Intelligent power grid management method based on electric energy metering box

Abstract

The invention relates to the technical field of smart power grids, and discloses a smart power grid management method based on electric energy metering boxes, and the method comprises the steps: synchronously collecting the voltage, current waveform and phase data of all electric energy metering boxes in a target region, calculating the power time sequence data, and carrying out the signal decomposition, thereby obtaining a reference load component and a fluctuation load component, then calculating a lagging cross-correlation function of a fluctuation signal between any two electric energy metering boxes to obtain a lagging cross-correlation analysis result, reversely correcting the reference load component based on the result until the model converges to obtain a converged weight matrix, and constructing an operation state topological structure based on the converged weight matrix to obtain an operation state topological structure; inverting a confidence score of the operating state topological structure based on the fluctuation load component, and generating a personalized demand management strategy of the electric energy metering box based on the confidence score; the intelligent power grid management efficiency based on the electric energy metering box can be improved.

Description

A Smart Grid Management Method Based on Electricity Metering Box Technical Field

[0001] This invention relates to the field of smart grid technology, and in particular to a smart grid management method based on an electricity metering box. Background Technology

[0002] In smart grid operation and management, the electricity metering box is the core terminal for load data acquisition. Its data application efficiency directly affects the accuracy of grid control. Existing technologies mostly rely on preset fixed topology models for grid management, failing to fully explore the dynamic correlation information in the load data of the metering box, and lacking effective signal decomposition methods to separate the baseline load and fluctuating load. This results in the topology structure being unable to match the grid operating status in real time, making it difficult to accurately capture the propagation path and correlation characteristics of load fluctuations, thereby affecting the accuracy of disturbance tracing and fault location, and failing to provide dynamic and reliable topology support for grid management.

[0003] Meanwhile, existing power grid demand management strategies mostly adopt a unified control model, failing to consider the differentiated roles of different metering boxes in the topology and the differences in load characteristics. Traditional methods have neither established a confidence assessment mechanism for the topology nor combined fluctuation source identification and propagation path analysis to formulate personalized strategies, resulting in insufficient targeting of control measures. Even when some solutions attempt to optimize management strategies by combining load data, the lack of quantitative judgment on topology reliability makes it difficult to balance control accuracy and system stability, failing to meet the refined and personalized management needs of smart grids and restricting the improvement of power grid operating efficiency and power supply quality. Therefore, how to provide dynamic and reliable topology support for power grid management and improve power grid operating efficiency and power supply quality has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a smart grid management method based on an electricity metering box to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, this invention provides a smart grid management method based on electricity metering boxes, comprising: S1, calculating power time-series data and performing signal decomposition based on voltage, current waveforms, and phase data of all electricity metering boxes within a target area, to obtain a reference load component and a fluctuating load component; S2, calculating the hysteresis cross-correlation function of the fluctuating signal between any two electricity metering boxes based on the reference load component and the fluctuating load component, obtaining the hysteresis cross-correlation analysis result, and back-correcting the reference load component based on the hysteresis cross-correlation analysis result until the model converges, to obtain a converged weight matrix; S3, constructing an operating topology based on the converged weight matrix, and inverting the confidence score of the operating topology based on the fluctuating load component; S4, generating a personalized demand management strategy for the electricity metering boxes based on the confidence score.

[0006] In a preferred embodiment, the step of calculating power time-series data and performing signal decomposition based on the voltage, current waveforms, and phase data of all energy metering boxes within the target area, collected synchronously, to obtain a reference load component and a fluctuating load component includes: calculating the instantaneous power at each moment based on the instantaneous voltage, instantaneous current, and instantaneous phase difference of all energy metering boxes within the target area, generating power time-series data corresponding to each energy metering box; performing wavelet decomposition processing on the power time-series data based on a preset wavelet transform basis function to obtain a low-frequency approximation coefficient sequence and a high-frequency detail coefficient sequence; reconstructing the low-frequency approximation coefficient sequence to obtain the reference load component, and reconstructing the high-frequency detail coefficient sequence to obtain the fluctuating load component.

[0007] In a preferred embodiment, the step of performing wavelet decomposition on the power time series data based on a preset wavelet transform basis function to obtain a low-frequency approximation coefficient sequence and a high-frequency detail coefficient sequence includes: performing multi-scale decomposition on the power time series data based on the preset wavelet transform basis function to obtain approximation coefficients and detail coefficients at each decomposition scale, wherein the mathematical expression of the preset wavelet transform basis function is as follows: In the formula, For scale j k The approximation coefficients below, For scale j k The detail coefficient, j k For decomposition scale index, k is the coefficient position index. and These are the basis functions obtained by the wavelet scaling function and the wavelet function through translation and scaling, respectively. P[n] is the instantaneous power value at the nth sampling time, N is the data length, and n is the sampling time index. The approximate coefficient sequence at the lowest decomposition scale is taken as the low-frequency approximate coefficient sequence, and the detail coefficient sequences at all scales are merged as the high-frequency detail coefficient sequence.

