Method and system for monitoring and analyzing operation data of electric energy meter based on electric red-oned process

Abstract

The invention belongs to the technical field of operation monitoring of electric energy meters, and discloses an operation data monitoring and analyzing method and system based on an electric red-keyed electric energy meter, and the method comprises the steps: deploying the electric red-keyed electric energy meter in a low-voltage transformer area, collecting multi-dimensional electric energy operation parameters in real time, carrying out the collection and time axis reconstruction of the multi-dimensional electric energy operation parameters according to nodes, and carrying out the monitoring and analysis of the operation data of the electric red-keyed electric energy meter. Generating a load behavior time sequence, identifying a load behavior event by using a feature detection algorithm, and representing the load behavior event as a behavior lexical element; analyzing a hierarchical dependency relationship between a trunk mode and a modification structure of a load behavior event by constructing a syntactic tree structure of a load behavior, identifying a structural abnormal load mode, and generating an abnormal disturbance subgraph; fusing the abnormal disturbance sub-graph with a preset power supply topological structure, constructing a path graph with a body, executing a graph transformation operation group in the path construction process, and generating a multi-disturbance path sub-graph; and the accuracy, timeliness and reliability of load anomaly detection of the power distribution area are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric energy meter operation monitoring, and more specifically, to a method and system for monitoring and analyzing electric energy meter operation data based on electric Honghua. Background Art

[0002] Patent publication number CN112684399A discloses an artificial intelligence-based method and system for fitting data of electric energy meter operation error monitoring. First, the electric energy meter operation data and its corresponding error monitoring data set are obtained. Secondly, the current monitoring data is determined to perform error analysis to obtain the power loss error data of the power load node data. Then, based on the power loss error data, the current monitoring data corresponding to the electric energy meter operation data is error fitted to obtain the current error fitting data and added to the corresponding error monitoring data set. Then, when the error monitoring data set meets the set conditions, the operation error identification result of the electric energy meter to be monitored is obtained. In this way, by analyzing, fitting and iterating the operation errors of different electric energy meter operation data of the electric energy meter to be monitored, the global and continuous operation error identification results of the electric energy meter to be monitored can be obtained, thereby providing an accurate and reliable correction basis for the error correction of the electric energy meter to be monitored.

[0003] The existing electricity meter operation data monitoring and analysis methods and systems have the following major problems: Existing technologies generally use methods based on time windows, threshold judgment or pattern classification to detect load anomalies. They only consider the local amplitude changes of behavioral events, but ignore the dependencies between events, structural combination forms and logical evolution paths, which makes it difficult to effectively identify complex abnormal patterns caused by structural disturbances. In existing methods, all fluctuation behaviors that exceed the threshold are often regarded as abnormal, but there are a large number of normal fluctuation behaviors in the actual power operation process. Traditional methods lack hierarchical analysis of behavioral structures and cannot distinguish structural anomalies from occasional disturbances based on event combination patterns, which may lead to false alarms and missed reports. Most existing technologies output detection results in the form of time series or statistical indicators, and cannot extract structured abnormal subgraphs for further analysis. This limits the support capabilities for subsequent topology adaptation, path risk propagation or graph isomorphism calculations.

[0004] Existing technologies rely on a single electric energy parameter or simple statistical features, which makes it difficult to accurately reflect the fine-grained changes in the load of the substation on different strategic paths, resulting in insufficient anomaly detection and evolution analysis. Traditional methods usually ignore the correlation between load disturbance paths and the evolution of topological structures, lack a dynamic path evolution mechanism based on graph structure, and are difficult to capture the evolution trend and nonlinear deviation of the load state. Existing load state distance metrics mostly use Euclidean distance or other symmetric metrics, which cannot effectively express the asymmetry and directionality of state changes during disturbances, resulting in insufficient sensitivity to load evolution deviations. Traditional parameter update methods mostly use optimization strategies such as gradient descent, ignoring the guiding role of changes in the load disturbance path structure on the dynamic adjustment of parameters. It is difficult to ensure the consistency of parameter evolution with the actual load topology evolution direction, affecting the accuracy of early warning.

[0005] In view of this, the present invention proposes a method and system for monitoring and analyzing the operating data of electric energy meters based on Dianhonghua to solve the above problems. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for monitoring and analyzing the operating data of an electric energy meter based on Dianhonghua, comprising: S1. Deploy electric energy meters in the preset low-voltage area to collect multi-dimensional electric energy operation parameters in real time, aggregate and reconstruct the time axis according to the collection nodes, generate load behavior time series, use feature detection algorithms to identify load behavior events, and represent load behavior events as behavior words; S2. Construct a syntactic tree structure of load behavior, analyze the hierarchical dependency between the main pattern and modified structure of load behavior events, identify structural abnormal load patterns, and generate abnormal disturbance subgraphs; S3. Integrate the abnormal perturbation subgraph with the preset power supply topology to construct an embodied path graph. During the path construction process, perform a graph transformation operation group to generate multiple perturbation path subgraphs. Based on the perturbation structure isomorphism index, the multiple perturbation path subgraphs are scored for their fitness and a perturbation path similarity matrix is ​​constructed. S4. Based on the disturbance path similarity matrix, a parameterized probability distribution family is established to construct the manifold space of load states. The geodesic distance between load states is calculated based on the Fisher information degree to dynamically characterize the load state evolution deviation. S5. Based on multi-dimensional electric energy operation parameters, a fuzzy comprehensive evaluation matrix is ​​constructed to perform substation health scoring; and early warning is issued for load state evolution deviations, generating a set of abnormal events with timestamps, and transmitting the substation health score and abnormal event set to the data monitoring and analysis terminal.

