High-voltage user power utilization state health assessment system and method based on time sequence abrupt change point detection

Abstract

The invention provides a high-voltage user power utilization state health assessment system and method based on time sequence abrupt change point detection, and the system comprises a data collection and fusion module, an abrupt change point detection engine, a health assessment and root cause analysis module, and a visual early warning and report generation module. The data acquisition and fusion module is used for acquiring multi-dimensional time sequence power consumption data and external environment data of a high-voltage user, the internal power consumption data and the external environment data are integrated through the data acquisition and fusion module, a unified multi-source time sequence data set is formed, data integrity and consistency are improved, and a reliable basis is provided for subsequent analysis; a multi-algorithm detection unit and an integrated decision-making unit of a catastrophe point detection engine are combined, a mixed detection strategy is adopted, the accuracy and robustness of catastrophe point recognition are remarkably improved, the false alarm rate is effectively reduced, and the timeliness and reliability of power utilization state monitoring are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of smart grid technology, and in particular to a system and method for assessing the health status of high-voltage users' electricity consumption based on time-series mutation point detection. Background Technology

[0002] With the rapid development of smart grids, real-time monitoring and health assessment of the power consumption status of high-voltage users have become a key link in ensuring the safe and stable operation of the power grid.

[0003] According to the patent titled "A Method, System, Equipment, and Storage Medium for Assessing the Health Status of Power Plant Equipment under Time Series" (Patent Publication No.: CN120387360A, Patent Publication Date: 2025-07-29), the method includes the following steps: S1. Real-time acquisition of operating status data of power plant equipment and storage of the data in time series; S2. Preprocessing of the acquired data; S3. Processing the equipment operating status data to obtain the equipment predicted degradation trend coefficient and the equipment environmental coefficient, calculating the change in equipment operating status based on the changes in the previous historical period, and determining whether to issue an equipment degradation trend warning based on the comparison results; S4. Marking the equipment that has issued a degradation trend warning, acquiring various health indicators of the marked equipment, calculating the health status assessment value of the marked equipment after combining various health indicators, and performing a health assessment; S5. Comparing the health assessment value of the marked equipment with the preset health zone assessment interval to achieve the assessment of the health status of the power plant equipment.

[0004] Based on the aforementioned existing technologies, current high-voltage user power consumption health assessment systems based on time-series mutation point detection still have the following problems: traditional systems often rely on a single source for data acquisition, lacking the fusion of multi-source data, resulting in incomplete or inconsistent data; at the same time, mutation point detection methods are singular, easily producing false alarms or missed alarms, and existing systems have difficulty quickly locating the root cause after detecting anomalies, lacking in-depth causal analysis of mutation events; health assessments rely heavily on human experience, lacking objective and quantitative scoring mechanisms. Therefore, this invention provides a high-voltage user power consumption health assessment system based on time-series mutation point detection. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a high-voltage user power consumption health assessment system and method based on time series mutation point detection. This invention addresses the following problems existing in current high-voltage user power consumption health assessment systems based on time series mutation point detection: traditional systems often rely on a single data source for data acquisition, lacking the fusion of multi-source data, resulting in incomplete or inconsistent data; at the same time, mutation point detection methods are singular, easily producing false alarms or missed alarms, and existing systems struggle to quickly locate the root cause after detecting anomalies, lacking in-depth causal analysis of mutation events; health assessments rely heavily on human experience, lacking an objective and quantitative scoring mechanism.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a high-voltage user power consumption status health assessment system based on time series abrupt change point detection, comprising: The data acquisition and fusion module (1) is used to acquire multi-dimensional time-series electricity consumption data of high-voltage users and external environmental data, and to clean, align and fuse the data to form a unified multi-source time-series dataset. The mutation point detection engine (2) is connected to the data acquisition and fusion module (1) to receive multi-source time series datasets and use an integrated mutation point detection algorithm to identify and locate mutation points in the power consumption status. The health assessment and root cause analysis module (3) is connected to the mutation point detection engine (2) to calculate the user's health score and perform root cause inference based on the identified mutation point characteristics, combined with the power grid topology and user profile knowledge graph; The visualization warning and report generation module (4) is connected to the health assessment and root cause analysis module (3) to display the health assessment results, mutation point alarms and root cause analysis conclusions in a visual form and generate an automated assessment report.