[0008] In a preferred embodiment, the step of calculating the hysteresis cross-correlation function of the fluctuating signals between any two energy metering boxes based on the reference load component and the fluctuating load component to obtain the hysteresis cross-correlation analysis results includes: for any two energy metering boxes, obtaining the time series of the fluctuating load components corresponding to these two energy metering boxes respectively; and calculating the hysteresis cross-correlation function values ​​of the time series of the fluctuating load components corresponding to the two energy metering boxes at different hysteresis times based on a preset hysteresis time range, wherein the mathematical expression for calculating the hysteresis cross-correlation function value is as follows: In the formula, R ij (τ) is the value of the lagged cross-correlation function, N f F is the total length of the time series of the fluctuating load components, τ is the lag time, and F is the total length of the time series of the fluctuating load components. i (t) and F j (t) represents the time series of two fluctuating load components, μ i and μ j These are sequences F i (t) and F j The sample mean of (t), σ i and σ j These are sequences F i (t) and F j The sample standard deviation of (t), where t is the time series index, and i and j are the identification indices of any two electricity metering boxes; the maximum correlation value and its corresponding optimal lag time are extracted from the lag cross-correlation function value to form the lag cross-correlation analysis results for the metering box.

[0009] In a preferred embodiment, the step of reversely correcting the reference load component based on the hysteresis cross-correlation analysis results includes: initializing a fully connected virtual topology network with all electricity metering boxes in the target area as nodes, wherein the edge weights between any two nodes are initialized to preset values; calculating the updated virtual topology network edge weights using a topology weight update function based on the maximum correlation value and optimal hysteresis time in the hysteresis cross-correlation analysis results; and performing reverse correction on the reference load component based on the updated virtual topology network edge weights to obtain the corrected reference load component.

[0010] In a preferred embodiment, reverse correction processing is continuously performed until the difference between the weight matrices formed by the edge weights of the virtual topology network generated in two adjacent iterations is less than a preset convergence threshold. The weight matrix that finally meets the condition is then used as the converged weight matrix.

[0011] In a preferred embodiment, the step of constructing the operational topology based on the converged weight matrix and inverting the confidence score of the operational topology based on the fluctuating load components includes: generating the operational topology by performing connection pruning using a preset weight threshold and the minimum spanning tree method in graph theory based on the converged weight matrix; calculating the stability index of the converged weight matrix based on the historical iterative weight sequence; calculating the explanatory power index of the topology for the fluctuating data based on the operational topology and the fluctuating load components; and calculating the confidence score of the operational topology by weighted fusion based on the stability index and the explanatory power index of the topology for the fluctuating data.

[0012] In a preferred embodiment, the graph theory minimum spanning tree method is as follows: starting from any node, continuously select the edge with the smallest weight between the currently connected set of nodes and the unconnected set of nodes.

[0013] In a preferred embodiment, generating a personalized demand management strategy for the electricity metering box based on the confidence score includes: comparing the confidence score with a preset confidence threshold; if the confidence score is greater than or equal to the threshold, executing subsequent strategy generation steps; otherwise, outputting alarm information; aggregating the baseline load components based on the operating topology to determine the load baseline value of each metering box in the topology; identifying key fluctuation sources and key propagation paths based on the operating topology and the fluctuating load components; and generating a corresponding personalized demand management strategy for each electricity metering box by combining the load baseline value, key fluctuation sources, key propagation paths, and the confidence score.

[0014] In a preferred embodiment, identifying key fluctuation sources and key propagation paths based on the operating topology and the fluctuating load components includes: calculating the corresponding fluctuation intensity index based on the time series of the fluctuating load components of each energy metering box; comparing the fluctuation intensity index with a preset fluctuation intensity threshold, and identifying energy metering boxes with fluctuation intensity indices greater than the threshold as key fluctuation sources; for each key fluctuation source, based on the operating topology, finding and determining a downstream path that starts from the key fluctuation source node, propagates along the topological connection direction, and whose path length does not exceed a preset depth value, as the key propagation path.

[0015] Compared with existing technologies, the present invention has the following beneficial effects: 1. The present invention significantly improves the accuracy and dynamic adaptability of the power grid operating state topology structure by combining signal decomposition and hysteresis cross-correlation analysis. By using wavelet transform to decompose the power time series data into a reference load component and a fluctuating load component, the stable load and instantaneous disturbance information are effectively separated, providing an accurate data foundation for topology analysis. By calculating the hysteresis cross-correlation function of the fluctuating signals between any two metering boxes, the correlation strength and propagation delay of load fluctuations are quantified, and the reference load component is corrected in reverse based on the result. The weight matrix obtained by closed-loop iterative convergence can truly reflect the actual electrical connection relationship between metering boxes. This data-driven topology inversion method breaks through the limitations of traditional fixed topology models, can capture the dynamic changes of the power grid operating state in real time, accurately identify the disturbance propagation path, and provide reliable topology support for power grid fault tracing and load regulation, greatly improving the scientific nature and foresight of power grid management.

