A power grid accident reconfiguration data fusion verification method, device and equipment and storage medium

Abstract

The application discloses a power grid accident reconstruction data fusion verification method and device, equipment and a storage medium, relates to the technical field of power system safety analysis, and the method comprises the following steps: preprocessing multi-source power grid data to obtain preprocessed multi-source data; calculating the random fluctuation uncertainty and system deviation uncertainty of each data source according to the preprocessed multi-source data, and obtaining the uncertainty data of each data source; calculating the correlation between multi-source uncertainties according to the uncertainty data of each data source, and obtaining multi-source uncertainty correlation data; adaptively performing evidence reasoning fusion on the preprocessed multi-source data according to the multi-source uncertainty correlation data, and obtaining a multi-source fusion result; and performing credibility evaluation according to the multi-source fusion result, obtaining a confidence interval and a credibility score, quantifying the uncertainty of multi-source data and adaptively performing evidence fusion, thereby reducing the uncertainty of power grid accident reconstruction and improving the credibility of the reconstruction result.

Description

Technical Field

[0001] This application relates to the field of power system security analysis technology, and in particular to a method, apparatus, equipment and storage medium for data fusion verification of power grid accident reconstruction. Background Technology

[0002] With the continuous growth of social electricity demand and the expansion of the power grid, the complexity and safety risks of power system operation are becoming increasingly prominent. Power grid accident reconstruction requires the comprehensive processing of heterogeneous data from multiple sources, including data acquisition and monitoring control systems, synchronous phasor measurement devices, and protection action signals, in order to quickly and accurately reconstruct the accident evolution process and support emergency decision-making.

[0003] In the process of multi-source data fusion, the propagation law of measurement uncertainty from each data source is often ignored, leading to significant deviations in the reconstruction results. Furthermore, the correlation between uncertainties from multiple sources is not effectively analyzed, potentially causing repeated calculations of errors originating from the same source during the fusion process. In addition, multi-source data fusion typically employs fixed weights or subjectively set weights, failing to adaptively adjust according to the uncertainty characteristics of each data source. Moreover, the fusion results lack quantitative evaluation methods such as confidence intervals and reliability scores, making it difficult to accurately assess the reliability of the reconstruction results. Summary of the Invention

[0004] The main purpose of this application is to provide a method, apparatus, equipment and storage medium for verifying data fusion in power grid accident reconstruction, which aims to solve the technical problem that the results of multi-source data fusion in power grid accident reconstruction are subject to large deviations due to uncertainty propagation and inter-source correlation, and lack of reliability quantitative assessment methods.

[0005] To achieve the above objectives, this application provides a method for data fusion verification of power grid accident reconfiguration, comprising the following steps: Preprocessing of multi-source power grid data yields preprocessed multi-source data. The random fluctuation uncertainty and systematic deviation uncertainty of each data source are calculated based on the preprocessed multi-source data to obtain the uncertainty data of each data source; The correlation between the uncertainties from each data source is calculated to obtain the multi-source uncertainty correlation data. Based on the multi-source uncertainty correlation data, adaptive evidence reasoning fusion is performed on the preprocessed multi-source data to obtain the multi-source fusion result; The credibility is assessed based on the multi-source fusion results to obtain the confidence interval and credibility score.

[0006] In one embodiment, the step of preprocessing multi-source power grid data to obtain preprocessed multi-source data includes: Collect multi-source power grid data; The average level and volatility of the time series data from each data source are calculated based on the multi-source data to obtain statistical characteristic data. Based on the statistical feature data, outliers in the multi-source power grid data are removed to obtain the filtered multi-source data; The filtered multi-source data is then filled with low-rank data to obtain the filled multi-source data. The completed multi-source data is then standardized to obtain preprocessed multi-source data.

[0007] In one embodiment, the step of performing low-rank completion on the filtered multi-source data to obtain completed multi-source data includes: The filtered multi-source data is used to construct an initial data matrix; Replace the positions corresponding to outliers in the initial data matrix with missing markers to obtain a matrix containing missing items; Singular value decomposition is performed on the matrix containing missing terms to obtain the left singular matrix, the singular value matrix, and the right singular matrix; The singular value matrix is ​​subjected to soft thresholding to obtain a truncated singular value matrix; The completed matrix is ​​reconstructed based on the left singular matrix, the truncated singular value matrix, and the right singular matrix to obtain the completed multi-source data.

[0008] In one embodiment, the step of calculating the random fluctuation uncertainty and systematic bias uncertainty of each data source based on the preprocessed multi-source data to obtain the uncertainty data of each data source includes: The first uncertainty is obtained by statistically analyzing the local fluctuations of the preprocessed multi-source data using a sliding window. The systematic bias of the preprocessed multi-source data is calculated based on the physical engineering model to obtain the second uncertainty. The total standard uncertainty is synthesized from the first uncertainty and the second uncertainty to obtain the uncertainty data of each data source.

[0009] In one embodiment, the step of calculating the correlation between multiple uncertainties based on the uncertainty data from each data source to obtain multi-source uncertainty correlation data includes: The Pearson correlation coefficient is calculated based on the uncertainty data of each data source to obtain the static correlation. The dynamic time bending distance is calculated based on the uncertainty data of each data source to obtain the dynamic correlation; Multi-source uncertainty correlation data are generated based on the static correlation and the dynamic correlation.

[0010] In one embodiment, the step of performing adaptive evidence reasoning fusion on the preprocessed multi-source data based on the multi-source uncertainty correlation data to obtain a multi-source fusion result includes: Based on the multi-source uncertainty correlation data, a basic probability allocation function is constructed to obtain the evidence from each data source; Based on the evidence from each data source, an improved particle swarm optimization algorithm is used to determine the adaptive weights of each data source, resulting in an adaptive weight allocation. The evidence from each data source is weighted and fused based on the adaptive weight allocation result to obtain a multi-source fusion result.

[0011] In one embodiment, the step of performing a credibility assessment based on the multi-source fusion result to obtain a confidence interval and a credibility score includes: Based on the multi-source fusion results, a basic probability allocation is determined, and probability allocation data is obtained. The trust function value is calculated based on the probability allocation data to obtain the trust level; The likelihood function value is calculated based on the probability allocation data to obtain the likelihood. The confidence interval is determined based on the confidence level and the likelihood level; A credibility score is calculated based on the multi-source fusion results to obtain the credibility score.

[0012] Furthermore, to achieve the above objectives, this application also proposes a power grid fault reconfiguration data fusion verification device, the device comprising: The data preprocessing module is used to preprocess multi-source power grid data to obtain preprocessed multi-source data; The uncertainty calculation module is used to calculate the random fluctuation uncertainty and systematic deviation uncertainty of each data source based on the preprocessed multi-source data, and obtain the uncertainty data of each data source. The correlation calculation module is used to calculate the correlation between multiple uncertainties based on the uncertainty data of each data source, and obtain multi-source uncertainty correlation data. The evidence fusion module is used to perform adaptive evidence reasoning fusion on the preprocessed multi-source data based on the multi-source uncertainty correlation data to obtain the multi-source fusion result; The credibility assessment module is used to assess credibility based on the multi-source fusion results and obtain confidence intervals and credibility scores.

[0013] In addition, to achieve the above objectives, this application also proposes a power grid accident reconfiguration data fusion verification device, the device comprising: a memory, a processor, and a power grid accident reconfiguration data fusion verification program stored in the memory and executable on the processor, the power grid accident reconfiguration data fusion verification program being configured to implement the steps of the power grid accident reconfiguration data fusion verification method as described above.

[0014] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, storing a power grid accident reconfiguration data fusion verification program, which, when executed by a processor, implements the steps of the power grid accident reconfiguration data fusion verification method as described above.

[0015] In addition, to achieve the above objectives, the present invention also provides a computer program product, which includes a power grid accident reconfiguration data fusion verification program. When the power grid accident reconfiguration data fusion verification program is executed by a processor, it implements the steps of the power grid accident reconfiguration data fusion verification method as described above.

[0016] This application preprocesses multi-source power grid data to eliminate outliers and missing items, laying a high-quality data foundation for subsequent analysis. Based on this, by calculating the random fluctuation uncertainty and systematic deviation uncertainty of each data source separately, the error level of each data source is accurately quantified. Furthermore, by analyzing the correlation between multi-source uncertainties, the common-source association and time lag characteristics of errors from different data sources are identified, thus avoiding repeated calculations of common-source errors during the fusion process. Subsequently, adaptive evidence reasoning fusion is performed on the multi-source data based on the aforementioned uncertainties and correlations, reasonably suppressing the weight contribution of data sources with high uncertainty in the fusion process, effectively reducing the propagation and amplification of uncertainty during fusion. Finally, by evaluating the credibility of the fusion results, confidence intervals and credibility scores are output, quantifying the reliability of the reconstruction results. Therefore, the accuracy of power grid accident reconstruction is improved, and the reliability of the reconstruction results can be effectively evaluated and output. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the first embodiment of the power grid accident reconstruction data fusion verification method of this application; Figure 2 This is a flowchart illustrating the overall process of the first embodiment of the power grid fault reconstruction data fusion verification method of this application. Figure 3 This is a flowchart illustrating the second embodiment of the power grid accident reconstruction data fusion verification method of this application; Figure 4 This is a flowchart illustrating the third embodiment of the power grid accident reconstruction data fusion verification method of this application; Figure 5 This is a structural block diagram of the first embodiment of the power grid accident reconstruction data fusion verification device of this application; Figure 6 This is a schematic diagram of the power grid accident reconstruction data fusion verification device of this application.

