classification results. Based on the anomaly classification results, data reconstruction and verification are performed. If the reconstructed transaction data passes the verification, a target standardized transaction dataset is generated.

2. The method according to claim 1, characterized in that, The data reconstruction and verification based on the anomaly classification results include: The correlation strength of each level of transaction data in the initial standardized transaction dataset is determined according to a preset time-series correlation algorithm, and a time-series correlation matrix is ​​constructed based on the correlation strength. Based on the anomaly classification results and the temporal correlation matrix, the initial standardized transaction dataset is reconstructed to obtain the reconstructed transaction data. The reconstructed transaction data is verified based on the time-series correlation matrix to obtain a verification result; wherein, the verification result is used to characterize whether the reconstructed transaction data passes the verification.

3. The method according to claim 2, characterized in that, The step of constructing a time-series correlation matrix based on the correlation strength includes: Based on the correlation strength, construct cross-level correlation matrices and multiple intra-level correlation matrices respectively; For each level of the association matrix, determine the mean of the association matrix within that level; Using the cross-level correlation matrix as a benchmark, the mean of the intra-level correlation matrix is ​​superimposed with the corresponding elements of the cross-level correlation matrix to obtain the time-series correlation matrix.

4. The method according to claim 2, characterized in that, The step of reconstructing the initial standardized transaction dataset based on the anomaly classification results and the temporal correlation matrix to obtain reconstructed transaction data includes: In the case where the anomaly classification result is a fatal anomaly, the transaction data at each level is reconstructed according to a preset reconstruction optimization algorithm and constraints to obtain the reconstructed transaction data; the constraints include end-to-end constraints and total power constraints. If the anomaly classification result is a severe anomaly, the association level of the abnormal data in the time series correlation matrix is ​​determined, and the abnormal data is reconstructed based on the normal transaction data of the association level and the preset reconstruction optimization algorithm to obtain the reconstructed transaction data. If the anomaly classification result is a general anomaly, determine the normal time period data within the transaction level to which the anomaly data belongs, and reconstruct the anomaly data based on the normal time period data to obtain the reconstructed transaction data. If the anomaly classification result is a minor anomaly, the historical data of the same period of the anomaly is determined, and the anomaly is reconstructed based on the historical data of the same period of the anomaly to obtain the reconstructed transaction data.

5. The method according to claim 2, characterized in that, The step of verifying the reconstructed transaction data based on the time-series correlation matrix to obtain the verification result includes: Substitute the reconstructed transaction data into the time-series correlation matrix to determine the correlation strength and deviation between the reconstructed transaction data and the transaction data at the correlation level; The verification result is obtained by comparing the correlation strength with the preset comprehensive temporal correlation strength and comparing the deviation with the preset deviation threshold.

6. The method according to any one of claims 1-5, characterized in that, The virtual power plant transaction data includes annual data, monthly data, day-ahead data, intraday data, real-time data, and deviation assessment data; the mapping of the virtual power plant transaction data to generate an initial standardized transaction dataset includes: The virtual power plant transaction data is uniformly mapped to the same time-series reference axis according to the time sequence. Based on the forward decomposition link, the annual data is decomposed layer by layer according to the time and planning logic of power trading to obtain initial transaction data at different levels. Based on the time-series feedback chain, and using real-time data and deviation assessment data as benchmarks, the initial transaction data is corrected layer by layer upwards to obtain the feedback-received initial transaction data, and the initial standardized transaction dataset is determined based on the feedback-received initial transaction data.

7. A feature reconstruction device for virtual power plant transaction data, characterized in that, The device includes: The mapping module is used to acquire the virtual power plant transaction data to be detected, map the virtual power plant transaction data, and generate an initial standardized transaction dataset; the virtual power plant transaction data includes multi-level transaction data with different time granularities and collection frequencies; The determination module is used to perform anomaly classification on the initial standardized transaction dataset, obtain anomaly classification results, and determine anomaly handling operations based on the anomaly classification results. The reconstruction module is used to reconstruct and verify the data based on the anomaly classification results, and generate a target standardized transaction dataset if the reconstructed transaction data passes the verification.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images