[0016] 2. This invention generates personalized demand management strategies based on topology confidence scoring, achieving refined and differentiated power grid control. Through weighted fusion of stability indicators and fluctuation data interpretability indicators, it quantitatively assesses the confidence level of the inverted topology, ensuring the reliability of subsequent strategy formulation. Under the premise of meeting confidence standards, it combines topology to aggregate load baseline values, identify key fluctuation sources and propagation paths, and tailors management strategies for different metering boxes. It provides time-of-use pricing suggestions for nodes with high load baseline values, issues rapid response commands to key fluctuation sources, and strengthens monitoring and resource deployment for nodes on key propagation paths. Simultaneously, through confidence-weighted strategy execution, it achieves precise matching between control measures and actual operational needs. This model avoids the blindness of traditional unified control, improving power grid operating efficiency and power supply stability while reducing unnecessary control costs, fully meeting the personalized and refined management needs of smart grids. Attached Figure Description

[0017] Figure 1 is a flowchart illustrating a smart grid management method based on an energy metering box according to an embodiment of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] This application provides a smart grid management method based on an electricity metering box. The executing entity of this smart grid management method based on an electricity metering box includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the smart grid management method based on an electricity metering box can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0021] Referring to Figure 1, a flowchart illustrating a smart grid management method based on an energy metering box according to an embodiment of the present invention is shown. In this embodiment, the smart grid management method based on an energy metering box includes: S1, calculating power time-series data and performing signal decomposition based on the voltage, current waveforms, and phase data of all energy metering boxes within a synchronously acquired target area to obtain a reference load component and a fluctuating load component; In this embodiment, calculating power time-series data and performing signal decomposition based on the voltage, current waveforms, and phase data of all energy metering boxes within a synchronously acquired target area to obtain a reference load component and a fluctuating load component includes: calculating the instantaneous power at each moment based on the instantaneous voltage value, instantaneous current value, and instantaneous phase difference of all energy metering boxes within the synchronously acquired target area, generating power time-series data corresponding to each energy metering box; performing wavelet decomposition processing on the power time-series data based on a preset wavelet transform basis function to obtain a low-frequency approximation coefficient sequence and a high-frequency detail coefficient sequence; reconstructing the low-frequency approximation coefficient sequence to obtain the reference load component, and reconstructing the high-frequency detail coefficient sequence to obtain the fluctuating load component.

[0022] It should be noted that the instantaneous power is calculated by multiplying the instantaneous voltage and current values ​​acquired synchronously point by point, and then multiplying the product by the cosine of the instantaneous phase difference at the corresponding moment to obtain the instantaneous power value at that moment. This process is executed continuously to generate power time-series data that varies with time.

[0023] It should be noted that the specific method for reconstructing the reference load component and the fluctuating load component is to perform inverse wavelet transform on the low-frequency approximation coefficient sequence and the high-frequency detail coefficient sequence, respectively. The mathematical expression of the inverse wavelet transform is as follows: In the formula, B(t) is the baseline load component, t is a continuous time variable, and K...J Let be the length of the high-frequency detail coefficient sequence at scale j, and k be the coefficient position index, CA J [k] represents the k-th low-frequency approximation coefficient at the maximum scale J. To reconstruct the scaling function, F(t) represents the fluctuating load component, J represents the maximum decomposition scale, and j... k For decomposition scale index, For scale j k The k-th high-frequency detail coefficient, To reconstruct the wavelet function.

[0024] It should be noted that the reference load component is a time series signal obtained by reconstructing the low-frequency approximation coefficient sequence through wavelet inverse transform. Its frequency components are low and change slowly, reflecting the long-term average power level and basic composition of the load connected to the power metering box, and is the main part of the load curve.

[0025] It should be noted that the fluctuating load component is a time series signal obtained by reconstructing the high-frequency detail coefficient sequence through inverse wavelet transform. Its frequency components are high and change rapidly. Superimposed on the reference load, it reflects the instantaneous fluctuations, random interference, and potential faults or abnormal disturbances of the load. It is the main data source for subsequent topological relationship analysis and disturbance tracing.

[0026] In this embodiment of the invention, the step of performing wavelet decomposition on the power time series data based on a preset wavelet transform basis function to obtain a low-frequency approximation coefficient sequence and a high-frequency detail coefficient sequence includes: performing multi-scale decomposition on the power time series data based on a preset wavelet transform basis function to obtain approximation coefficients and detail coefficients at each decomposition scale, wherein the mathematical expression of the preset wavelet transform basis function is as follows: In the formula, For scale j k The approximation coefficients below, For scale j k The detail coefficient, j k For decomposition scale index, k is the coefficient position index. and These are the basis functions obtained by the wavelet scaling function and the wavelet function through translation and scaling, respectively. P[n] is the instantaneous power value at the nth sampling time, N is the data length, and n is the sampling time index. The approximate coefficient sequence at the lowest decomposition scale is taken as the low-frequency approximate coefficient sequence, and the detail coefficient sequences at all scales are merged as the high-frequency detail coefficient sequence.