[0020] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of this application and are not intended to limit this application.

[0022] It should be noted that the executing entity of the embodiments of this application can be a computing service device with data processing, network communication, and program execution functions, such as a personal computer or mobile phone, or an electronic device capable of realizing the above functions, such as the aforementioned power grid accident reconstruction data fusion verification device. The following embodiments will be described using the power grid accident reconstruction data fusion verification device as an example.

[0023] This application provides a method for data fusion and verification of power grid fault reconfiguration, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the power grid accident reconstruction data fusion verification method of this application.

[0024] In this embodiment, the power grid fault reconstruction data fusion verification method includes the following steps: Step S10: Preprocess the multi-source power grid data to obtain preprocessed multi-source data; It should be noted that during the data preparation phase of power grid accident reconstruction, the raw data from different monitoring methods need to be processed to improve their quality.

[0025] like Figure 2 As shown, Figure 2The overall flowchart of this scheme is as follows: After the raw data is completed using the SVT algorithm, it enters the stage of constructing a multi-source measurement uncertainty quantification model. Then, it undergoes multi-source data fusion modeling based on evidence reasoning theory, and finally obtains the fusion result evaluation and output. Among them, the data completion using the SVT algorithm includes outlier detection, construction of missing matrices, and reconstruction of the completion matrix; the construction of the multi-source measurement uncertainty quantification model includes Type A uncertainty, Type B uncertainty, and combined uncertainty; the multi-source data fusion modeling based on evidence reasoning theory includes establishing an identification framework and DS synthesis.

[0026] It should be noted that multi-source power grid data refers to a collection of data from different monitoring methods in power grid dispatching, including data acquisition and monitoring control system data, synchronous phasor measurement device data, and protection action signals, etc., which are used to reflect multi-dimensional information about the power grid's operating status. Preprocessing refers to the process of cleaning, filtering, and completing the raw data, namely, eliminating data noise, removing outliers, and filling in missing items, which is used to improve data quality. Preprocessed multi-source data refers to multi-source power grid data that has undergone cleaning, filtering, completion, and standardization, i.e., a data set with a uniform format and reliable quality, which is used to support subsequent uncertainty calculations.

[0027] Understandably, preprocessing multi-source power grid data to obtain preprocessed multi-source data refers to collecting power grid operation data from different monitoring methods, sequentially performing outlier removal, missing data completion, and standardization processing to obtain preprocessed multi-source data with a unified format and reliable quality.

[0028] Preprocessing eliminates noise and missing terms in multi-source power grid data, providing a high-quality data foundation for subsequent uncertainty calculations.

[0029] Step S20: Calculate the random fluctuation uncertainty and systematic bias uncertainty of each data source based on the preprocessed multi-source data to obtain the uncertainty data of each data source; It should be noted that during the uncertainty quantification phase of power grid accident reconfiguration, it is necessary to assess the magnitude of random errors and systematic errors in each data source separately.

[0030] It should be noted that each data source refers to an independent data channel from different monitoring methods within a multi-source power grid data set, namely, the data acquisition and monitoring control system channel, the synchronous phasor measurement device channel, and the protection action signal channel, etc., which are used to provide power grid operation information from different dimensions. Random fluctuation uncertainty refers to the uncontrollable but statistically significant error component caused by sensor noise and minor environmental fluctuations; it is an uncertainty index reflecting the random fluctuation characteristics of the data and is used to quantify the level of random error in the data. Systematic deviation uncertainty refers to the controllable and modelable error component caused by installation errors, zero-point drift, and nonlinear factors; it is an error index reflecting the inherent deviation of the measuring device and is used to quantify the level of systematic error in the data. Uncertainty data refers to the total standard uncertainty set obtained after uncertainty evaluation of each data source; it is structured data containing comprehensive error information from all data sources, used to support subsequent correlation analysis and fusion processing.

[0031] Understandably, calculating the random fluctuation uncertainty and systematic bias uncertainty of each data source based on the preprocessed multi-source data to obtain the uncertainty data of each data source means assessing the random fluctuation uncertainty and systematic bias uncertainty of each data source separately, and then combining the two types of uncertainty into the total standard uncertainty, thereby obtaining the uncertainty data of each data source.

[0032] By calculating the random fluctuation uncertainty and the systematic deviation uncertainty separately and synthesizing the total standard uncertainty, the error level of each data source can be fully quantified, providing a quantitative basis for weight allocation in the subsequent fusion process.

[0033] In the specific implementation, an A / B uncertainty model is constructed using the preprocessed data, for each data source. Type A and Type B uncertainties were calculated separately, and the total standard uncertainty was synthesized. Type A uncertainty refers to random fluctuations (uncontrollable but statistically significant), mainly from sensor noise and minor environmental fluctuations, and was calculated using a sliding window statistical method. Type B uncertainty refers to systematic deviations (controllable but requiring modeling), mainly from installation errors, zero-point drift, and nonlinearity, and was calculated using a physical / engineering model.

[0034] In one feasible implementation, step S20 includes: Step S201: Obtain the first uncertainty by statistically analyzing the local fluctuations of the preprocessed multi-source data through a sliding window; It should be noted that in the stage of assessing the random fluctuation uncertainty of power grid accident reconfiguration, it is necessary to estimate the short-term fluctuation characteristics of the data through local statistical methods.

[0035] It should be noted that a sliding window refers to a fixed-length data interval that moves continuously across a time series; that is, a data segment of a predetermined duration is extracted from the current time. It is used to estimate local statistical characteristics. Local fluctuation refers to the dispersion of data within the sliding window relative to the local average level; that is, a statistical measure reflecting the random fluctuation characteristics of the data in a short period of time, used to characterize the short-term stability of the data. The first uncertainty refers to the random fluctuation uncertainty obtained through the sliding window statistics; that is, a quantitative index reflecting the level of local random error in the data, used to characterize the short-term fluctuation characteristics of the data.

[0036] Understandably, obtaining the first uncertainty by using a sliding window to statistically analyze the local fluctuations of preprocessed multi-source data involves moving a fixed-length window across the time series of each data source and calculating the dispersion of the data within each window relative to the local average level, thereby obtaining the first uncertainty that reflects the random fluctuation characteristics of the data.

[0037] By using sliding window statistics, the local random fluctuation characteristics of each data source can be dynamically estimated, providing a random error component for the overall standard uncertainty.

[0038] In the specific implementation, the input is the nth data source time series after completion. Using a length of Using a sliding window to estimate local fluctuations:

[0039]

[0040] in For Centered The standard deviation of the samples within the point window. The larger the value, the more intense the random fluctuations around time t; This represents the mean of the nth data source over w time points centered at time t. Let represent the Type A uncertainty of the nth data source at time t.

[0041] Step S202: Calculate the systematic bias of the preprocessed multi-source data based on the physical engineering model to obtain the second uncertainty; It should be noted that during the stage of assessing the system deviation uncertainty in power grid accident reconstruction, it is necessary to calculate the inherent deviation of the measuring device based on physical laws.

[0042] It should be noted that the physical engineering model refers to a mathematical description based on the physical characteristics of the equipment and engineering experience; that is, a parameterized expression reflecting the inherent error law of the measuring device, used to calculate the theoretical value of the systematic deviation. The systematic deviation refers to the fixed deviation between the measured value and the true value, i.e., the deterministic error component caused by the inherent characteristics of the measuring device, used to characterize the magnitude of the systematic error in the data. The second uncertainty refers to the systematic deviation uncertainty calculated according to the physical engineering model, i.e., a quantitative index reflecting the level of systematic error in the data, used to characterize the inherent deviation characteristics of the data.

[0043] Understandably, calculating the systematic deviation of preprocessed multi-source data based on a physical engineering model to obtain the second uncertainty means establishing an error model based on the physical characteristics of the measuring device and engineering experience, substituting the preprocessed multi-source data into the model to calculate the systematic deviation, thereby obtaining the second uncertainty that reflects the inherent deviation characteristics of the data.

[0044] Through calculations using physical engineering models, the systematic deviations of each data source can be theoretically estimated, providing systematic error components for the overall standard uncertainty.

[0045] In practical implementation, the total standard uncertainty of a single data source is synthesized as follows:

[0046] Total uncertainty sequence of all N data sources .