[0007] In a preferred embodiment: the data acquisition and fusion module (1) specifically includes: The internal data acquisition unit (11) is used to collect multi-dimensional internal power consumption data of high-voltage users from the power consumption information acquisition system and SCADA system. The internal power consumption data includes at least: total power consumption, three-phase current, voltage, active power, reactive power, power factor and daily load curve data. External data access unit (12) is used to access external data sources through API interface, the external data including meteorological data and calendar information; The data fusion processing unit (13) is used to perform time alignment, missing value filling and data standardization processing on the internal electricity consumption data and external data based on the timestamp, so as to generate the multi-source time series dataset.

[0008] In a preferred embodiment: the system further includes edge computing nodes deployed on the user side, and the edge computing nodes are used for: Collect high-frequency local power consumption data, the sampling frequency of which is higher than that of the internal power consumption data in the data acquisition and fusion module (1); A lightweight mutation point detection algorithm is run locally to perform preliminary identification and alarm for transient faults; The preliminary identification results, alarm information and data after feature extraction are uploaded to the mutation point detection engine (2).

[0009] In a preferred embodiment: the mutation point detection engine (2) specifically includes: The multi-algorithm detection unit (21) is used to run at least two different mutation point detection algorithms in parallel, wherein the algorithms are selected from at least two of CUSUM, PELT, Binary Segmentation, and Wavelet Transform. An integrated decision unit (22) is connected to the multi-algorithm detection unit (21) and configured to perform fusion decision on the mutation point location and confidence level of each algorithm output, and determine the final mutation point set by weighted average method; The pattern recognition unit (23) is used to perform seasonal and trend decomposition on the multi-source time series dataset and to detect mutation points on the decomposed residual sequence in order to identify at least one of mean mutation, variance mutation and trend reversal.

[0010] In a preferred embodiment: when the integrated decision-making unit (22) makes a fusion decision, the specific steps are as follows: A dynamic weight is assigned to each mutation point detection algorithm, and the dynamic weight is adaptively adjusted based on the accuracy and recall metrics of the algorithm on historical data; Calculate the weighted number of votes for each candidate mutation point across all algorithms; When the weighted number of votes exceeds a preset threshold, and the spatial density of the point in different algorithm results meets the clustering requirements, it is determined as the final effective mutation point.

[0011] In a preferred embodiment: the health assessment and root cause analysis module (3) specifically includes: The causal discovery unit (31) is used to extract multivariate data within the mutation time window after detecting the mutation point, and to use the causal discovery algorithm to analyze the causal relationship between the power consumption state mutation and external environmental data and internal power consumption data of other dimensions, and to generate a candidate root cause set. The knowledge graph query unit (32) stores the equipment files, historical operation and maintenance records and power grid topology connection relationships of high voltage users. It is used to receive the candidate root cause set and combine it with the knowledge graph for reasoning and verification in order to locate the most likely faulty equipment and external factors. The health calculation unit (33) is used to calculate a quantitative health score based on the frequency, intensity, type of mutation points and the conclusions of root cause analysis. Mutations caused by internal equipment failures will be given a higher unhealthy weight.

[0012] In a preferred embodiment: the causal discovery unit (31) is specifically used for: At least one of the constraint-based PC algorithm, the fraction-based GES algorithm, and the nearest neighbor-based CCI algorithm is used to calculate the conditional independence between variables and construct a causal graph. In the causal graph, the direct cause parent node and indirect cause node that lead to the sudden change in power consumption state are identified; The physical variables corresponding to the direct cause parent node are output to the candidate root cause set.