[0027] It should be noted that multi-scale decomposition of power time series data utilizes the time-frequency localization property of wavelet transform to decompose the power time series data into different scales. Each scale corresponds to a frequency band, and the larger the scale, the lower the corresponding frequency. At each scale, wavelet transform is used to obtain approximate coefficients that reflect the low-frequency profile of the signal and detail coefficients that reflect the high-frequency details of the signal.

[0028] It should be noted that the low-frequency approximation coefficient sequence is a set of dimensionless coefficients with the lowest corresponding frequency range. It reflects the slowly changing, long-term trend components in the power time series data and is the basis for reconstructing the reference load components.

[0029] It should be noted that the high-frequency detail coefficient sequence is a collection of multiple dimensionless coefficients that cover detailed information from lower to higher frequencies. It reflects the rapidly changing and short-term fluctuating components in the power time series data and is the basis for reconstructing the fluctuating load components.

[0030] S2, based on the reference load component and the fluctuating load component, calculate the hysteresis cross-correlation function of the fluctuating signals between any two energy metering boxes to obtain the hysteresis cross-correlation analysis result. Based on the hysteresis cross-correlation analysis result, reverse-correct the reference load component until the model converges to obtain the converged weight matrix. In this embodiment of the invention, calculating the hysteresis cross-correlation function of the fluctuating signals between any two energy metering boxes based on the reference load component and the fluctuating load component to obtain the hysteresis cross-correlation analysis result includes: for any two energy metering boxes, obtaining the time series of the fluctuating load components corresponding to these two energy metering boxes respectively; based on a preset hysteresis time range, calculating the hysteresis cross-correlation function values ​​of the time series of the fluctuating load components corresponding to the two energy metering boxes at different hysteresis times, wherein the mathematical expression for calculating the hysteresis cross-correlation function value is as follows: In the formula, R ij (τ) is the value of the lagged cross-correlation function, N f F is the total length of the time series of the fluctuating load components, τ is the lag time, and F is the total length of the time series of the fluctuating load components. i (t) and F j (t) represents the time series of two fluctuating load components, μ i and μ j These are sequences F i (t) and F j The sample mean of (t), σ i and σ j These are sequences F i (t) and F jThe sample standard deviation of (t), where t is the time series index, and i and j are the identification indices of any two electricity metering boxes; the maximum correlation value and its corresponding optimal lag time are extracted from the lag cross-correlation function value to form the lag cross-correlation analysis results for the metering box.

[0031] It should be noted that obtaining the time series of fluctuating load components involves extracting a series of data points arranged in chronological order within the corresponding time window from the fluctuating load component results of each metering box obtained in the signal decomposition step, forming a time series signal characterizing the high-frequency fluctuation characteristics of the load of that metering box.

[0032] It should be noted that τ∈[-τ] max , τ max And τ is an integer, representing the sequence F j (t) relative to F i (t) The number of steps shifted on the time axis, in the sampling interval, where τ is a positive value representing F. j Lagging behind F i τ being negative indicates F j Ahead of F i .

[0033] It should be noted that extracting the maximum correlation value means finding the lag cross-correlation function value with the largest absolute value among all the calculated lag time values. It represents the maximum possible correlation strength between the fluctuation signals of the two metering boxes.

[0034] It should be noted that the optimal lag time is the lag time corresponding to the maximum correlation value. It indicates the most likely time delay in the transmission of fluctuations from metering box i to metering box j, and the unit is the sampling interval time.

[0035] It should be noted that the result of the lag cross-correlation analysis is a data pair containing the maximum correlation value and the optimal lag time, which quantifies the strength of the synchronicity or causal relationship between the load fluctuation patterns of the two metering boxes. The closer the value is to 1 or -1, the stronger the correlation.

[0036] In this embodiment of the invention, the step of reversely correcting the reference load component based on the hysteresis cross-correlation analysis results includes: initializing a fully connected virtual topology network with all electricity metering boxes in the target area as nodes, wherein the edge weights between any two nodes are initialized to preset values; calculating the updated virtual topology network edge weights using a topology weight update function based on the maximum correlation value and optimal hysteresis time in the hysteresis cross-correlation analysis results; and performing reverse correction on the reference load component based on the updated virtual topology network edge weights to obtain the corrected reference load component.

[0037] It should be noted that the initialization of the fully connected virtual topology network is to construct a graph structure in which each electricity meter box represents a node, and there is an edge between any two nodes. The weight of the edge is initialized to the same preset value, indicating that the connection probability between all meter boxes is equal at the beginning.

[0038] Furthermore, the weights of the edges are initialized to the same preset value, which defaults to 0.5.