[0047] Step S203: Combine the first uncertainty and the second uncertainty to obtain the total standard uncertainty and the uncertainty data of each data source.

[0048] It should be noted that in the uncertainty synthesis stage of power grid accident reconstruction, random errors and systematic errors need to be combined into a comprehensive index.

[0049] It should be noted that total standard uncertainty refers to the combined index of random fluctuation uncertainty and systematic deviation uncertainty, that is, the final result obtained by combining the two types of uncertainty through the root-square method, which is used to comprehensively characterize the uncertainty level of the data. Uncertainty data from each data source refers to the set of total standard uncertainties from each data source, that is, structured data containing comprehensive error information from all data sources, which is used to support subsequent correlation analysis and fusion processing.

[0050] Understandably, obtaining the uncertainty data of each data source by combining the first and second uncertainties to obtain the total standard uncertainty means combining the first and second uncertainties using the root-squaring method to obtain the total standard uncertainty of each data source, thereby obtaining the uncertainty data of each data source.

[0051] By synthesizing the total standard uncertainty, the random and systematic errors from various data sources are combined into a single index, which facilitates subsequent correlation analysis and comparison among multiple sources of uncertainty.

[0052] Step S30: Calculate the correlation between multiple uncertainties based on the uncertainty data from each data source to obtain the multi-source uncertainty correlation data; It should be noted that in the multi-source error correlation analysis stage of power grid accident reconstruction, it is necessary to identify the correlation characteristics between errors from different data sources.

[0053] It should be noted that multi-source uncertainty correlation data refers to a quantitative description of the degree of correlation between uncertainties from different data sources. In other words, it is a set of statistical indicators that reflect the commonality of errors and time lag relationships among various data sources. It is used to guide the deduplication of errors in the subsequent fusion process.

[0054] Understandably, calculating the correlation between multiple uncertainties based on the uncertainty data from various data sources to obtain multi-source uncertainty correlation data means calculating the Pearson correlation coefficient and dynamic time curvature distance for the uncertainty sequences of each data source, thereby obtaining multi-source uncertainty correlation data that reflects the static homology and dynamic time lag characteristics between multiple uncertainties.

[0055] By analyzing the correlation between multiple sources of uncertainty, the common-source association and time lag characteristics of errors from different data sources can be identified, thereby avoiding the repeated calculation of common-source errors during the fusion process.

[0056] Step S40: Perform adaptive evidence reasoning fusion on the preprocessed multi-source data based on the multi-source uncertainty correlation data to obtain the multi-source fusion result; It should be noted that in the multi-source data integration stage of power grid accident reconstruction, the fusion weights of each data source need to be dynamically adjusted according to the uncertainty characteristics. To overcome the sensitivity of DS theory to highly conflicting evidence, an adaptive weighting mechanism is introduced.

[0057] It should be noted that adaptive evidence-based reasoning fusion refers to a data synthesis method that dynamically adjusts the fusion strategy according to the characteristics of multi-source uncertainty. Specifically, it is a multi-source information fusion process based on evidence-based reasoning theory and combined with adaptive weight optimization, used to reduce the impact of uncertainty propagation on the fusion result. The multi-source fusion result refers to the comprehensive data output obtained after adaptive evidence-based reasoning fusion, i.e., the final data after fusing information from various data sources and suppressing the influence of high-uncertainty data sources. It is used to characterize the fused power grid fault reconstruction information.

[0058] Understandably, the process of adaptively fusing preprocessed multi-source data with multi-source uncertainty correlation data to obtain multi-source fusion results refers to constructing an evidence reasoning framework based on multi-source uncertainty correlation data, determining the weights of each data source through an adaptive optimization algorithm, and then performing weighted fusion on each data source to obtain multi-source fusion results with reduced uncertainty propagation impact.

[0059] Through adaptive evidence reasoning fusion, the weight contribution of data sources with high uncertainty in the fusion process is reasonably suppressed, which effectively reduces the propagation and amplification of uncertainty in the fusion process and improves the accuracy of the fusion results.

[0060] In one feasible implementation, step S40 includes: Step S401: Construct a basic probability allocation function based on the multi-source uncertainty correlation data to obtain the evidence body of each data source; It should be noted that in the evidence construction stage of power grid accident reconstruction, it is necessary to transform the uncertainty information from various data sources into a form of evidence that can be fused.

[0061] It should be noted that the basic probability assignment function is a mathematical tool in evidence reasoning theory that transforms uncertainty into probability assignment. In other words, it maps the uncertainty information from various data sources to the degree of trust in the evidence body, and is used to construct the basic evidence required for fusion. The evidence body refers to the set of trust assignments from various data sources output by the basic probability assignment function; that is, structured information representing the degree of support a single data source provides for each proposition, and it is used in subsequent weighted fusion calculations.

[0062] Understandably, constructing a basic probability allocation function based on multi-source uncertainty correlation data to obtain the evidence body of each data source means establishing a basic probability allocation function for each data source based on multi-source uncertainty correlation data, mapping the uncertainty information of each data source to a trust allocation set, thereby obtaining the evidence body of each data source.

[0063] By constructing a basic probability allocation function, the uncertainty information from each data source is transformed into a standardized form of evidence, providing a unified data foundation for subsequent adaptive weight determination and weighted fusion.

[0064] In the specific implementation, the initial weights are:

[0065] The initial weights for the nth data source; In the objective function design, the error between the fusion result and the reference value (such as the verification value of the standard material) is minimized:

[0066] in The fusion mean; The value is a reference value, and α represents smoothness. The weight vector from the previous optimization round:

[0067] in, The smaller the value, the more accurate the fusion result and the more stable the weight.

[0068] Step S402: Based on the evidence from each data source, determine the adaptive weights of each data source using an improved particle swarm optimization algorithm to obtain the adaptive weight allocation results; It should be noted that in the adaptive weight optimization stage of power grid fault reconstruction, intelligent algorithms are needed to find the optimal fusion weights for each data source.

[0069] It should be noted that the improved particle swarm optimization algorithm refers to an adaptive optimization method that improves upon the traditional particle swarm optimization algorithm. Its full English name is Particle Swarm Optimization, abbreviated as PSO. It's an optimization algorithm that finds the optimal solution by simulating the foraging behavior of bird flocks and introducing an adaptive mechanism. It is used to determine the optimal fusion weights for each data source. Adaptive weights are fusion weight coefficients dynamically calculated based on the characteristics of the evidence from each data source; they reflect the relative importance of each data source in the fusion process and are used to distinguish the contribution levels of different data sources. The adaptive weight allocation result refers to the set of optimal weights for each data source calculated by the improved particle swarm optimization algorithm. It contains the configuration information of the adaptive weight values ​​for all data sources and is used to guide subsequent weighted fusion operations.

[0070] Understandably, determining the adaptive weights of each data source based on the evidence from each data source using an improved particle swarm optimization algorithm, and obtaining the adaptive weight allocation result, means taking the evidence from each data source as input and using the improved particle swarm optimization algorithm to iteratively search for the optimal weight combination, thereby obtaining the adaptive weight allocation result that reflects the relative importance of each data source.

[0071] By improving the particle swarm optimization algorithm to determine adaptive weights, the fusion weights of each data source can be dynamically optimized according to the characteristics of its evidence body, avoiding fusion bias caused by fixed weights or subjectively set weights.

[0072] Step S403: Based on the adaptive weight allocation result, the evidence from each data source is weighted and fused to obtain the multi-source fusion result.

[0073] It should be noted that in the weighted synthesis phase of power grid accident reconstruction, the evidence from each data source needs to be merged based on the optimized weights.

[0074] It should be noted that weighted fusion refers to the process of weighting and synthesizing evidence from various data sources based on the results of adaptive weight allocation. In other words, it is an operation that merges multiple evidences according to the importance coefficients of each data source, which is used to generate the final multi-source fusion result.

[0075] Understandably, weighted fusion of evidence from each data source based on adaptive weight allocation results to obtain multi-source fusion results means performing weighted comprehensive calculations on the evidence from each data source according to the adaptive weight allocation results, merging the trust allocation information of multiple data sources, and thus obtaining multi-source fusion results that integrate information from each data source.

[0076] By using weighted fusion, information from each data source is synthesized in a differentiated manner based on their reliability differences, thereby suppressing the impact of high-uncertainty data sources and improving the overall reliability of the fusion results.

[0077] In the specific implementation, particles represent weight vectors:

[0078] Satisfy constraints:

[0079]

[0080] Update speed and location:

[0081]

[0082] in Weighted inertia (taken as 0.5~0.9, representing the ability of a particle to maintain its original velocity). The larger the value, the more likely it is to perform a global search. , For learning factors; Represents the historical best position of the i-th particle (making The smallest weight vector); Represents the global optimal position of all particles (in the current iteration) The smallest weight vector); This represents the position of the i-th particle in the k-th round. The iteration stops when the number of iterations reaches the upper limit or the change in J(w) is less than the threshold, and the globally optimal weights are output. .