[0007] Preferably, the method for obtaining the multi-dimensional electric energy operation parameters includes: Dianhonghua electricity meters are deployed at the main incoming line location and key branch outgoing lines of the preset low-voltage substation. The Dianhonghua electricity meters are equipped with voltage transformers, current transformers and analog-to-digital converters to collect multi-dimensional electricity operation parameters in real time; multi-dimensional electricity operation parameters include voltage, current, active power, reactive power, apparent power, power factor, system frequency, electric energy, harmonic distribution and equipment temperature; Dianhonghua electricity meters perform clock calibration based on the global positioning system GPS to unify the time between different collection nodes.

[0008] Preferably, the method of expressing a load behavior event as a behavior word comprises: Dianhonghua's electric energy meter encapsulates the collected multi-dimensional electric energy operation parameters according to the node ID and timestamp through the communication network and transmits them to the data monitoring and analysis terminal. The data monitoring and analysis terminal performs verification, decoding and interpolation compensation processing on the collected data, aligns and fuses the multi-dimensional electric energy operation parameters of different nodes according to the timestamp, and reconstructs the time axis sequence of the electric energy operation parameters. A feature detection algorithm is applied based on the reconstructed timeline sequence to identify load change events, which include power surges, current starts, and voltage fluctuations. Each detected load change event is represented as a behavior word containing behavior type, start and end time, amplitude change, and phase information.

[0009] Preferably, the method for identifying structural abnormal load patterns and generating abnormal disturbance subgraphs includes: The behavior tokens are arranged in chronological order into a behavior token sequence, and the behavior token sequence is directly mapped into a syntactic tree structure representing hierarchical dependencies. Different nodes in the syntactic tree structure represent load behavior events, including trunk nodes and modification nodes. Edges represent the dependency relationship between trunk nodes and modification nodes. A trunk modification judgment function is defined to identify trunk nodes and modification nodes in the syntactic tree structure, and a structural pattern analysis is performed on the constructed syntactic tree structure to match the structural abnormal load pattern; the structural abnormal load pattern includes trunk mutation type abnormality, modification abnormality type abnormality and hierarchical mismatch type abnormality; the nodes and hierarchical dependencies are extracted from the syntactic tree structure matched to the structural abnormal load pattern to form an abnormal disturbance subgraph.

[0010] Preferably, the method for generating the multi-perturbation path subgraph includes: The load behavior events contained in the identified abnormal disturbance subgraph are mapped to the corresponding power supply nodes in the preset power supply topology to form an embodied path graph. The embodied path graph is based on the node and line connection relationship in the preset power supply topology, and the nodes where the load behavior events occur are marked in the embodied path graph; On the embodied path graph, based on the perturbation propagation modeling requirements, an operation group containing different graph transformation operations is executed. The operation group includes pruning operations, removing redundant branches, expansion operations, path replacement operations and edge weight adjustment operations; through the operation group of graph transformation operations, a multi-perturbation path subgraph is constructed.

[0011] Preferably, the method for constructing the disturbance path similarity matrix includes: By mapping the node and edge structures in the abnormal perturbation subgraph to the corresponding node and edge structures in each perturbation path subgraph, the perturbation structure isomorphism index is calculated. The perturbation structure isomorphism index is the ratio of the number of edges retained in the perturbation path subgraph in the abnormal perturbation subgraph to the total number of edges in the abnormal perturbation subgraph. The perturbation structural isomorphism index is calculated for all perturbation path subgraphs to form a path fitness score vector. The structural similarity between any two perturbation path subgraphs is compared using the Jaccard similarity coefficient, and the structural similarities between all perturbation path subgraphs are organized into a perturbation path similarity matrix.

[0012] Preferably, the method for constructing the manifold space of load states includes: The constructed perturbation path similarity matrix is ​​used to perform dimensionality reduction embedding processing on different perturbation path subgraphs. The dimensionality reduction embedding process adopts a manifold learning algorithm, including any one of local linear embedding and isometric mapping, to embed the perturbation path subgraph into a preset manifold space. In the manifold space, based on the load state changes caused by each perturbation path subgraph in the corresponding time period, a joint sample set between the perturbation path embedding vector and the load state change is constructed. The load state change includes the amplitude fluctuations of current, voltage, active power and power factor.

[0013] Preferably, the method for dynamically characterizing the load state evolution deviation includes: Based on the perturbation path similarity matrix between multiple perturbation path subgraphs, a manifold learning algorithm is used to reduce the dimensionality of each perturbation path subgraph to obtain the corresponding low-dimensional vector set. The current, voltage, and power changes within the corresponding time period of the perturbation path subgraph are then combined to construct a joint sample. The exponential family distribution is used to model the joint samples, and a probability density function is constructed. The Fisher information matrix is ​​constructed based on the probability density function. The Fisher information matrix is ​​used to measure the identifiability and sensitivity between the disturbance path and the load state. The geodesic distance between the load states corresponding to the parameter vectors at any two moments is defined as ; A geometric migration mechanism guided by perturbation graphs is introduced to embed the perturbation path subgraphs at adjacent moments into a unified manifold space. The graph migration parameter vector is calculated based on the graph structure difference. The parameter vector is updated and adjusted to make the evolution trend of the parameter vector consistent with the topological evolution direction of the perturbation path subgraph, thereby dynamically characterizing the load state evolution deviation.