[0039] It should be noted that the updated edge weights are calculated using the topological weight update function based on the results of lag cross-correlation analysis. The maximum correlation value and the optimal lag time are used as inputs to calculate the edge weights between nodes, reflecting the strength of the disturbance propagation and the connection probability implied by the delay.

[0040] Furthermore, the mathematical expression for the topological weight update function is as follows: Among them, W ij It is the updated edge weight from node i to node j. It is the maximum correlation value. α is the optimal lag time, β is the normalization coefficient, and β is the time delay decay coefficient.

[0041] Furthermore, the updated edge weights range from [0,1], representing the connection strength or influence probability from metering box i to metering box j; the larger the absolute value of the maximum correlation value, the stronger the correlation of the fluctuation signal, and the larger the edge weight should be; the larger the absolute value of the optimal lag time, the longer the propagation delay, and the smaller the edge weight should be. It is a negative exponential decay function used to simulate the impact of delay on connection strength; the normalization coefficient is used to ensure that the weight is within a reasonable range and takes a value of 1; the time delay decay coefficient is used to control the decay rate and takes a value of 0.1.

[0042] It should be noted that the reverse correction of the baseline load component utilizes the edge weights of the updated virtual topology network and a load propagation simulation algorithm to simulate the propagation process of load or disturbance in the topology network, thereby adjusting the baseline load component of each metering box to better reflect the load distribution under the actual topology. The mathematical expression of the load propagation simulation algorithm is as follows: ;in, B is the corrected reference load component of metering box i. i γ is the original reference load component of metering box i, and W is the correction factor. ji It is the updated edge weight from node j to node i. This represents the summation over all other metering boxes, where i and j are the identifier indices of any two electricity metering boxes.

[0043] It should be noted that the corrected reference load component is the result of adjusting the original reference load component after considering the virtual topology connection relationship. By introducing the topology propagation effect, the reference load not only reflects the local stable load, but also includes the load influence transmitted by adjacent metering boxes through electrical connections, thus more accurately representing the power grid operating status and providing a more accurate input for the subsequent convergence of the weight matrix.

[0044] In this embodiment of the invention, reverse correction processing is continuously performed until the difference between the weight matrices formed by the edge weights of the virtual topology network generated in two adjacent iterations is less than a preset convergence threshold. The weight matrix that finally meets the condition is then used as the converged weight matrix.

[0045] It should be noted that the difference between the old and new weight matrices is calculated by calculating the difference in norms between the weight matrices obtained from the two iterations. When this difference is less than a preset convergence threshold, the final weight matrix that meets the condition is taken as the converged weight matrix. The preset convergence threshold is a very small positive number, with a value of 10. -3 .

[0046] It should be noted that each iteration constitutes a closed-loop process. First, the baseline load component is corrected based on the current weight matrix. Then, the fluctuating load component is updated using the corrected baseline load component. Next, the hysteresis cross-correlation analysis results are recalculated based on the updated fluctuating load component. Finally, a new generation of weight matrix is ​​generated using the topological weight update function based on the new hysteresis cross-correlation analysis results.

[0047] S3, constructing a running-state topology based on the converged weight matrix, and inverting a confidence score for the running-state topology based on the fluctuating load components; In this embodiment of the invention, constructing a running-state topology based on the converged weight matrix and inverting a confidence score for the running-state topology based on the fluctuating load components includes: generating a running-state topology by performing connection pruning using a preset weight threshold and the minimum spanning tree method in graph theory based on the converged weight matrix; calculating a stability index of the converged weight matrix based on the historical iterative weight sequence; calculating an explanatory power index of the topology for fluctuating data based on the running-state topology and the fluctuating load components; and calculating a confidence score for the running-state topology by weighted fusion based on the stability index and the explanatory power index of the topology for fluctuating data.

[0048] It should be noted that the connection pruning is performed by using a preset weight threshold and the minimum spanning tree method in graph theory. First, a weight threshold is set, and all elements in the converged weight matrix that are less than the preset weight threshold are set to zero, resulting in a sparse weight matrix. Then, each electricity metering box is regarded as a node in the graph, and the connections corresponding to the non-zero elements in the sparse weight matrix are regarded as weighted edges, thus constructing a weighted connected graph. Finally, the minimum spanning tree method in graph theory is applied to this weighted connected graph to find the tree structure that connects all nodes and has the minimum sum of edge weights. This tree structure is the most likely operating topology structure inferred.

[0049] It should be noted that the operational topology is a tree or directed acyclic graph structure, where the nodes correspond to the electricity metering boxes and the edges represent the actual inferred electrical connections. This structure reflects the actual physical connection hierarchy and path of the electricity metering boxes in the target area during operation, and serves as the basic network model for subsequent personalized demand management.

[0050] It should be noted that the stability index of the weight matrix is ​​used to evaluate the reliability and robustness of the final result of the iterative optimization process. By analyzing the weight matrix sequence generated in the last few iterations, the degree of variation between them is calculated. The higher the stability, the more reliable the model convergence result.