[0083] Step S50: Based on the multi-source fusion results, perform a credibility assessment to obtain the confidence interval and credibility score.

[0084] It should be noted that during the reliability evaluation phase of the power grid accident reconstruction results, a quantitative analysis of the credibility of the fusion results is required.

[0085] It should be noted that credibility assessment refers to the process of quantitatively analyzing the reliability of multi-source fusion results. Specifically, it's an evaluation method that determines the confidence interval of the fusion results by calculating trust level and likelihood, and it characterizes the credibility of the fusion results. Credibility score, on the other hand, is an indicator that quantitatively scores the overall credibility of the fusion results; it's a numerical evaluation reflecting the strength of the fusion results' reliability, and it's used to intuitively characterize the quality level of the fusion results.

[0086] Understandably, assessing the credibility of multi-source fusion results to obtain confidence intervals and credibility scores involves conducting evidence reasoning analysis on the multi-source fusion results, calculating their trust and likelihood to determine confidence intervals, and calculating the overall credibility score, thereby obtaining confidence intervals and credibility scores that quantitatively characterize the reliability of the fusion results.

[0087] By conducting a credibility assessment of the fusion results, the reliability of the fusion results can be quantitatively expressed, enabling dispatchers to accurately grasp the credibility level of the reconstruction results and providing a reliable quantitative basis for emergency decision-making.

[0088] In one feasible implementation, step S50 includes: Step S501: Determine the basic probability allocation based on the multi-source fusion results to obtain probability allocation data; It should be noted that in the probability allocation extraction stage of power grid accident reconstruction, it is necessary to separate the basic probability information used for credibility calculation from the fusion results.

[0089] It should be noted that basic probability allocation refers to the fundamental operation in evidence reasoning theory that maps uncertainty into probability values. This involves transforming the trust information in the fusion results into a computable probabilistic form, which supports the calculation of trust and likelihood. Probability allocation data refers to the basic probability allocation information extracted from the multi-source fusion results; this is the foundational data used for subsequent trust and likelihood calculations, serving as input for credibility assessment.

[0090] Understandably, determining the basic probability allocation based on the multi-source fusion results and obtaining probability allocation data means extracting trust allocation information from the multi-source fusion results, transforming it into a basic probability allocation form, and thus obtaining probability allocation data for subsequent trust calculations.

[0091] By determining the basic probability assignment, the trust information in the multi-source fusion results is transformed into a standardized probability form, providing a unified data foundation for subsequent calculations of trust and likelihood.

[0092] Step S502: Calculate the trust function value based on the probability allocation data to obtain the trust level; It should be noted that during the trust lower bound calculation stage of power grid accident reconfiguration, it is necessary to assess the minimum level of trust that the fusion results provide for the proposition.

[0093] It should be noted that the trust function value refers to the lower limit of the degree of trust in a proposition in evidence reasoning theory. In other words, it is an indicator reflecting the minimum degree of trust that the fusion result provides for a proposition, and it is used to determine the lower limit of the confidence interval. The degree of trust refers to the lower limit of the credibility of the fusion result, characterized by the trust function value. It is a quantitative value reflecting the minimum level of assurance of the fusion result, and it is used in determining the confidence interval.

[0094] Understandably, calculating the trust function value based on the probability allocation data to obtain the trust degree means substituting the probability allocation data into the trust function for calculation, obtaining the minimum value of the degree of support of the fusion result for each proposition, and thus obtaining the trust degree that reflects the lower limit of the credibility of the fusion result.

[0095] By calculating the trust function value, the minimum level of trust in the fusion result for each proposition can be quantified, providing a basis for determining the lower limit of the confidence interval.

[0096] Step S503: Calculate the likelihood function value based on the probability allocation data to obtain the likelihood. It should be noted that during the trust ceiling calculation phase of power grid accident reconfiguration, it is necessary to assess the maximum degree of trust that the fusion results provide for the proposition.

[0097] It should be noted that the likelihood function value refers to the upper limit of the degree of trust in a proposition in evidence reasoning theory; that is, it is an indicator reflecting the maximum degree of trust that the fusion result supports for a proposition, and it is used to determine the upper limit of the confidence interval. Likelihood, on the other hand, refers to the upper limit of the credibility of the fusion result, characterized by the likelihood function value; that is, it is a quantitative value reflecting the highest possible level of the fusion result, and it is used in determining the confidence interval.

[0098] Understandably, calculating the likelihood function value based on the probability allocation data to obtain the likelihood means substituting the probability allocation data into the likelihood function for calculation, obtaining the maximum value of the degree of support of the fusion result for each proposition, and thus obtaining the likelihood that reflects the upper limit of the credibility of the fusion result.

[0099] By calculating the likelihood function value, the maximum confidence level of the fusion result in supporting each proposition can be quantified, providing a basis for determining the upper limit of the confidence interval.

[0100] Step S504: Determine the confidence interval based on the confidence level and likelihood. It should be noted that in the confidence interval determination stage of power grid accident reconstruction, the fluctuation range of the result needs to be obtained by combining the upper and lower limits of confidence.

[0101] It should be noted that the confidence interval refers to the upper and lower bounds of the possible values ​​of the fusion result, that is, the range of result fluctuation determined by the confidence level and the likelihood, which is used to quantify the range of uncertainty of the fusion result.

[0102] Understandably, determining the confidence interval based on the confidence level and likelihood means using the calculated confidence level as the lower limit and the likelihood level as the upper limit to construct the range of possible values ​​of the fusion result, thereby obtaining a confidence interval that quantitatively represents the uncertainty of the fusion result.

[0103] By determining the confidence interval through a combination of trust level and likelihood, the range of uncertainty in the fusion result can be clearly defined, providing schedulers with a quantitative basis for the reliability of the result.

[0104] Step S505: Calculate the credibility score based on the multi-source fusion results to obtain the credibility score.

[0105] It should be noted that in the overall credibility quantification stage of power grid accident reconstruction, a comprehensive score needs to be given to the fusion results.

[0106] It should be noted that the credibility score is an indicator that quantifies the overall credibility of the fusion results, that is, a numerical evaluation that reflects the strength of the reliability of the fusion results, and is used to intuitively characterize the quality level of the fusion results.

[0107] Understandably, calculating a credibility score based on the multi-source fusion results means conducting a comprehensive analysis and quantitative evaluation of the multi-source fusion results, calculating a numerical index that reflects the strength of its overall reliability, and thus obtaining a credibility score used to intuitively characterize the quality of the fusion results.

[0108] By calculating the credibility score, the overall reliability of the fusion result can be intuitively expressed by a single numerical value, making it easier for schedulers to quickly judge the quality level of the reconstruction result.

[0109] In the specific implementation, input fusion confidence allocation ; Let the probability distribution after fusion be... , It is the value of the k-th feature / index to be evaluated in the input fusion confidence assignment m, all Together, they constitute the set of discrete values ​​to be analyzed. Find the smallest closed interval. satisfy:

[0110] Expected value:

[0111] in Fusion confidence is the probability distribution.

[0112] The credibility score is based on three indicators as follows:

[0113]

[0114] in, DS conflict factor (the smaller the better); The standard uncertainty after fusion (the smaller the better); The mean of the Pearson correlation coefficient among all data sources is the average of the Pearson correlation coefficients among the data sources. The larger the size, the more synchronized the multi-source data. Adjustable weights (equal weights are recommended, or adjustments can be made based on the scenario). For high reliability; if For medium reliability; If the information is deemed unreliable, an alarm will be triggered, prompting a manual verification.

[0115] This embodiment calculates the random fluctuation uncertainty and systematic deviation uncertainty of each data source separately, accurately quantifying the error characteristics of each data source. Simultaneously, by analyzing the correlation between multi-source uncertainties, the common-source association and time lag characteristics of errors from different data sources are effectively identified, avoiding redundant calculations during the fusion process. Subsequently, adaptive evidence reasoning fusion is performed on the preprocessed multi-source data based on the multi-source uncertainty correlation data, allowing the fusion weights of each data source to adaptively adjust according to their uncertainty characteristics. This effectively suppresses the weight contribution of high-uncertainty data sources and reduces the propagation and amplification of uncertainty during the fusion process. Finally, by evaluating the credibility of the multi-source fusion results, confidence intervals and credibility scores are output, allowing the reliability of the fusion results to be quantitatively expressed. Therefore, this embodiment improves the accuracy of power grid accident reconstruction and enables the reliability of the reconstruction results to be effectively assessed.

[0116] Reference Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the power grid accident reconstruction data fusion verification method of this application.

[0117] In one feasible implementation, step S10 may include: Step S101: Collect multi-source power grid data; It should be noted that during the data acquisition phase of power grid accident reconstruction, it is necessary to obtain raw operational data from different monitoring methods.

[0118] It should be noted that multi-source power grid data refers to a collection of data from different monitoring methods in power grid dispatching, namely, data from data acquisition and monitoring control systems, data from synchronous phasor measurement devices, and protection action signals, which are used to reflect multi-dimensional information about the power grid's operating status.