[0014] Preferably, the method for performing area health scoring includes: Based on the actual value of each power parameter in the multidimensional power operation parameters, different levels of fuzzy membership functions are constructed to map the power parameter values ​​to membership values ​​of different health levels, including excellent, good, medium, and poor. The membership of each type of power parameter to different health levels is combined into a fuzzy evaluation matrix, with parameter types as rows and health levels as columns. The weight vector corresponding to each electric energy parameter is preset, and the fuzzy evaluation matrix and the weight vector are fuzzy weighted synthesized to obtain the comprehensive membership vector of each health level; the maximum membership method is used to convert the comprehensive membership vector into a substation health score value to perform substation health scoring.

[0015] Based on the Dianhonghua electric energy meter operation data monitoring and analysis system, including: The time series acquisition activation module deploys electric energy meters in the preset low-voltage area to collect multi-dimensional power operation parameters in real time, aggregate them according to the acquisition nodes and reconstruct the time axis to generate load behavior time series, use feature detection algorithms to identify load behavior events, and represent load behavior events as behavior words; The load structured analysis module constructs a syntactic tree structure of load behavior, analyzes the hierarchical dependency between the main pattern and modified structure of load behavior events, identifies structural abnormal load patterns, and generates abnormal disturbance subgraphs; The path load response module integrates the abnormal disturbance subgraph with the preset power supply topology to construct an embodied path graph. During the path construction process, a graph transformation operation group is executed to generate multiple disturbance path subgraphs. Based on the disturbance structure isomorphism index, the multiple disturbance path subgraphs are scored for their fitness and a disturbance path similarity matrix is ​​constructed. The manifold adaptive evolution module establishes a parameterized probability distribution family based on the disturbance path similarity matrix, constructs the manifold space of load states, and calculates the geodesic distance between load states based on the Fisher information degree, dynamically characterizing the load state evolution deviation; The comprehensive evaluation dynamic early warning module constructs a fuzzy comprehensive evaluation matrix based on multi-dimensional power operation parameters to perform substation health scoring; it also issues early warnings for load state evolution deviations, generates a set of abnormal events with timestamps, and transmits the substation health score and abnormal event set to the data monitoring and analysis terminal.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention models load behavior events as word units and constructs a syntactic tree structure of behavior word units to express the master-slave dependency and modification relationship between events, so that the originally isolated behavior points are transformed into directed dependency structures with contextual relationships, which greatly improves the semantic expression ability of behavior modeling. The defined trunk modification judgment function comprehensively considers the significance score of the event and its dependency depth, so that the system can automatically identify the core driving nodes and modification nodes in the behavior chain, supports hierarchical analysis of the behavior tree, and effectively identifies structural abnormality patterns caused by trunk transitions, modification disorders or hierarchical mismatches; structural abnormalities are divided into three categories: trunk mutation type, modification abnormality type and hierarchical mismatch type. The substructure with abnormal pattern in the syntax tree is identified through structural pattern analysis, avoiding the situation where all mutation behaviors are classified as abnormal, effectively reducing misjudgment and improving the accuracy of substation operation status assessment.

[0017] By introducing multiple perturbation path subgraphs and their similarity matrices, and employing advanced manifold learning algorithms, this approach achieves a low-dimensional representation of complex load perturbation paths, effectively preserving the structural characteristics and dynamic correlations of multi-strategy paths and improving the precision and expressiveness of load state characterization. By probabilistically modeling joint samples and constructing a Fisher information matrix, this approach quantifies the sensitivity and distinguishability between perturbation paths and load states, enhancing the accuracy and robustness of load state identification. By replacing the traditional symmetric distance with the geodesic distance in parameter space, this approach enhances the ability to capture nonlinear and asymmetric features of load state evolution and improves the sensitivity of dynamic deviation characterization. A structural migration transformation function based on the graph edit distance guides the dynamic update of parameter vectors, aligning parameter evolution trends with the topological changes of the perturbation path subgraph. This addresses the lack of structural awareness in traditional parameter updates and improves the accuracy and reliability of load state evolution prediction. By reflecting the asymmetric evolution of load states at multiple moments in real time, this approach improves the efficiency and accuracy of identifying abnormal perturbations, providing strong technical support for substation health assessment and intelligent early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of the method for monitoring and analyzing the operating data of electric energy meters based on Dianhonghua technology of the present invention; Figure 2 This is a schematic diagram of the structure of the electric energy meter operation data monitoring and analysis system based on Dianhonghua of the present invention; Figure 3 This is a flow chart of the method provided by the present invention for representing load behavior events as behavior tokens. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0020] Example 1 See also Figure 1 and Figure 3 As shown, the embodiment 1 further illustrates the method for monitoring and analyzing the operation data of the electric energy meter based on the electric Honghua power meter proposed in the present invention, including: With the continuous development of power systems, the monitoring and analysis of electricity meter operating data plays a vital role in load management, anomaly detection, and power quality assurance within distribution substations. Existing electricity meter operating data monitoring and analysis methods and systems primarily rely on time window analysis, threshold judgment, or pattern classification techniques to detect and identify abnormal load behavior. However, these traditional methods suffer from the following major shortcomings: Existing technologies often focus solely on local amplitude changes in behavioral events, ignoring the dependencies between load behavioral events, their structural combinations, and their logical evolution paths. This makes it difficult to effectively identify abnormal patterns caused by complex structural disturbances. Typically, all fluctuations exceeding a preset threshold are considered abnormal, failing to distinguish between normal fluctuations and true anomalies, which can easily lead to false positives and missed reports. Furthermore, traditional methods lack a hierarchical analysis of load behavioral structures and are unable to distinguish structural anomalies from sporadic disturbances based on event combination patterns, limiting the accuracy of anomaly identification.