[0051] Furthermore, the mathematical expression for the stability index is as follows: In the formula, SI is the stability index, and W (K-tr+1) W is the weight matrix for the (K-tr+1)th iteration. (K-tr) This is the weight matrix for the (K-tr)th iteration, where tr is the iteration index, m is the final iteration number, and ||W (k-tr+1) -W (k-tr) || F It is the norm difference between two consecutive iterations of the weight matrix. Let K denote the norm of the matrix, and K be the total number of iterations.

[0052] It should be noted that the explanatory power index of the computational topology for fluctuation data is to evaluate the extent to which the generated operating topology can explain the observed fluctuation load components. The principle is that if the topology is correct, then the propagation pattern contained in the fluctuation data should be able to match the propagation model under the constraints of the topology well.

[0053] Furthermore, the mathematical expression for calculating the explanatory power index is as follows: In the formula, EI is the explanatory power index, and F... obs H is the actual observed fluctuating load component matrix, and H is the system matrix constructed based on the topology A. T H) -1 HT It is the projection matrix projected onto the column space of H. The norm of a matrix is ​​represented.

[0054] It should be noted that the confidence score calculated by weighted fusion combines the evaluation information of the stability index and the interpretability index, which are two different dimensions, into a single score to comprehensively reflect the credibility of the inverted operating topology. The stability index and the interpretability index are multiplied by their respective stability weights and interpretability weights, and the product results are added together to obtain the confidence score. The higher the score, the more credible the inverted operating topology is.

[0055] It should be noted that the confidence score is a comprehensive quantitative evaluation result. It evaluates the reliability of the power grid operating state topology based on data-driven inversion from two core dimensions: the stability of the model convergence process and the ability of the inference results to explain the actual observation data. A high score indicates that the topology is not only generated by a stable iterative process, but can also effectively explain the actual disturbance propagation phenomenon in the power grid.

[0056] In this embodiment of the invention, the specific content of the graph theory minimum spanning tree method is as follows: starting from any node, continuously select the edge with the smallest weight between the currently connected set of nodes and the unconnected set of nodes.

[0057] It should be noted that the minimum spanning tree method in graph theory is as follows: Starting from any node, continuously select the edge with the smallest weight between the currently connected set of nodes and the unconnected set of nodes, and add the corresponding new node to the connected set until all nodes are connected. The resulting tree structure ensures that all metering boxes are connected and the total connection cost is the lowest, that is, the overall connection probability is the highest.

[0058] S4. Generate a personalized demand management strategy for the electricity metering box based on the confidence score.

[0059] In this embodiment of the invention, generating a personalized demand management strategy for the electricity metering box based on the confidence score includes: comparing the confidence score with a preset confidence threshold; if the confidence score is greater than or equal to the threshold, executing subsequent strategy generation steps; otherwise, outputting alarm information; aggregating the baseline load components based on the operating topology to determine the load base value of each metering box in the topology; identifying key fluctuation sources and key propagation paths based on the operating topology and the fluctuating load components; and generating a corresponding personalized demand management strategy for each electricity metering box by combining the load base value, key fluctuation sources, key propagation paths, and the confidence score.

[0060] It should be noted that comparing the confidence score with the preset confidence threshold is to determine whether the retrieved running topology has sufficient reliability to support the formulation of subsequent strategies. The preset confidence threshold is 0.75.

[0061] It should be noted that the aggregation of the baseline load components to determine the load base value is based on the hierarchical relationship defined by the operating topology. The baseline load components of all lower-level metering boxes belonging to the same upper-level node are summed from bottom to top. The load base value of the upper-level node is composed of its own baseline load and the baseline load of all its direct lower-level nodes aggregated together. This process quantifies the total stable load level within the jurisdiction of each metering box.

[0062] It should be noted that generating a personalized demand management strategy for each electricity metering box involves synthesizing differentiated strategies based on its load baseline, whether it is a critical source of fluctuations, whether it is located on a critical propagation path, and topology confidence level. The strategy content may include, but is not limited to: time-of-use pricing recommendations or load control targets based on the load baseline; rapid response or filtering device switching commands for critical fluctuation sources; enhanced monitoring of nodes on critical propagation paths; or deployment of flexible control resources. The confidence score is used to weight the recommended execution strength of the strategy; the higher the confidence level, the higher the recommendation level and control intensity of the strategy.

[0063] In this embodiment of the invention, identifying key fluctuation sources and key propagation paths based on the operating topology and the fluctuating load components includes: calculating the corresponding fluctuation intensity index based on the time series of the fluctuating load components of each power metering box; comparing the fluctuation intensity index with a preset fluctuation intensity threshold, and identifying power metering boxes with fluctuation intensity indices greater than the threshold as key fluctuation sources; for each key fluctuation source, based on the operating topology, finding and determining a downstream path that starts from the key fluctuation source node, propagates along the topological connection direction, and whose path length does not exceed a preset depth value, as the key propagation path.