[0119] As is understandable, collecting multi-source power grid data refers to obtaining power grid operation data from data acquisition and monitoring control systems, synchronous phasor measurement devices, and protection action signal acquisition terminals, thereby obtaining multi-source power grid data containing multi-dimensional power grid status information.

[0120] By collecting multi-source power grid data, a comprehensive raw data foundation was provided for subsequent accident reconstruction.

[0121] Step S102: Calculate the average level and volatility of the time series data from each data source based on the multi-source data to obtain statistical characteristic data; It should be noted that during the data statistics phase of power grid accident reconstruction, it is necessary to analyze the distribution characteristics of the time series of each data source.

[0122] It should be noted that statistical characteristic data refers to a set of indicators obtained through statistical calculations that reflect the central tendency and dispersion of data. This is structured data containing the average level and volatility of time series data from various data sources, used to identify abnormal patterns in the data. Various data sources refer to independent data channels from different monitoring methods in multi-source power grid data, such as data acquisition and monitoring control system channels, synchronous phasor measurement device channels, and protection action signal channels, which provide different dimensions of power grid operation information. Furthermore, the average level refers to a measure of the central tendency of the data, a statistical measure reflecting the overall size level of the data, used to characterize the baseline level of the data. Further, volatility refers to a measure of the dispersion of the data, a statistical measure reflecting the magnitude of fluctuations around the average level, used to characterize the stability of the data.

[0123] Understandably, calculating the average level and volatility of time series data from multiple data sources to obtain statistical characteristic data means calculating the average level and volatility of time series data from each data source separately, thereby obtaining statistical characteristic data that reflects the distribution characteristics of data from each data source.

[0124] By calculating statistical feature data, the data distribution characteristics of each data source can be quantified, providing a basis for subsequent outlier identification.

[0125] Step S103: Remove outliers from the multi-source power grid data based on statistical characteristic data to obtain the filtered multi-source data; It should be noted that during the data screening stage of power grid accident reconstruction, data points that deviate significantly from the normal range need to be removed based on statistical characteristics.

[0126] It should be noted that outliers refer to isolated data points that significantly deviate from the normal distribution range of the data. These are erroneous measurements caused by sensor malfunctions or external interference, and are used to identify and remove them to improve data quality. The filtered multi-source data refers to the set of normal data retained after removing outliers; that is, the valid data after removing outliers, which is used to ensure the accuracy of subsequent analysis.

[0127] Understandably, removing outliers from multi-source power grid data based on statistical feature data to obtain filtered multi-source data means comparing the data points from each data source with the normal range determined based on statistical feature data, and removing data points that exceed the normal range as outliers, thereby obtaining filtered multi-source data after removing outliers.

[0128] By removing outliers, obvious erroneous measurements in the data can be eliminated, avoiding interference from outlier data with subsequent reconstruction results.

[0129] In the specific implementation, raw data is collected, and for each data source's time series, it is processed in the order of filling missing data, identifying anomalies, repairing anomalies, and standardization to ensure the data quality of a single data source. If data is missing, due to the complex non-linear correlation between multiple data sources, a Singular Value Thresholding (SVT) algorithm is used to fill in the missing data. First, a 3D model is used... Criteria for filtering outliers: For the nth data source, extract its time series data. The average level of the nth data source:

[0130] In the formula, m represents the number of non-NaN elements (missing terms are excluded if data is missing), and the standard deviation is... This indicates data fluctuation; a larger value indicates greater instability. It can be represented as:

[0131] Normal data range is Check each data point one by one, if or Then mark This is an outlier.

[0132] Step S104: Perform low-rank completion on the filtered multi-source data to obtain the completed multi-source data; It should be noted that during the data completion phase of power grid accident reconstruction, it is necessary to fill the data gaps left after removing outliers.

[0133] It should be noted that low-rank imputation refers to a method of filling in missing data by utilizing the low-rank structure of a data matrix. It is a matrix recovery technique based on the assumption of inherent data correlation, used to fill in missing items while preserving the main characteristics of the data. The imputed multi-source data refers to the complete dataset after low-rank imputation processing, i.e., the data with missing items effectively filled, which provides complete time-series information.

[0134] Understandably, performing low-rank completion on the filtered multi-source data to obtain the completed multi-source data means constructing a data matrix from the filtered multi-source data, and using the inherent low-rank structure of the data to fill in and restore the missing positions, thereby obtaining the multi-source data with the missing items filled in.

[0135] By using low-rank completion, missing items in the data matrix can be filled based on the inherent correlation of the data, ensuring the integrity and availability of multi-source data.

[0136] In one feasible implementation, step S104 may include: Step S1041: Construct an initial data matrix from the filtered multi-source data; It should be noted that in the matrix construction phase of power grid accident reconstruction, the selected multi-source data needs to be organized into a matrix form for low-rank decomposition.

[0137] It should be noted that the initial data matrix refers to a two-dimensional array formed by arranging the filtered multi-source data according to the time dimension and the data source dimension. That is, it is a structured matrix containing time series data from each data source, and it is used as input for low-rank completion. The data matrix refers to a mathematical structure that organizes data in rows and columns, that is, a two-dimensional arrangement containing time series information from multiple data sources, and it is used to support matrix factorization operations.

[0138] Understandably, constructing an initial data matrix from filtered multi-source data means arranging and combining time-series data from each data source according to preset dimensions to form a structured initial data matrix.

[0139] By constructing an initial data matrix, scattered multi-source data are organized into a unified mathematical structure, providing standardized input for subsequent matrix decomposition and missing term processing.

[0140] Step S1042: Replace the positions corresponding to outliers in the initial data matrix with missing markers to obtain a matrix containing missing items; It should be noted that during the missing marker phase of power grid accident reconstruction, the locations where outliers have been removed need to be marked for subsequent processing.

[0141] It should be noted that missing value markers are specific symbols or values ​​used to identify missing data locations in a data matrix. In other words, they indicate that the original data anomalies at that location have been removed, and they guide subsequent completion algorithms in identifying the locations to be filled. A matrix containing missing items refers to a data matrix after replacing outlier locations with missing value markers; that is, it contains a matrix structure that explicitly identifies missing locations, and it is used as input for singular value decomposition.

[0142] Understandably, replacing the positions corresponding to outliers in the initial data matrix with missing markers to obtain a matrix containing missing items means locating the positions where outliers have been removed in the initial data matrix, replacing the data at these positions with missing markers, and thus obtaining a matrix containing missing items that clearly indicates the positions to be filled.

[0143] By replacing the missing markers, the locations of the data to be completed are clearly identified, enabling subsequent decomposition operations to effectively recover these locations.

[0144] Step S1043: Perform singular value decomposition on the matrix containing missing terms to obtain the left singular matrix, the singular value matrix, and the right singular matrix; It should be noted that in the matrix decomposition stage of power grid accident reconstruction, the matrix containing missing terms needs to be decomposed into low-rank components.

[0145] It's important to note that Singular Value Decomposition (SVD) is a mathematical method that decomposes a matrix into the product of a left singular matrix, a singular value matrix, and a right singular matrix. In other words, it's a matrix decomposition technique that achieves dimensionality reduction and structural analysis by extracting the main eigenvalues ​​of the matrix, revealing the inherent low-rank structure of data. The left singular matrix is ​​one of the orthogonal matrices obtained from SVD, containing the set of eigenvectors representing the main patterns of change along the row directions, and is used to characterize the correlation patterns between data sources. The singular value matrix is ​​the diagonal matrix obtained from SVD, containing the diagonal arrangement of the importance values ​​of the principal components, and is used to characterize the energy magnitude of each principal component. The right singular matrix is ​​the other orthogonal matrix obtained from SVD, containing the set of eigenvectors representing the main patterns of change along the column directions, and is used to characterize the main patterns of change in the time dimension.

[0146] Understandably, performing singular value decomposition on a matrix containing missing terms to obtain a left singular matrix, a singular value matrix, and a right singular matrix means decomposing the matrix containing missing terms into the product of three matrices through singular value decomposition, thereby obtaining a left singular matrix, a singular value matrix, and a right singular matrix that respectively represent the row direction features, principal component energy, and column direction features.

[0147] Through singular value decomposition, the intrinsic low-rank structure of the data matrix is ​​revealed as three characteristic matrices, providing a decomposition basis for subsequent thresholding and matrix reconstruction.

[0148] Step S1044: Perform soft thresholding on the singular value matrix to obtain the truncated singular value matrix; It should be noted that during the noise suppression phase of power grid fault reconstruction, secondary feature components need to be filtered to retain the main correlation structure of the data.

[0149] It should be noted that soft thresholding refers to a filtering method that progressively attenuates singular values. This involves compressing singular values ​​below a preset threshold while retaining those above it—a non-linear processing operation used to suppress minor feature components corresponding to noise. The truncated singular value matrix is ​​the diagonal matrix that retains the main singular values ​​after soft thresholding; that is, a simplified diagonal matrix that removes the noise components corresponding to small singular values. It is used to represent the main low-rank structure of the data.