[0021] Existing technologies mostly output anomaly detection results in the form of time series or statistical indicators, lacking structured expression of abnormal behavior. It is difficult to form anomaly subgraphs that are convenient for subsequent topology adaptation, path risk propagation, and graph isomorphism calculation, which reduces the depth and efficiency of anomaly analysis and decision support.

[0022] Traditional methods often rely on single power parameters or simple statistical features, making it difficult to fully reflect the fine-grained dynamic changes in distribution network loads along multi-strategy disturbance paths, resulting in insufficient anomaly detection and load evolution analysis capabilities. Existing technologies also generally ignore the correlation and topological evolution between load disturbance paths, lack a graph-based dynamic path evolution mechanism, and have difficulty capturing nonlinear changes in load states.

[0023] Current methods for measuring load state evolution distances often use Euclidean distance or other symmetric metrics, which cannot effectively represent the asymmetry and directionality of load disturbances, reducing sensitivity to load evolution deviations. Traditional parameter update strategies, primarily based on numerical optimization methods like gradient descent, lack the ability to perceive and guide topological changes in load disturbance paths. This makes it difficult to ensure consistency between parameter evolution trends and the actual direction of load topology evolution, thus impacting the accuracy and real-time nature of anomaly warnings.

[0024] In order to effectively solve the above problems, the present invention proposes a method for monitoring and analyzing the operating data of electric energy meters based on Dianhonghua, including: S1. Deploy electric energy meters in the preset low-voltage area to collect multi-dimensional electric energy operation parameters in real time, aggregate and reconstruct the time axis according to the collection nodes, generate load behavior time series, use feature detection algorithms to identify load behavior events, and represent load behavior events as behavior words; S2. Construct a syntactic tree structure of load behavior, analyze the hierarchical dependency between the main pattern and modified structure of load behavior events, identify structural abnormal load patterns, and generate abnormal disturbance subgraphs; S3. Integrate the abnormal perturbation subgraph with the preset power supply topology to construct an embodied path graph. During the path construction process, perform a graph transformation operation group to generate multiple perturbation path subgraphs. Based on the perturbation structure isomorphism index, the multiple perturbation path subgraphs are scored for their fitness and a perturbation path similarity matrix is ​​constructed. S4. Based on the disturbance path similarity matrix, a parameterized probability distribution family is established to construct the manifold space of load states. The geodesic distance between load states is calculated based on the Fisher information degree to dynamically characterize the load state evolution deviation. S5. Based on multi-dimensional electric energy operation parameters, a fuzzy comprehensive evaluation matrix is ​​constructed to perform substation health scoring; and early warning is issued for load state evolution deviations, generating a set of abnormal events with timestamps, and transmitting the substation health score and abnormal event set to the data monitoring and analysis terminal.

[0025] The method for obtaining multi-dimensional electric energy operation parameters includes: Dianhonghua electricity meters are deployed at the main incoming line location and key branch outgoing lines of the preset low-voltage substation. The Dianhonghua electricity meters are equipped with voltage transformers, current transformers and analog-to-digital converters to collect multi-dimensional electricity operation parameters in real time; multi-dimensional electricity operation parameters include voltage, current, active power, reactive power, apparent power, power factor, system frequency, electric energy, harmonic distribution and equipment temperature; Dianhonghua electricity meters perform clock calibration based on the global positioning system GPS to unify the time between different collection nodes.

[0026] Methods for representing load behavior events as behavior tokens include: Dianhonghua's electric energy meter encapsulates the collected multi-dimensional electric energy operation parameters according to the node ID and timestamp through the communication network and transmits them to the data monitoring and analysis terminal. The data monitoring and analysis terminal performs verification, decoding and interpolation compensation processing on the collected data, aligns and fuses the multi-dimensional electric energy operation parameters of different nodes according to the timestamp, and reconstructs the time axis sequence of the electric energy operation parameters. A feature detection algorithm is applied based on the reconstructed timeline sequence to identify load change events, which include power surges, current starts, and voltage fluctuations. Each detected load change event is represented as a behavior word containing behavior type, start and end time, amplitude change, and phase information.

[0027] The method for identifying abnormal structural load patterns and generating abnormal disturbance submaps includes: The behavior tokens are arranged in chronological order into a behavior token sequence, and the behavior token sequence is directly mapped into a syntactic tree structure representing hierarchical dependencies. Different nodes in the syntactic tree structure represent load behavior events, including trunk nodes and modification nodes. Edges represent the dependency relationship between trunk nodes and modification nodes. Define the trunk modification judgment function to identify the trunk nodes and modification nodes in the syntax tree structure, and perform structural pattern analysis on the constructed syntax tree structure to match the structural abnormal load pattern; The backbone modification decision function is ;in, Represents a node in a syntax tree structure; Representation node The behavioral significance score is composed of a weighted combination of the power change amount, change duration and number of changes in the electric energy meter; represents the behavior significance score threshold, only When , the node is qualified to be considered as a backbone; Representation node The dependency depth in the syntax tree is the number of layers from the node to the root node. , indicating that it is the root node, it is regarded as the trunk node; the structural abnormal load pattern includes trunk mutation type abnormality, modification abnormality type abnormality and hierarchical mismatch type abnormality; the nodes and hierarchical dependencies are extracted from the syntactic tree structure matched to the structural abnormal load pattern to form an abnormal disturbance subgraph.