[0064] It should be noted that the fluctuation intensity index is calculated by extracting features from the time series of the fluctuating load component of each metering box to obtain a scalar value to quantify the magnitude of the overall fluctuation amplitude of the series. This index reflects the severity of random disturbances or abnormal changes in the load monitored by the metering box.

[0065] Furthermore, the fluctuation intensity index is calculated by taking the root mean square value of its fluctuation load component time series, and its mathematical expression is: In the formula, I i F is the fluctuation intensity index of metering box i. i(t) is the value of its fluctuating load component at time t, N is the total length of the time series, i is the identifier index of any energy meter box, and N F This is the time series of the fluctuating load component of the metering box, where t is the time index.

[0066] It should be noted that by comparing the fluctuation intensity index with a preset fluctuation intensity threshold, the electricity metering boxes whose fluctuation intensity index exceeds the preset fluctuation intensity threshold are screened out. This method uses a preset threshold value to screen out electricity metering boxes with abnormal fluctuation levels. The preset fluctuation intensity threshold is set as follows: calculate the average standard deviation of the fluctuation intensity index of all electricity metering boxes in the target area, and then set the threshold as the mean plus the standard deviation multiplied by an adjustable coefficient. The adjustable coefficient is set to 1.5, which means that when the fluctuation intensity index of a metering box exceeds this statistical threshold, it is considered a potential source of abnormal fluctuation.

[0067] It should be noted that the key fluctuation source is the metering box node where the amplitude of the fluctuating load component is abnormally prominent and exceeds the normal range. It identifies the "source" location of the main random disturbances or impact loads in the power grid and is the primary target for disturbance management and suppression.

[0068] It should be noted that finding critical propagation paths based on the running topology involves using the electrical connections and directions defined by the topology to start from the identified critical fluctuation source node, traverse along the edges, and find the downstream metering box sequence that the fluctuation may directly affect.

[0069] Furthermore, determining that "the path length does not exceed the preset depth value" means that only the number of topological edges traversed from the critical fluctuation source is considered to be less than or equal to a preset maximum value of the propagation path. This preset maximum value of the propagation path reflects the characteristic of fluctuation energy attenuation with electrical distance in the power grid, focusing on local networks that are significantly affected by the source. The preset maximum value of the propagation path is 3.

[0070] It should be noted that the critical propagation path refers to the sequence of consecutive nodes and edges that connect the critical source of the disturbance to its directly downstream affected areas and have a short electrical distance. It reveals the main electrical channels on which the disturbance spreads from the source to other parts of the power grid, and clarifies the line sections that need to be monitored, protected, or have suppression measures deployed. This provides a direct path basis for implementing precise and efficient power grid disturbance management.

[0071] In the several embodiments provided by this invention, it should be understood that the disclosed method can be implemented in other ways.

[0072] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0073] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, and technology that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smart grid management method based on an electricity metering box, characterized in that, The method includes: S1, calculating power time-series data and performing signal decomposition based on the voltage, current waveforms, and phase data of all energy metering boxes within the target area collected synchronously, to obtain a reference load component and a fluctuating load component; S2, calculating the hysteresis cross-correlation function of the fluctuating signal between any two energy metering boxes based on the reference load component and the fluctuating load component, obtaining the hysteresis cross-correlation analysis result, and back-correcting the reference load component based on the hysteresis cross-correlation analysis result until the model converges, to obtain a converged weight matrix; S3, constructing an operating topology based on the converged weight matrix, and inverting the confidence score of the operating topology based on the fluctuating load component; S4, generating a personalized demand management strategy for the energy metering boxes based on the confidence score.

2. The smart grid management method based on an electricity metering box as described in claim 1, characterized in that, The process involves calculating power time-series data and performing signal decomposition based on the voltage, current waveforms, and phase data of all energy metering boxes within the target area, obtained synchronously. This includes: calculating the instantaneous power at each moment based on the instantaneous voltage, instantaneous current, and instantaneous phase difference of all energy metering boxes within the target area, generating power time-series data corresponding to each energy metering box; performing wavelet decomposition on the power time-series data based on a preset wavelet transform basis function to obtain a low-frequency approximation coefficient sequence and a high-frequency detail coefficient sequence; reconstructing the low-frequency approximation coefficient sequence to obtain the reference load component, and reconstructing the high-frequency detail coefficient sequence to obtain the fluctuating load component.