[0013] In a preferred embodiment: the mutation point detection engine (2) further integrates a deep learning model, and the deep learning model is an autoencoder, LSTM network and Transformer architecture based on time-series data. The operation steps of the mutation point detection engine (2) are as follows: The multi-source time series dataset is input into the trained deep learning model to directly obtain the health anomaly probability and reconstruction error of the electricity consumption sequence. The reconstructed error sequence and the anomaly probability sequence are used as new detection dimensions and input into the multi-algorithm detection unit (21) to detect mutation points, so as to realize a hybrid detection strategy of traditional algorithms and deep learning models.

[0014] This invention also provides a method for assessing the power consumption status health of high-voltage users based on time-series abrupt change point detection, and a high-voltage user power consumption status health assessment system based on time-series abrupt change point detection; including the following steps: Step 1: Data Acquisition and Fusion: Acquire multi-dimensional time-series electricity consumption data and external environment data of high-voltage users, and clean, align and fuse the multi-dimensional time-series electricity consumption data and external environment data to form a unified multi-source time-series dataset; Step 2: Abrupt change detection: An integrated abrupt change detection algorithm is used to identify and locate abrupt changes in the power consumption status of the multi-source time series dataset; Step 3: Health assessment and root cause analysis: Based on the identified mutation point characteristics, combined with the power grid topology and user profile knowledge graph, calculate the power consumption status health score of high-voltage users, and infer the root causes of mutation points. Step 4: Visualization and Report Generation: The health score, mutation point alarm information, and root cause analysis conclusions are displayed in a visual format, and an automated evaluation report is generated.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This high-voltage user power consumption health assessment system based on time series mutation point detection integrates internal power consumption data and external environmental data through a data acquisition and fusion module to form a unified multi-source time series dataset, improving data integrity and consistency and providing a reliable foundation for subsequent analysis. Combining the multi-algorithm detection unit and integrated decision unit of the mutation point detection engine, a hybrid detection strategy is adopted to significantly improve the accuracy and robustness of mutation point identification, effectively reduce the false alarm rate, and ensure the timeliness and reliability of power consumption status monitoring.

[0016] 2. This high-voltage user power consumption status health assessment system based on time series mutation point detection achieves accurate location and reasoning of mutation point root causes through the causal discovery unit and knowledge graph query unit of the health assessment and root cause analysis module. Combined with the health degree calculation unit, it provides quantitative scoring, enhancing the depth and efficiency of fault diagnosis. At the same time, the visualization early warning and report generation module displays the results in an intuitive form, supports rapid decision-making and automated report generation, and improves the intelligence level of high-voltage user power consumption status management and operation and maintenance response speed.

[0017] 3. This high-voltage user power status health assessment system based on time series mutation point detection introduces edge computing nodes for local high-frequency data acquisition and lightweight mutation point detection, enabling preliminary identification and alarm of transient faults and reducing cloud processing pressure. Combined with the pattern recognition unit and deep learning model of the mutation point detection engine, it enhances the detection capabilities of mean mutations, variance mutations, and trend reversals, forming a multi-level analysis system, improving the system's adaptability and real-time performance, and providing comprehensive technical support for high-voltage user health status assessment. Attached Figure Description

[0018] Figure 1 This is a block diagram of the high-voltage user power consumption status health assessment system of the present invention; Figure 2 This is a block diagram of the data acquisition and fusion module of the present invention; Figure 3 This is a block diagram of the mutation point detection engine of the present invention; Figure 4 This is a block diagram of the health assessment and root cause analysis module of the present invention.