[0028] The following problems existing in the existing technology are solved: the existing technology generally adopts methods based on time windows, threshold judgment or pattern classification to detect load anomalies, which only considers the local amplitude changes of behavioral events, but ignores the dependencies between events, structural combination forms and logical evolution paths, resulting in difficulty in effectively identifying complex abnormal patterns caused by structural disturbances. In existing methods, all fluctuation behaviors that exceed the threshold are often regarded as abnormal, but there are a large number of normal fluctuation behaviors in the actual power operation process. Traditional methods lack hierarchical analysis of behavioral structures and cannot distinguish structural anomalies from occasional disturbances from the event combination pattern, which may lead to false alarms and missed reports. Most of the current existing technologies output detection results in the form of time series or statistical indicators, and cannot extract structured abnormal subgraphs for further analysis, which limits the support capabilities for subsequent topology adaptation, path risk propagation or graph isomorphism calculations.

[0029] Compared with the existing technology, the beneficial effects are as follows: load behavior events are modeled as word units, and by constructing a syntactic tree structure of behavior word units, the master-slave dependency and modification relationship between events is expressed, so that the originally isolated behavior points are transformed into directed dependency structures with contextual relationships, which greatly improves the semantic expression ability of behavior modeling. The defined trunk modification judgment function comprehensively considers the significance score of the event and its dependency depth, so that the system can automatically identify the core driving nodes and modification nodes in the behavior chain, support hierarchical analysis of the behavior tree, and effectively identify structural abnormality patterns caused by trunk transitions, modification disorders or hierarchical mismatches; structural abnormalities are divided into three categories: trunk mutation type, modification abnormality type and hierarchical mismatch type. Through structural pattern analysis, substructures with abnormal patterns in the syntax tree are identified, avoiding the situation where all mutation behaviors are classified as abnormal, effectively reducing misjudgments and improving the accuracy of substation operation status assessment.

[0030] The method for generating multiple perturbation path subgraphs includes: The load behavior events contained in the identified abnormal disturbance subgraph are mapped to the corresponding power supply nodes in the preset power supply topology to form an embodied path graph. The embodied path graph is based on the node and line connection relationship in the preset power supply topology, and the nodes where the load behavior events occur are marked in the embodied path graph; On the embodied path graph, based on the disturbance propagation modeling requirements, an operation group containing different graph transformation operations is executed. The operation group includes pruning operations, removing redundant branches, expansion operations, and extending the disturbance impact to redundant power supply paths; path replacement operations and edge weight adjustment operations; used to correct path propagation weights according to device status, current direction or historical load records; and through the operation group of graph transformation operations, a multi-disturbance path subgraph is constructed.

[0031] Methods for constructing the perturbation path similarity matrix include: By mapping the node and edge structures in the abnormal perturbation subgraph to the corresponding node and edge structures in each perturbation path subgraph, the perturbation structure isomorphism index is calculated. The perturbation structure isomorphism index is the ratio of the number of edges retained in the perturbation path subgraph in the abnormal perturbation subgraph to the total number of edges in the abnormal perturbation subgraph. The perturbation structural isomorphism index is calculated for all perturbation path subgraphs to form a path fitness score vector. The structural similarity between any two perturbation path subgraphs is compared using the Jaccard similarity coefficient, and the structural similarities between all perturbation path subgraphs are organized into a perturbation path similarity matrix.

[0032] Methods for constructing the manifold space of load states include: The constructed perturbation path similarity matrix is ​​used to perform dimensionality reduction embedding processing on different perturbation path subgraphs. The dimensionality reduction embedding process adopts a manifold learning algorithm, including any one of local linear embedding and isometric mapping, to embed the perturbation path subgraph into a preset manifold space. In the manifold space, based on the load state changes caused by each perturbation path subgraph in the corresponding time period, a joint sample set between the perturbation path embedding vector and the load state change is constructed. The load state change includes the amplitude fluctuations of current, voltage, active power and power factor.

[0033] Methods for dynamically characterizing load state evolution deviations include: Based on the perturbation path similarity matrix between multiple perturbation path subgraphs, a manifold learning algorithm is used to reduce the dimensionality of each perturbation path subgraph to obtain the corresponding low-dimensional vector set. The current, voltage, and power changes within the corresponding time period of the perturbation path subgraph are then combined to construct a joint sample. The exponential family distribution is used to model the joint sample and construct the probability density function. The probability density function is: ;in, Indicates that given a parameter vector Under the condition of Probability density of occurrence; represents the parameter vector; represents sufficient statistics; represents the normalization function; According to the probability density function, the Fisher information matrix is ​​constructed. Through the Fisher information matrix, the Fisher information matrix is ;in, Represents the second-order partial derivative of the parameter vector; measures the identifiability and sensitivity between the disturbance path and the load state; defines the geodesic distance between the load states corresponding to the parameter vectors at any two moments as ; The perturbation graph-guided geometric migration mechanism is introduced to embed the perturbation path subgraphs at adjacent moments into a unified manifold space. The graph migration parameter vector is calculated based on the graph structure difference. The parameter vector is updated and adjusted. The adjusted parameter vector is ;in, Indicates The parameter vector after the update at each moment; Indicates that at present The parameter vector of the moment; represents the step migration coefficient; Represents the graph structure migration transformation function, which can be defined by the graph edit distance change rate; Indicates the perturbation path subgraph from the previous moment To the present The structural transformation process at each moment is realized; the evolution trend of the parameter vector is made consistent with the topological evolution direction of the perturbation path subgraph, and the load state evolution deviation is dynamically characterized.