3. The smart grid management method based on an energy metering box as described in claim 2, characterized in that, The process of performing wavelet decomposition on the power time series data based on a preset wavelet transform basis function to obtain a low-frequency approximation coefficient sequence and a high-frequency detail coefficient sequence includes: performing multi-scale decomposition on the power time series data based on the preset wavelet transform basis function to obtain approximation coefficients and detail coefficients at each decomposition scale, wherein the mathematical expression of the preset wavelet transform basis function is as follows: In the formula, For scale j k The approximation coefficients below, For scale j k The detail coefficient, j k For decomposition scale index, k is the coefficient position index. and These are the basis functions obtained by the wavelet scaling function and the wavelet function through translation and scaling, respectively. P[n] is the instantaneous power value at the nth sampling time, N is the data length, and n is the sampling time index. The approximate coefficient sequence at the lowest decomposition scale is taken as the low-frequency approximate coefficient sequence, and the detail coefficient sequences at all scales are merged as the high-frequency detail coefficient sequence.

4. The smart grid management method based on an electricity metering box as described in claim 1, characterized in that, The step of calculating the hysteresis cross-correlation function of the fluctuating signals between any two energy metering boxes based on the reference load component and the fluctuating load component, and obtaining the hysteresis cross-correlation analysis results, includes: for any two energy metering boxes, obtaining the time series of the fluctuating load components corresponding to these two energy metering boxes respectively; and calculating the hysteresis cross-correlation function values ​​of the time series of the fluctuating load components corresponding to the two energy metering boxes at different hysteresis times based on a preset hysteresis time range, wherein the mathematical expression for calculating the hysteresis cross-correlation function value is as follows: In the formula, R ij (τ) is the value of the lagged cross-correlation function, N f F is the total length of the time series of the fluctuating load components, τ is the lag time, and F is the total length of the time series of the fluctuating load components. i (t) and F j (t) represents the time series of two fluctuating load components, μ i and μ j These are sequences F i (t) and F j The sample mean of (t), σ i and σ j These are sequences F i (t) and F j The sample standard deviation of (t), where t is the time series index, and i and j are the identification indices of any two electricity metering boxes; the maximum correlation value and its corresponding optimal lag time are extracted from the lag cross-correlation function value to form the lag cross-correlation analysis results for the metering box.

5. The smart grid management method based on an electricity metering box as described in claim 1, characterized in that, The step of reversing the baseline load component based on the hysteresis cross-correlation analysis results includes: initializing a fully connected virtual topology network with all electricity metering boxes in the target area as nodes, wherein the edge weights between any two nodes are initialized to preset values; calculating the updated virtual topology network edge weights using a topology weight update function based on the maximum correlation value and optimal hysteresis time in the hysteresis cross-correlation analysis results; and reversing the baseline load component based on the updated virtual topology network edge weights to obtain the corrected baseline load component.

6. The smart grid management method based on an electricity metering box as described in claim 1, characterized in that, The process of obtaining the converged weight matrix until the model converges includes: continuously performing reverse correction processing until the difference between the weight matrices formed by the edge weights of the virtual topology network generated in two adjacent iterations is less than a preset convergence threshold, and finally using the weight matrix that meets the condition as the converged weight matrix.

7. The smart grid management method based on an electricity metering box as described in claim 1, characterized in that, The process of constructing the operational topology based on the converged weight matrix and inverting the confidence score of the operational topology based on the fluctuating load components includes: generating the operational topology by performing connection pruning using a preset weight threshold and the minimum spanning tree method in graph theory based on the converged weight matrix; calculating the stability index of the converged weight matrix based on the historical iterative weight sequence; calculating the explanatory power index of the topology for the fluctuating data based on the operational topology and the fluctuating load components; and calculating the confidence score of the operational topology by weighted fusion based on the stability index and the explanatory power index of the topology for the fluctuating data.

8. The smart grid management method based on an electricity metering box as described in claim 7, characterized in that, The specific content of the graph theory minimum spanning tree method is as follows: Starting from any node, continuously select the edge with the smallest weight between the currently connected set of nodes and the unconnected set of nodes.

9. A smart grid management method based on an electricity metering box as described in claim 1, characterized in that, The process of generating a personalized demand management strategy for the electricity metering box based on the confidence score includes: comparing the confidence score with a preset confidence threshold; if the confidence score is greater than or equal to the threshold, executing subsequent strategy generation steps; otherwise, outputting alarm information; aggregating the baseline load components based on the operating topology to determine the load baseline value of each metering box in the topology; identifying key fluctuation sources and key propagation paths based on the operating topology and the fluctuating load components; and generating a corresponding personalized demand management strategy for each electricity metering box by combining the load baseline value, key fluctuation sources, key propagation paths, and the confidence score.

10. A smart grid management method based on an energy metering box as described in claim 9, characterized in that, The step of identifying key fluctuation sources and key propagation paths based on the operating topology and the fluctuating load components includes: calculating the corresponding fluctuation intensity index based on the time series of the fluctuating load components of each power metering box; comparing the fluctuation intensity index with a preset fluctuation intensity threshold, and identifying power metering boxes with fluctuation intensity indices greater than the threshold as key fluctuation sources; for each key fluctuation source, based on the operating topology, finding and determining a downstream path that starts from the key fluctuation source node, propagates along the topological connection direction, and whose path length does not exceed a preset depth value, as the key propagation path.

Images