[0019] In the diagram: 1. Data Acquisition and Fusion Module; 11. Internal Data Acquisition Unit; 12. External Data Access Unit; 13. Data Fusion Processing Unit; 2. Mutation Point Detection Engine; 21. Multi-Algorithm Detection Unit; 22. Integrated Decision Unit; 23. Pattern Recognition Unit; 3. Health Assessment and Root Cause Analysis Module; 31. Causal Discovery Unit; 32. Knowledge Graph Query Unit; 33. Health Calculation Unit; 4. Visual Early Warning and Report Generation Module. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0022] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations according to this application; as used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise; furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0023] Please see Figures 1-4 The present invention provides a technical solution: A high-voltage user power consumption health assessment system based on time series abrupt change point detection. The high-voltage user power consumption health assessment system includes: The data acquisition and fusion module 1 is used to acquire multi-dimensional time-series electricity consumption data of high-voltage users and external environmental data, and to clean, align and fuse the data to form a unified multi-source time-series dataset. The mutation point detection engine 2 is connected to the data acquisition and fusion module 1. It is used to receive multi-source time series datasets and use an integrated mutation point detection algorithm to identify and locate mutation points in the power consumption status. The health assessment and root cause analysis module 3 is connected to the mutation point detection engine 2. It is used to calculate the user's health score and perform root cause inference based on the identified mutation point characteristics, combined with the power grid topology and user profile knowledge graph. The visualization early warning and report generation module 4 is connected to the health assessment and root cause analysis module 3. It is used to display the health assessment results, mutation point alarms and root cause analysis conclusions in a visual form and generate automated assessment reports.

[0024] In this embodiment, the data acquisition and fusion module 1 specifically includes: The internal data acquisition unit 11 is used to collect multi-dimensional internal power consumption data of high-voltage users from the power consumption information acquisition system and SCADA system. The internal power consumption data includes at least: total power consumption, three-phase current, voltage, active power, reactive power, power factor and daily load curve data. External data access unit 12 is used to access external data sources through an API interface. External data includes meteorological data and calendar information. The data fusion processing unit 13 is used to perform time alignment, missing value imputation and data standardization processing on internal electricity consumption data and external data based on timestamps, and generate multi-source time series datasets.

[0025] It has achieved comprehensive collection and fusion processing of multi-dimensional internal power consumption data and external environmental data of high-voltage users, forming a unified multi-source time series dataset. This provides a complete and consistent data foundation for subsequent mutation point detection and health assessment, and improves the accuracy and reliability of system data processing.

[0026] In this embodiment, the system also includes edge computing nodes deployed on the user side, and the edge computing nodes are used for: High-frequency local power consumption data is collected, with a sampling frequency higher than that of the internal power consumption data in data acquisition and fusion module 1; A lightweight mutation point detection algorithm is run locally to perform preliminary identification and alarm for transient faults; The preliminary identification results, alarm information, and data after feature extraction are uploaded to the mutation point detection engine 2.

[0027] It realizes the acquisition of high-frequency local power consumption data and lightweight mutation point detection, and can perform preliminary identification and alarm of instantaneous faults. The preliminary identification results, alarm information and feature extraction data are uploaded to the mutation point detection engine (2), which reduces the cloud processing pressure, improves the real-time performance and response speed of the system, and enhances the efficiency of local fault handling.

[0028] In this embodiment, the mutation point detection engine 2 specifically includes: The multi-algorithm detection unit 21 is used to run at least two different mutation point detection algorithms in parallel, and the algorithms are selected from at least two of CUSUM, PELT, Binary Segmentation, and Wavelet Transform. The integrated decision unit 22 is connected to the multi-algorithm detection unit 21 and is configured to perform fusion decision on the mutation point location and confidence level of each algorithm output, and determine the final mutation point set by weighted average method. The pattern recognition unit 23 is used to perform seasonal and trend decomposition on the multi-source time series dataset and to detect abrupt change points in the decomposed residual sequence to identify at least one of mean abrupt change, variance abrupt change, and trend reversal.

[0029] By employing multiple mutation point detection algorithms in parallel operation and integrated decision-making, combined with seasonality and trend decomposition, the accuracy and robustness of mutation point identification are significantly improved. It can effectively detect various mutation types such as mean mutation, variance mutation and trend reversal, and reduce the risk of false alarms and false negatives.

[0030] In this embodiment, when the integrated decision-making unit 22 performs fusion decision-making, the specific steps are as follows: Each mutation detection algorithm is assigned a dynamic weight, which is adaptively adjusted based on the algorithm's accuracy and recall metrics on historical data. Calculate the weighted number of votes for each candidate mutation point across all algorithms; When the weighted number of votes exceeds a preset threshold, and the spatial density of the point in different algorithm results meets the clustering requirements, it is determined as the final effective mutation point.