[0150] Understandably, performing soft thresholding on the singular value matrix to obtain a truncated singular value matrix means comparing each singular value in the singular value matrix with a preset threshold, and attenuating the singular values ​​that are smaller than the preset threshold, thereby obtaining a truncated singular value matrix that retains the main feature components and suppresses noise components.

[0151] By using soft thresholding, small singular values ​​representing noise and minor correlations can be suppressed, allowing subsequent reconstruction to focus on the main low-rank structure of the data.

[0152] Step S1045: Reconstruct the complete matrix based on the left singular matrix, the truncated singular value matrix, and the right singular matrix to obtain the completed multi-source data.

[0153] It should be noted that in the matrix recovery phase of power grid accident reconstruction, the processed feature matrix needs to be recombined into a complete data matrix.

[0154] It should be noted that the completed matrix refers to the complete data matrix reconstructed based on the processed left singular matrix, the truncated singular value matrix, and the right singular matrix. In other words, it is a low-rank approximate matrix in which missing positions have been filled and restored, which is used to replace the original matrix containing missing terms.

[0155] Understandably, reconstructing the complete matrix based on the left singular matrix, the truncated singular value matrix, and the right singular matrix to obtain the completed multi-source data means performing matrix multiplication on the left singular matrix, the truncated singular value matrix, and the right singular matrix to recombine them, thereby obtaining the complete matrix where the missing positions have been filled based on the main features, i.e., the completed multi-source data.

[0156] By reconstructing and completing the matrix, the matrix containing missing terms is restored to a low-rank approximately complete matrix, so that the time series from each data source can be reasonably filled in the missing positions.

[0157] In the specific implementation, a data matrix is ​​constructed. ,in, Indicates the first The time point, the first The values ​​from each data source are used; all detected outliers are replaced with NaN to obtain a matrix containing missing items. The low-rank complete matrix is ​​found using the Singular Value Thresholding (SVT) algorithm. ,satisfy ,and The smallest rank is the one that best fits the latent association of multiple features.

[0158] Fill the NaN values ​​in each column with the average amount of non-NaN data to avoid slow convergence caused by initial values ​​of all zeros. Perform singular value decomposition:

[0159] in, r is rank, , , These are the time points and the number of data sources, respectively. It is a left singular matrix. It is a singular value matrix. (here) That is, extraction (the diagonal element), and The larger the singular value, the more important the feature correlation. It is a right singular matrix, and Apply soft thresholding to outliers:

[0160] Only the top k largest singular values ​​are retained; those less than 0 are set to 0. The soft threshold parameter has the following values:

[0161]

[0162] in, The larger the value, the more small singular values ​​are set to zero, and the stronger the low-rank constraint after completion. If the value is too small, the low-rank constraint is too weak, and the completion result is prone to retaining noise. max(M,N) is the maximum dimension of the matrix; taking the square root of its reciprocal is to ensure... The scale is adapted to the matrix dimension. k is a parameter in SVT that limits the number of core singular values ​​retained. Core correlations in low-rank matrices are typically concentrated in the first 50% of singular values, while the latter half correspond to noise / weak correlations. It is an empirical value that can preserve the main correlations of multiple features while ensuring that the rank of the completed matrix is ​​low enough, thus balancing information and constraints.

[0163] Reconstruct the matrix using the singular values ​​after soft thresholding to obtain a new complete matrix. .

[0164]

[0165] in (Singular value diagonal matrix after soft thresholding) It is the first k columns of U. It represents the first k columns of V. Complete the matrix. Not only did it fill in the missing and outlier data, it also replaced the normal original data observations. At this point, the completion matrix needs to be... The value at the known location of the observation is replaced with the original observation.

[0166] Furthermore, after step S1045, the method further includes: calculating the relative error between the current completed matrix and the previous completed matrix to obtain relative error data; when the relative error data is less than a preset convergence threshold, determining the current completed matrix as the completed multi-source data; when the relative error data is not less than the preset convergence threshold, returning to step S1043.

[0167] In the iterative convergence determination and result output stage of power grid fault reconstruction, it is necessary to quantitatively evaluate the stability of the completion iteration process and control the iteration termination. It should be noted that the current completion matrix refers to the matrix reconstructed in this iteration after singular value decomposition and soft thresholding, i.e., the output result of the low-rank completion operation in this round, which is used for comparison with the result of the previous round. The previous completion matrix refers to the matrix reconstructed in the previous iteration, i.e., the output result of the previous low-rank completion operation, which serves as the basis for comparison in this round. Relative error data refers to the quantitative index of the degree of difference between the current completion matrix and the previous completion matrix, i.e., the relative rate of change calculated through the differences in matrix elements, which is used to determine whether the iteration has converged. The preset convergence threshold refers to the pre-set iteration termination judgment standard, i.e., the numerical limit reflecting the stability of the completion result, with a value of 10 to the power of -6, which is used to control the iteration accuracy. The completed multi-source data refers to the final low-rank completion matrix output after iterative convergence, which is the set of data that satisfies the condition that the known observations remain unchanged and the missing items are reasonably filled. It is used as the output of the data completion process.

[0168] Understandably, the relative error between the current completed matrix and the previous completed matrix is ​​calculated to obtain the relative error data. When the relative error data is less than a preset convergence threshold, the current completed matrix is ​​determined as the completed multi-source data. When the relative error data is not less than the preset convergence threshold, the process returns to step S1043, which means comparing the current completed matrix obtained in this round of reconstruction with the previous completed matrix element by element and calculating the relative error data. The relative error data is then compared with the preset convergence threshold. If the relative error data is less than the preset convergence threshold, the iteration is determined to be converged, and the current completed matrix is ​​output as the completed multi-source data that satisfies the condition that the known observations remain unchanged and fits the low-rank correlation. If the relative error data is not less than the preset convergence threshold, the process is determined to be non-converged, and the current completed matrix is ​​returned to step S1043 to continue iterating until the convergence condition is met.

[0169] Through iterative convergence judgment and cyclic optimization, the completion result can gradually approach the optimal low-rank solution, avoiding the completion deviation caused by a single decomposition, while ensuring that the known observations remain unchanged during the iteration process, thus improving the accuracy and stability of data completion.

[0170] In the specific implementation, the current completion matrix is ​​calculated. Compared to the previous round Relative Frobenius norm error:

[0171] Relative error:

[0172] Based on environmental conditions, tol = 10⁻⁶ is chosen. If convergence is achieved, the process is considered complete; otherwise, the steps are repeated.

[0173] The completed matrix after iterative convergence is the final result. This satisfies the condition that the known observed values ​​remain unchanged: Complete the terms to fit low-rank associations: The rank is much smaller than min(M,N).

[0174] Step S105: Standardize the completed multi-source data to obtain preprocessed multi-source data.

[0175] It should be noted that during the data standardization phase of power grid accident reconstruction, it is necessary to eliminate the dimensional differences between different data sources.

[0176] It should be noted that standardization refers to the process of uniformly scaling data from different data sources, that is, the standardization operation that eliminates differences in dimensions and numerical ranges, and is used to achieve comparability of multi-source data. Preprocessed multi-source data refers to the final data set obtained after collection, filtering, completion, and standardization, that is, data with uniform format, reliable quality, and can be directly used for uncertainty calculation, which is used to support subsequent accident reconstruction analysis.

[0177] Understandably, standardizing the completed multi-source data to obtain preprocessed multi-source data means performing a uniform scale transformation on each of the completed data sources to eliminate differences in dimensions and numerical ranges, thereby obtaining preprocessed multi-source data with a uniform format and comparability.

[0178] Through standardization, the differences in dimensions and numerical ranges between different data sources are eliminated, providing a unified and comparable data foundation for subsequent joint analysis of multi-source data and uncertainty calculation.

[0179] This embodiment calculates the average level and volatility of time series data from each data source, quantifying the data distribution characteristics of each source. This allows for the accurate identification and removal of outliers, eliminating erroneous measurements caused by sensor malfunctions or external interference, and preventing abnormal data from interfering with subsequent analysis. Furthermore, by constructing an initial data matrix from the filtered multi-source data, and replacing missing markers at outlier locations with singular value decomposition and soft thresholding, the main low-rank structure of the data matrix is ​​preserved while noise components are suppressed. The resulting completed matrix allows for the reasonable filling of missing positions, ensuring the integrity and usability of the multi-source data. Finally, by standardizing the completed multi-source data, differences in dimensions and numerical ranges between different data sources are eliminated, resulting in a unified and comparable data format. Therefore, this embodiment provides a high-quality, complete, and unified data foundation for subsequent uncertainty calculations and incident reconstruction.

[0180] This embodiment refers to Figure 4 , Figure 4 This is a flowchart illustrating the third embodiment of the power grid accident reconstruction data fusion verification method of this application.