[0034] The following technical problems existing in the existing technology are solved: the existing technology relies on a single electric energy parameter or simple statistical features, which makes it difficult to accurately reflect the fine-grained changes in the load of the substation on different strategic paths, resulting in insufficient anomaly detection and evolution analysis. Traditional methods usually ignore the correlation between load disturbance paths and the evolution of topological structures, lack a dynamic path evolution mechanism based on graph structure, and are difficult to capture the evolution trend and nonlinear deviation of the load state. Existing load state distance metrics mostly use Euclidean distance or other symmetric metrics, which cannot effectively express the asymmetry and directionality of state changes during disturbances, resulting in insufficient sensitivity to load evolution deviations. Traditional parameter update methods mostly use optimization strategies such as gradient descent, ignoring the guiding role of changes in the load disturbance path structure on the dynamic adjustment of parameters. It is difficult to ensure the consistency of parameter evolution with the actual load topology evolution direction, affecting the accuracy of early warning.

[0035] Compared with existing technologies, this approach offers several advantages: By introducing multiple perturbation path subgraphs and their similarity matrices, and employing advanced manifold learning algorithms, it achieves a low-dimensional representation of complex load perturbation paths, effectively preserving the structural characteristics and dynamic correlations of multi-strategy paths and improving the precision and expressiveness of load state characterization. By constructing a Fisher information matrix through probabilistic modeling of joint samples, it quantifies the sensitivity and distinguishability between perturbation paths and load states, enhancing the accuracy and robustness of load state identification. By replacing the traditional symmetric distance with the geodesic distance in parameter space, it enhances the ability to capture nonlinear and asymmetric features of load state evolution and improves the sensitivity of dynamic deviation characterization. A structural migration transformation function based on the graph edit distance guides the dynamic update of parameter vectors, aligning parameter evolution trends with the topological changes of the perturbation path subgraph. This addresses the lack of structural awareness in traditional parameter updates and improves the accuracy and reliability of load state evolution prediction. It also reflects the asymmetric evolution of load states at multiple moments in real time, significantly improving the efficiency and accuracy of abnormal perturbation identification, providing strong technical support for substation health assessment and intelligent early warning.

[0036] Methods for conducting area health scoring include: Based on the actual value of each power parameter in the multidimensional power operation parameters, different levels of fuzzy membership functions are constructed to map the power parameter values ​​to membership values ​​of different health levels, including excellent, good, medium, and poor. The membership of each type of power parameter to different health levels is combined into a fuzzy evaluation matrix, with parameter types as rows and health levels as columns. The weight vector corresponding to each electric energy parameter is preset, and the fuzzy evaluation matrix and the weight vector are fuzzy weighted synthesized to obtain the comprehensive membership vector of each health level; the maximum membership method is used to convert the comprehensive membership vector into a substation health score value to perform substation health scoring.

[0037] This embodiment, by modeling load behavior events as words and constructing a syntactic tree structure of behavior words to express the master-slave dependency and modification relationship between events, turns the originally isolated behavior points into directed dependency structures with contextual relationships, greatly improving the semantic expression ability of behavior modeling. The defined trunk modification judgment function comprehensively considers the significance score of the event and its dependency depth, so that the system can automatically identify the core driving nodes and modification nodes in the behavior chain, supports hierarchical analysis of the behavior tree, and effectively identifies structural abnormality patterns caused by trunk transitions, modification disorders or hierarchical mismatches; divides structural abnormalities into three categories: trunk mutation type, modification abnormality type and hierarchical mismatch type, and identifies substructures with abnormal patterns in the syntax tree through structural pattern analysis, avoiding the situation where all mutation behaviors are classified as abnormal, effectively reducing misjudgments and improving the accuracy of substation operation status assessment.

[0038] By introducing multiple perturbation path subgraphs and their similarity matrices, and employing advanced manifold learning algorithms, this approach achieves a low-dimensional representation of complex load perturbation paths, effectively preserving the structural characteristics and dynamic correlations of multi-strategy paths and improving the precision and expressiveness of load state characterization. By probabilistically modeling joint samples and constructing a Fisher information matrix, this approach quantifies the sensitivity and distinguishability between perturbation paths and load states, enhancing the accuracy and robustness of load state identification. By replacing the traditional symmetric distance with the geodesic distance in parameter space, this approach enhances the ability to capture nonlinear and asymmetric features of load state evolution and improves the sensitivity of dynamic deviation characterization. A structural migration transformation function based on the graph edit distance guides the dynamic update of parameter vectors, aligning parameter evolution trends with the topological changes of the perturbation path subgraph. This addresses the lack of structural awareness in traditional parameter updates and improves the accuracy and reliability of load state evolution prediction. By reflecting the asymmetric evolution of load states at multiple moments in real time, this approach improves the efficiency and accuracy of identifying abnormal perturbations, providing strong technical support for substation health assessment and intelligent early warning.