[0031] By dynamically assigning weights, calculating weighted votes, and verifying clustering requirements, the system achieves intelligent fusion of results from multiple mutation point detection algorithms, ensuring the reliability and effectiveness of the final mutation point set. This further optimizes the accuracy and stability of mutation point detection and enhances the scientific nature of the system's decision-making.

[0032] In this embodiment, the health assessment and root cause analysis module 3 specifically includes: The causal discovery unit 31 is used to extract multivariate data within the mutation time window after detecting a mutation point, and to use the causal discovery algorithm to analyze the causal relationship between the power consumption state mutation and external environmental data and internal power consumption data of other dimensions, and to generate a candidate root cause set. The knowledge graph query unit 32 stores the equipment files, historical operation and maintenance records and power grid topology connection relationships of high-voltage users. It is used to receive the candidate root cause set and combine it with the knowledge graph for reasoning and verification in order to locate the most likely faulty equipment and external factors. The health calculation unit 33 is used to calculate a quantitative health score based on the frequency, intensity, type of mutation points and the conclusions of root cause analysis. Mutations caused by internal equipment failures will be given a higher unhealthy weight.

[0033] It enables precise localization and reasoning of the root causes of mutation points, and calculates quantitative health scores based on mutation point characteristics, thereby enhancing the depth and objectivity of fault diagnosis and providing a comprehensive basis for health status assessment.

[0034] In this embodiment, the causal discovery unit 31 is specifically used for: At least one of the constraint-based PC algorithm, the fraction-based GES algorithm, and the nearest neighbor-based CCI algorithm is used to calculate the conditional independence between variables and construct a causal graph. In a cause-effect graph, identify the direct cause parent node and the indirect cause node that lead to a sudden change in the power consumption state; Output the physical variables corresponding to the parent node of the direct cause to the candidate root cause set.

[0035] It can accurately analyze the causal relationship between sudden changes in electricity consumption status and external environmental data, as well as internal electricity data from other dimensions, providing a scientific basis for root cause analysis and improving the accuracy and efficiency of causal inference.

[0036] In this embodiment, the mutation point detection engine 2 further integrates a deep learning model, which is an autoencoder, LSTM network, and Transformer architecture based on time-series data. The operation steps of the mutation point detection engine 2 are as follows: By inputting multi-source time-series datasets into a trained deep learning model, the probability of health anomalies and reconstruction errors of electricity consumption sequences can be directly obtained. The reconstructed error sequence and the anomaly probability sequence are used as new detection dimensions and input into the multi-algorithm detection unit 21 to detect mutation points, so as to realize a hybrid detection strategy of traditional algorithms and deep learning models.

[0037] A hybrid detection strategy combining traditional algorithms and deep learning models was implemented, which improved the sensitivity and adaptability of mutation point detection and enhanced the system's ability to recognize complex temporal patterns.

[0038] Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0039] During operation, the system integrates internal power consumption data from high-voltage users with external environmental data to form a unified, high-quality multi-source time-series dataset. Subsequently, the mutation point detection engine 2 employs a hybrid strategy combining various traditional algorithms and deep learning models to collaboratively analyze the power consumption time-series data and accurately identify and locate mutation points through an integrated decision-making mechanism. Upon detecting a mutation point, the health assessment and root cause analysis module 3 is activated. It first uses a causal discovery algorithm to infer the potential causes of the mutation, then combines this with a knowledge graph containing equipment files and grid topology for verification and precise location. Finally, it comprehensively analyzes the characteristics of the mutation point and the results of the root cause analysis to calculate a quantitative health score. The results of the entire process, including health status, mutation alarms, and root cause conclusions, are presented intuitively by the visualization warning and report generation module 4, which automatically generates an assessment report, thus achieving a fully intelligent, end-to-end assessment of the power consumption health status of high-voltage users, from data to decision.