[0181] In one feasible implementation, step S30 includes: Step S301: Calculate the Pearson correlation coefficient based on the uncertainty data of each data source to obtain the static correlation; It should be noted that in the static correlation quantification stage of power grid accident reconstruction, it is necessary to calculate the degree of linear correlation between the uncertainty sequences of each data source.

[0182] It should be noted that the Pearson correlation coefficient (PCC) is a statistical indicator that measures the degree of linear correlation between two variables. It is a numerical measure that reflects the direction and strength of the linear correlation between two sequences by using the ratio of covariance to standard deviation. It is used to characterize the static linear correlation characteristics of uncertainty sequences. Uncertainty data from various data sources refers to the total standard uncertainty set obtained after uncertainty assessment of each data source; that is, structured data containing comprehensive error information from all data sources, which is used as input to support correlation analysis. Static correlation refers to the degree of linear correlation between uncertainties from different data sources regardless of time offset; it is a statistic reflecting the synchronous change characteristics of errors and is used to identify common-source associations of errors from different data sources.

[0183] Understandably, calculating the Pearson correlation coefficient based on the uncertainty data of each data source to obtain the static correlation means extracting the uncertainty sequences of two data sources, calculating the covariance of the two sequences, and then dividing by the product of the standard deviations of the two sequences, thereby obtaining the static correlation that reflects the degree of linear correlation between the uncertainties of the two data sources.

[0184] In the specific implementation, it is necessary to analyze whether the uncertainties of different preprocessed data sources have the same origin (such as two sensors being affected by the same temperature interference, and their uncertainties being positively correlated), so as to lay the foundation for avoiding repeated calculation of the same origin error during subsequent fusion.

[0185] First, build Uncertainty correlation coefficient matrix , This represents the uncertainty correlation coefficient between the i-th data source and the j-th data source, within a certain range. , This indicates that the errors change synchronously and are completely positively correlated; This indicates that the errors are perfectly negatively correlated; if This indicates that the errors are independent and uncorrelated.

[0186] The steps for calculating the Pearson correlation coefficient (static correlation) include: Input the uncertainty sequences of the i-th and j-th data sources , :

[0187] Step S302: Calculate the dynamic time bending distance based on the uncertainty data of each data source to obtain the dynamic correlation; It should be noted that in the dynamic correlation quantification stage of power grid accident reconfiguration, it is necessary to calculate the similarity between the uncertainty sequences of each data source, taking into account the time offset.

[0188] It's important to note that Dynamic Time Warping (DTW) is a metric for measuring the similarity between two time series. It involves using dynamic programming to find the optimal time warping alignment path and calculate the minimum cumulative distance, characterizing the dynamic time lag correlation characteristics of uncertainty sequences. Dynamic correlation refers to the degree of similarity between uncertainties from different data sources, considering time offsets. It's a statistic reflecting the characteristics of error changes with time lag, used to identify the time delay patterns of error propagation from different data sources. A distance matrix is ​​a two-dimensional array composed of point-by-point distance values ​​between the two sequences. It reflects the degree of difference between the two sequences at different time points and serves as the foundational data for dynamic programming pathfinding.

[0189] Understandably, calculating the dynamic time curvature distance based on the uncertainty data of each data source to obtain the dynamic correlation means constructing a distance matrix of the uncertainty sequences of two data sources, using dynamic programming to find the optimal curvature path from the starting point to the ending point in the distance matrix, so that the cumulative distance on the path is minimized, and the minimum cumulative distance is used as the dynamic correlation reflecting the degree of dynamic time lag correlation between the uncertainties of the two data sources.

[0190] In practical implementation, for variable pairs with time lag (such as...) With fuel input, (Concentration changes take several minutes after fuel input to begin changing). The DTW algorithm is used to calculate the optimal alignment path and extract the delay time. And time alignment is performed during fusion. Two uncertainty sequences are input. , Constructing the distance matrix: Calculating Each point and Euclidean distance of each point:

[0191] Finding the optimal path: Using dynamic programming, find the alignment path with the minimum total distance; first, construct a... Distance matrix Each element represents the difference between the two sequences at the corresponding time point:

[0192] This matrix records the local distances between all possible combinations of time points. A cumulative cost matrix is ​​introduced. ,in Indicates starting from the origin Go to current location The minimum cumulative distance.

[0193] The recursive formula is as follows: ; The boundary conditions are: ; This recursive process ensures that each step selects the direction that minimizes the cumulative cost, reflecting the optimal substructure property.

[0194] Complete the cumulative cost matrix After filling, from the endpoint Begin reverse tracing, gradually returning to the starting point. Each time, the selection is made from three preceding states. The direction with the minimum value forms the complete optimal alignment path. .

[0195] The path is in the form of a series of coordinate pairs:

[0196] in It satisfies the monotonically increasing constraint.

[0197] Based on the optimal path The average time delay between the two sequences can be further calculated. :

[0198] in The sampling period (e.g., 1 second, 1 minute). If ,illustrate Lagging behind Conversely, it is ahead of schedule.

[0199] Before subsequent fusion modeling, the uncertainty sequence of the lag side is shifted forward. Each time step achieves timeline alignment.

[0200] Step S303: Generate multi-source uncertainty correlation data based on static correlation and dynamic correlation.

[0201] It should be noted that in the multi-source error correlation integration stage of power grid accident reconstruction, the static and dynamic analysis results need to be merged into a unified correlation description.

[0202] It should be noted that multi-source uncertainty correlation data refers to the set of correlation descriptions obtained by integrating static and dynamic correlations. That is, it is structured data that reflects the homogeneity and time lag relationship between uncertainties of various data sources. It is used to guide the error deduplication process in the subsequent fusion process.

[0203] Understandably, generating multi-source uncertainty correlation data based on static and dynamic correlations means integrating the linear co-originating correlation information reflected by static correlations with the time-lag correlation information reflected by dynamic correlations to obtain multi-source uncertainty correlation data that comprehensively describes the correlation characteristics between uncertainties from various data sources.

[0204] This embodiment quantifies the static linear correlation between uncertainties from different data sources by calculating the Pearson correlation coefficient, thus identifying the common origin of synchronous error changes. Based on this, the dynamic time lag correlation between uncertainties from different data sources is determined by calculating the dynamic time curvature distance, revealing the time delay pattern of error propagation. Furthermore, by comprehensively integrating static and dynamic correlations, the common origin correlation characteristics and time lag characteristics between multi-source uncertainties are uniformly described, thereby avoiding repeated calculations of common origin errors in subsequent fusion processes and providing a correlation basis for adaptive weight allocation. Therefore, this embodiment improves the accuracy of multi-source data fusion.

[0205] Reference Figure 5 , Figure 5 This is a structural block diagram of the first embodiment of the power grid accident reconstruction data fusion verification device of this application.

[0206] like Figure 5 As shown in the embodiment of this application, the power grid fault reconfiguration data fusion verification device includes: Data preprocessing module 10 is used to preprocess multi-source power grid data to obtain preprocessed multi-source data; Uncertainty calculation module 20 is used to calculate the random fluctuation uncertainty and systematic deviation uncertainty of each data source based on the preprocessed multi-source data, and obtain the uncertainty data of each data source; The correlation calculation module 30 is used to calculate the correlation between multiple uncertainties based on the uncertainty data from each data source, and to obtain multi-source uncertainty correlation data. The evidence fusion module 40 is used to perform adaptive evidence reasoning fusion on the preprocessed multi-source data based on the multi-source uncertainty correlation data to obtain the multi-source fusion result. The credibility assessment module 50 is used to assess credibility based on the multi-source fusion results and obtain confidence intervals and credibility scores.

[0207] Based on the first embodiment of the power grid accident reconstruction data fusion verification device described in this application, a second embodiment of the power grid accident reconstruction data fusion verification device of this application is proposed.

[0208] In this embodiment, the data preprocessing module 10 is also used to collect multi-source power grid data; Statistical characteristic data are obtained by calculating the average level and volatility of time series data from multiple data sources. Outliers in the multi-source power grid data are removed based on statistical characteristic data to obtain the filtered multi-source data; The filtered multi-source data is imputed to obtain the imputed multi-source data. The completed multi-source data is standardized to obtain preprocessed multi-source data.

[0209] In this embodiment, the data preprocessing module 10 is also used to construct an initial data matrix from the filtered multi-source data; Replace the positions corresponding to outliers in the initial data matrix with missing value markers to obtain a matrix containing missing items; Singular value decomposition is performed on the matrix containing missing terms to obtain the left singular matrix, the singular value matrix, and the right singular matrix; The singular value matrix is ​​subjected to soft thresholding to obtain the truncated singular value matrix; The completed matrix is ​​reconstructed based on the left singular matrix, the truncated singular value matrix, and the right singular matrix, resulting in the completed multi-source data.