[0039] Example 2 See also Figure 2 As shown, this embodiment is based on the Dianhonghua electric energy meter operation data monitoring and analysis system, including: The time series acquisition activation module deploys electric energy meters in the preset low-voltage area to collect multi-dimensional power operation parameters in real time, aggregate them according to the acquisition nodes and reconstruct the time axis to generate load behavior time series, use feature detection algorithms to identify load behavior events, and represent load behavior events as behavior words; The load structured analysis module constructs a syntactic tree structure of load behavior, analyzes the hierarchical dependency between the main pattern and modified structure of load behavior events, identifies structural abnormal load patterns, and generates abnormal disturbance subgraphs; The path load response module integrates the abnormal disturbance subgraph with the preset power supply topology to construct an embodied path graph. During the path construction process, a graph transformation operation group is executed to generate multiple disturbance path subgraphs. Based on the disturbance structure isomorphism index, the multiple disturbance path subgraphs are scored for their fitness and a disturbance path similarity matrix is ​​constructed. The manifold adaptive evolution module establishes a parameterized probability distribution family based on the disturbance path similarity matrix, constructs the manifold space of load states, and calculates the geodesic distance between load states based on the Fisher information degree, dynamically characterizing the load state evolution deviation; The comprehensive evaluation dynamic early warning module constructs a fuzzy comprehensive evaluation matrix based on multi-dimensional power operation parameters to perform substation health scoring; it also issues early warnings for load state evolution deviations, generates a set of abnormal events with timestamps, and transmits the substation health score and abnormal event set to the data monitoring and analysis terminal.

[0040] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0041] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for users of ordinary skill in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. Based on the Dianhonghua electric energy meter operation data monitoring and analysis method, it is characterized by: include: S1. Deploy electric energy meters in the preset low-voltage area to collect multi-dimensional electric energy operation parameters in real time, aggregate and reconstruct the time axis according to the collection nodes, generate load behavior time series, use feature detection algorithms to identify load behavior events, and represent load behavior events as behavior words; S2. Construct a syntactic tree structure of load behavior, analyze the hierarchical dependency between the main pattern and modified structure of load behavior events, identify structural abnormal load patterns, and generate abnormal disturbance subgraphs; S3, integrating the abnormal disturbance subgraph with the preset power supply topology to construct an embodied path graph, and performing a graph transformation operation group during the path construction process to generate a multi-disturbance path subgraph; Based on the perturbation structure isomorphism index, the fitness of multiple perturbation path subgraphs is scored and the perturbation path similarity matrix is ​​constructed; S4. Based on the disturbance path similarity matrix, a parameterized probability distribution family is established to construct the manifold space of load states. The geodesic distance between load states is calculated based on the Fisher information degree to dynamically characterize the load state evolution deviation. S5. Based on multi-dimensional electric energy operation parameters, a fuzzy comprehensive evaluation matrix is ​​constructed to perform substation health scoring; and early warning is issued for load state evolution deviations, generating a set of abnormal events with timestamps, and transmitting the substation health score and abnormal event set to the data monitoring and analysis terminal.

2. The method for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua according to claim 1 is characterized in that: The method for obtaining the multi-dimensional electric energy operation parameters includes: Dianhonghua electricity meters are deployed at the main incoming line location and key branch outgoing lines of the preset low-voltage substation. The Dianhonghua electricity meters are equipped with voltage transformers, current transformers and analog-to-digital converters to collect multi-dimensional electricity operation parameters in real time; multi-dimensional electricity operation parameters include voltage, current, active power, reactive power, apparent power, power factor, system frequency, electric energy, harmonic distribution and equipment temperature; Dianhonghua electricity meters perform clock calibration based on the global positioning system GPS to unify the time between different collection nodes.

3. The method for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua according to claim 2 is characterized in that: The method for expressing load behavior events as behavior tokens includes: Dianhonghua's electric energy meter encapsulates the collected multi-dimensional electric energy operation parameters according to the node ID and timestamp through the communication network and transmits them to the data monitoring and analysis terminal. The data monitoring and analysis terminal performs verification, decoding and interpolation compensation processing on the collected data, aligns and fuses the multi-dimensional electric energy operation parameters of different nodes according to the timestamp, and reconstructs the time axis sequence of the electric energy operation parameters. A feature detection algorithm is applied based on the reconstructed timeline sequence to identify load change events, which include power surges, current starts, and voltage fluctuations. Each detected load change event is represented as a behavior word containing behavior type, start and end time, amplitude change, and phase information.

4. The method for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua according to claim 3 is characterized in that: The method for identifying structural abnormal load patterns and generating abnormal disturbance subgraphs includes: The behavior tokens are arranged in chronological order into a behavior token sequence, and the behavior token sequence is directly mapped into a syntactic tree structure representing hierarchical dependencies. Different nodes in the syntactic tree structure represent load behavior events, including trunk nodes and modification nodes. Edges represent the dependency relationship between trunk nodes and modification nodes. A trunk modification judgment function is defined to identify trunk nodes and modification nodes in the syntactic tree structure, and a structural pattern analysis is performed on the constructed syntactic tree structure to match the structural abnormal load pattern; the structural abnormal load pattern includes trunk mutation type abnormality, modification abnormality type abnormality and hierarchical mismatch type abnormality; the nodes and hierarchical dependencies are extracted from the syntactic tree structure matched to the structural abnormal load pattern to form an abnormal disturbance subgraph.