[0210] In this embodiment, the uncertainty calculation module 20 is also used to obtain the first uncertainty by statistically analyzing the local fluctuations of the preprocessed multi-source data through a sliding window. The second uncertainty is obtained by calculating the systematic bias of the preprocessed multi-source data based on the physical engineering model. The total standard uncertainty is synthesized from the first and second uncertainties to obtain the uncertainty data of each data source.

[0211] In this embodiment, the correlation calculation module 30 is also used to calculate the Pearson correlation coefficient based on the uncertainty data of each data source to obtain the static correlation. The dynamic time bending distance is calculated based on the uncertainty data from each data source to obtain the dynamic correlation; Multi-source uncertainty correlation data are generated based on static and dynamic correlations.

[0212] In this embodiment, the evidence fusion module 40 is also used to construct a basic probability allocation function based on the multi-source uncertainty correlation data to obtain the evidence body of each data source; Based on the evidence from each data source, an improved particle swarm optimization algorithm is used to determine the adaptive weights of each data source, resulting in an adaptive weight allocation. The evidence from each data source is weighted and fused based on the adaptive weight allocation results to obtain the multi-source fusion result.

[0213] In this embodiment, the credibility assessment module 50 is also used to determine the basic probability allocation based on the multi-source fusion result and obtain probability allocation data; The trust function value is calculated based on the probability allocation data to obtain the trust level. The likelihood function value is calculated based on the probability allocation data to obtain the likelihood. Determine the confidence interval based on the degree of trust and the likelihood. The credibility score is calculated based on the multi-source fusion results.

[0214] Other embodiments or specific implementations of the power grid accident reconstruction data fusion verification device of this application can be referred to the above-described method embodiments, and will not be repeated here.

[0215] This application provides a power grid accident reconfiguration data fusion verification device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the power grid accident reconfiguration data fusion verification method in the above embodiment 1.

[0216] The following reference Figure 6 This document illustrates a structural schematic diagram of a power grid fault reconstruction data fusion verification device suitable for implementing embodiments of this application. The power grid fault reconstruction data fusion verification device in this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 6 The power grid accident reconstruction data fusion verification device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0217] like Figure 6As shown, the power grid fault reconstruction data fusion verification device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the power grid fault reconstruction data fusion verification device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the power grid fault reconfiguration data fusion verification device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a power grid fault reconfiguration data fusion verification device with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0218] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0219] The power grid accident reconstructing data fusion verification device provided in this application, employing the power grid accident reconstructing data fusion verification method in the above embodiments, can solve the technical problem that the multi-source data fusion results in power grid accident reconstructing are subject to large deviations due to uncertainty propagation and inter-source correlation, and lacks a reliable quantitative assessment method. Compared with the prior art, the beneficial effects of the power grid accident reconstructing data fusion verification device provided in this application are the same as those of the power grid accident reconstructing data fusion verification method provided in the above embodiments, and other technical features in this power grid accident reconstructing data fusion verification device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0220] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0221] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0222] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the power grid accident reconfiguration data fusion verification method in the above embodiments.

[0223] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0224] The aforementioned computer-readable storage medium may be included in the power grid accident reconfiguration data fusion verification device; or it may exist independently and not be assembled into the power grid accident reconfiguration data fusion verification device.

[0225] The aforementioned computer-readable storage medium carries one or more programs that, when executed by the power grid fault reconstruction data fusion verification device, enable the device to write computer program code for performing the operations of this application in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++; and also conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet using an Internet service provider).

[0226] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0227] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0228] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described power grid accident reconstructing data fusion verification method. This addresses the technical problem that the fusion results of multi-source data in power grid accident reconstructing suffer from significant deviations due to uncertainty propagation and inter-source correlation, and the lack of reliable quantitative assessment methods. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the power grid accident reconstructing data fusion verification method provided in the above embodiments, and will not be elaborated upon here.

[0229] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the power grid accident reconfiguration data fusion verification method described above.

[0230] The computer program product provided in this application can solve the technical problem of data fusion verification for power grid accident reconstruction. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the power grid accident reconstruction data fusion verification method provided in the above embodiments, and will not be repeated here.

[0231] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.

Claims

1. A method for verifying data fusion during power grid fault reconstruction, characterized in that, The method includes: Preprocessing of multi-source power grid data yields preprocessed multi-source data. The random fluctuation uncertainty and systematic deviation uncertainty of each data source are calculated based on the preprocessed multi-source data to obtain the uncertainty data of each data source; The correlation between the uncertainties from each data source is calculated to obtain the multi-source uncertainty correlation data. Based on the multi-source uncertainty correlation data, adaptive evidence reasoning fusion is performed on the preprocessed multi-source data to obtain the multi-source fusion result; The credibility is assessed based on the multi-source fusion results to obtain the confidence interval and credibility score.

2. The method as described in claim 1, characterized in that, The step of preprocessing multi-source power grid data to obtain preprocessed multi-source data includes: Collect multi-source power grid data; The average level and volatility of the time series data from each data source are calculated based on the multi-source data to obtain statistical characteristic data. Based on the statistical feature data, outliers in the multi-source power grid data are removed to obtain the filtered multi-source data; The filtered multi-source data is then filled with low-rank data to obtain the filled multi-source data. The completed multi-source data is then standardized to obtain preprocessed multi-source data.

3. The method as described in claim 2, characterized in that, The step of performing low-rank completion on the filtered multi-source data to obtain the completed multi-source data includes: The filtered multi-source data is used to construct an initial data matrix; Replace the positions corresponding to outliers in the initial data matrix with missing markers to obtain a matrix containing missing items; Singular value decomposition is performed on the matrix containing missing terms to obtain the left singular matrix, the singular value matrix, and the right singular matrix; The singular value matrix is ​​subjected to soft thresholding to obtain a truncated singular value matrix; The completed matrix is ​​reconstructed based on the left singular matrix, the truncated singular value matrix, and the right singular matrix to obtain the completed multi-source data.

4. The method as described in claim 1, characterized in that, The step of calculating the random fluctuation uncertainty and systematic bias uncertainty of each data source based on the preprocessed multi-source data to obtain the uncertainty data of each data source includes: The first uncertainty is obtained by statistically analyzing the local fluctuations of the preprocessed multi-source data using a sliding window. The systematic bias of the preprocessed multi-source data is calculated based on the physical engineering model to obtain the second uncertainty. The total standard uncertainty is synthesized from the first uncertainty and the second uncertainty to obtain the uncertainty data of each data source.

5. The method as described in claim 1, characterized in that, The step of calculating the correlation between multiple uncertainties based on the uncertainty data from each data source to obtain multi-source uncertainty correlation data includes: The Pearson correlation coefficient is calculated based on the uncertainty data of each data source to obtain the static correlation. The dynamic time bending distance is calculated based on the uncertainty data of each data source to obtain the dynamic correlation; Multi-source uncertainty correlation data are generated based on the static correlation and the dynamic correlation.

6. The method as described in claim 1, characterized in that, The step of performing adaptive evidence reasoning fusion on the preprocessed multi-source data based on the multi-source uncertainty correlation data to obtain the multi-source fusion result includes: Based on the multi-source uncertainty correlation data, a basic probability allocation function is constructed to obtain the evidence from each data source; Based on the evidence from each data source, an improved particle swarm optimization algorithm is used to determine the adaptive weights of each data source, resulting in an adaptive weight allocation. The evidence from each data source is weighted and fused based on the adaptive weight allocation result to obtain a multi-source fusion result.

7. The method as described in claim 1, characterized in that, The step of performing a credibility assessment based on the multi-source fusion result to obtain a confidence interval and a credibility score includes: Based on the multi-source fusion results, a basic probability allocation is determined, and probability allocation data is obtained. The trust function value is calculated based on the probability allocation data to obtain the trust level; The likelihood function value is calculated based on the probability allocation data to obtain the likelihood. The confidence interval is determined based on the confidence level and the likelihood level; A credibility score is calculated based on the multi-source fusion results to obtain the credibility score.

8. A data fusion verification device for power grid accident reconfiguration, characterized in that, The device includes: The data preprocessing module is used to preprocess multi-source power grid data to obtain preprocessed multi-source data; The uncertainty calculation module is used to calculate the random fluctuation uncertainty and systematic deviation uncertainty of each data source based on the preprocessed multi-source data, and obtain the uncertainty data of each data source. The correlation calculation module is used to calculate the correlation between multiple uncertainties based on the uncertainty data of each data source, and obtain multi-source uncertainty correlation data. The evidence fusion module is used to perform adaptive evidence reasoning fusion on the preprocessed multi-source data based on the multi-source uncertainty correlation data to obtain the multi-source fusion result; The credibility assessment module is used to assess credibility based on the multi-source fusion results and obtain confidence intervals and credibility scores.

9. A data fusion verification device for power grid accident reconfiguration, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the power grid fault reconfiguration data fusion verification method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the power grid accident reconfiguration data fusion verification method as described in any one of claims 1 to 7.

Images