5. The method for monitoring and analyzing the operation data of electric energy meters according to claim 4 is characterized in that: The method for generating the multi-perturbation path subgraph includes: The load behavior events contained in the identified abnormal disturbance subgraph are mapped to the corresponding power supply nodes in the preset power supply topology to form an embodied path graph. The embodied path graph is based on the node and line connection relationship in the preset power supply topology, and the nodes where the load behavior events occur are marked in the embodied path graph; On the embodied path graph, based on the perturbation propagation modeling requirements, an operation group containing different graph transformation operations is executed. The operation group includes pruning operations, removing redundant branches, expansion operations, path replacement operations and edge weight adjustment operations; through the operation group of graph transformation operations, a multi-perturbation path subgraph is constructed.

6. The method for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua according to claim 5 is characterized in that: The method for constructing the disturbance path similarity matrix includes: By mapping the node and edge structures in the abnormal perturbation subgraph to the corresponding node and edge structures in each perturbation path subgraph, the perturbation structure isomorphism index is calculated. The perturbation structure isomorphism index is the ratio of the number of edges retained in the perturbation path subgraph in the abnormal perturbation subgraph to the total number of edges in the abnormal perturbation subgraph. The perturbation structural isomorphism index is calculated for all perturbation path subgraphs to form a path fitness score vector. The structural similarity between any two perturbation path subgraphs is compared using the Jaccard similarity coefficient, and the structural similarities between all perturbation path subgraphs are organized into a perturbation path similarity matrix.

7. The method for monitoring and analyzing the operation data of electric energy meters according to claim 6 is characterized in that: The method for constructing the manifold space of load states includes: The constructed perturbation path similarity matrix is ​​used to perform dimensionality reduction embedding processing on different perturbation path subgraphs. The dimensionality reduction embedding process adopts a manifold learning algorithm, including any one of local linear embedding and isometric mapping, to embed the perturbation path subgraph into a preset manifold space. In the manifold space, based on the load state changes caused by each perturbation path subgraph in the corresponding time period, a joint sample set between the perturbation path embedding vector and the load state change is constructed. The load state change includes the amplitude fluctuations of current, voltage, active power and power factor.

8. The method for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua according to claim 7 is characterized in that: The method for dynamically characterizing the load state evolution deviation includes: Based on the perturbation path similarity matrix between multiple perturbation path subgraphs, a manifold learning algorithm is used to reduce the dimensionality of each perturbation path subgraph to obtain the corresponding low-dimensional vector set. The current, voltage, and power changes within the corresponding time period of the perturbation path subgraph are then combined to construct a joint sample. The exponential family distribution is used to model the joint samples, and a probability density function is constructed. The Fisher information matrix is ​​constructed based on the probability density function. The Fisher information matrix is ​​used to measure the identifiability and sensitivity between the disturbance path and the load state. The geodesic distance between the load states corresponding to the parameter vectors at any two moments is defined as ; A geometric migration mechanism guided by perturbation graphs is introduced to embed the perturbation path subgraphs at adjacent moments into a unified manifold space. The graph migration parameter vector is calculated based on the graph structure difference. The parameter vector is updated and adjusted to make the evolution trend of the parameter vector consistent with the topological evolution direction of the perturbation path subgraph, thereby dynamically characterizing the load state evolution deviation.

9. The method for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua according to claim 8 is characterized in that: The method for performing area health scoring includes: Based on the actual value of each power parameter in the multidimensional power operation parameters, different levels of fuzzy membership functions are constructed to map the power parameter values ​​to membership values ​​of different health levels, including excellent, good, medium, and poor. The membership of each type of power parameter to different health levels is combined into a fuzzy evaluation matrix, with parameter types as rows and health levels as columns. The weight vector corresponding to each electric energy parameter is preset, and the fuzzy evaluation matrix and the weight vector are fuzzy weighted synthesized to obtain the comprehensive membership vector of each health level; the maximum membership method is used to convert the comprehensive membership vector into a substation health score value to perform substation health scoring.

10. A system for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua, used to implement the method for monitoring and analyzing the operation data of electric energy meters based on Dianhonghua according to any one of claims 1 to 9, characterized in that: include: The time series acquisition activation module deploys electric energy meters in the preset low-voltage area to collect multi-dimensional power operation parameters in real time, aggregate them according to the acquisition nodes and reconstruct the time axis to generate load behavior time series, use feature detection algorithms to identify load behavior events, and represent load behavior events as behavior words; The load structured analysis module constructs a syntactic tree structure of load behavior, analyzes the hierarchical dependency between the main pattern and modified structure of load behavior events, identifies structural abnormal load patterns, and generates abnormal disturbance subgraphs; The path load response module integrates the abnormal disturbance subgraph with the preset power supply topology to construct an embodied path graph. During the path construction process, a graph transformation operation group is executed to generate multiple disturbance path subgraphs. Based on the disturbance structure isomorphism index, the multiple disturbance path subgraphs are scored for their fitness and a disturbance path similarity matrix is ​​constructed. The manifold adaptive evolution module establishes a parameterized probability distribution family based on the disturbance path similarity matrix, constructs the manifold space of load states, and calculates the geodesic distance between load states based on the Fisher information degree, dynamically characterizing the load state evolution deviation; The comprehensive evaluation dynamic early warning module constructs a fuzzy comprehensive evaluation matrix based on multi-dimensional power operation parameters to perform substation health scoring; it also issues early warnings for load state evolution deviations, generates a set of abnormal events with timestamps, and transmits the substation health score and abnormal event set to the data monitoring and analysis terminal.

Images