[0040] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0041] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-voltage user power consumption state health assessment system based on time series mutation point detection, characterized in that, The system comprises: a data acquisition and fusion module (1) configured to acquire multi-dimensional time-series power consumption data and external environment data of a high-voltage user, and to perform cleaning, alignment and fusion processing on the data to form a unified multi-source time-series data set; a mutation point detection engine (2) connected to the data acquisition and fusion module (1) and configured to receive the multi-source time-series data set and to perform mutation point identification and positioning on the power consumption state by using an integrated mutation point detection algorithm; a health assessment and root cause analysis module (3) connected to the mutation point detection engine (2) and configured to calculate a health score of the user and to perform root cause inference based on the identified mutation point features and in combination with a power grid topology and a user profile knowledge graph; a visualized early warning and report generation module (4) connected to the health assessment and root cause analysis module (3) and configured to display the health assessment results, mutation point alarms and root cause analysis conclusions in a visualized form and to generate an automatic assessment report.

2. The high-voltage user power consumption state health assessment system based on time series mutation point detection according to claim 1, characterized in that: The data acquisition and fusion module (1) specifically comprises: an internal data acquisition unit (11) configured to acquire multi-dimensional internal power consumption data of a high-voltage user from a power consumption information acquisition system and a SCADA system, wherein the internal power consumption data at least includes total power consumption, three-phase current, voltage, active power, reactive power, power factor and daily load curve data; an external data access unit (12) configured to access external data sources through an API interface, wherein the external data includes weather data and calendar information; a data fusion processing unit (13) configured to perform time alignment, missing value filling and data standardization processing on the internal power consumption data and the external data based on a timestamp, thereby generating the multi-source time-series data set. 3.The high-voltage user power consumption state health assessment system based on time series mutation point detection of claim 1, wherein: The system further comprises an edge computing node deployed at the user side, and the edge computing node is configured to: acquire high-frequency local power consumption data with a sampling frequency higher than that of the internal power consumption data in the data acquisition and fusion module (1); locally run a lightweight mutation point detection algorithm to preliminarily identify and alarm transient faults; upload the preliminary identification results, alarm information and data after feature extraction to the mutation point detection engine (2).

4. The high voltage user power consumption state health assessment system based on time series mutation point detection according to claim 1, characterized in that: The mutation point detection engine (2) specifically comprises: a multi-algorithm detection unit (21) configured to run at least two different mutation point detection algorithms in parallel, wherein the algorithms are selected from at least two of CUSUM, PELT, Binary Segmentation and Wavelet Transform; an integrated decision unit (22) connected to the multi-algorithm detection unit (21) and configured to make a fusion decision on the mutation point positions and confidence levels output by each algorithm, and to determine a final set of mutation points by using a weighted average method; a pattern recognition unit (23) configured to perform seasonal and trend decomposition on the multi-source time-series data set, and to perform mutation point detection on the decomposed residual sequences to identify at least one of mean mutation, variance mutation and trend turning.

5. The high-voltage user power consumption state health assessment system based on time series mutation point detection according to claim 4, characterized in that: When the integrated decision unit (22) makes a fusion decision, the specific steps are: A dynamic weight is assigned to each mutation point detection algorithm, which is adaptively adjusted based on the accuracy and recall rate of the algorithm on historical data; The weighted vote of each candidate mutation point in all algorithms is calculated; When the weighted vote exceeds a preset threshold and the spatial density of the point in different algorithm results meets the clustering requirement, it is determined as the final effective mutation point. 6.The high-voltage user power consumption state health assessment system based on time series mutation point detection of claim 1, wherein: The health assessment and root cause analysis module (3) specifically includes: A causal discovery unit (31) is configured to extract multivariate data in a mutation time window after detecting a mutation point, analyze the causal relationship between the mutation of electricity consumption state and external environmental data and internal other dimension electricity data using a causal discovery algorithm, and generate a candidate root cause set; A knowledge graph query unit (32) stores the equipment archives, historical operation and maintenance records, and power grid topology connection relationships of high-voltage users, receives the candidate root cause set, and combines the knowledge graph to perform reasoning and verification to locate the most possible fault equipment and external factors; A health degree calculation unit (33) is configured to calculate a quantitative health degree score based on the frequency, intensity, and type of mutation points and the conclusion of root cause analysis, wherein a mutation caused by a device internal fault is given a higher unhealthy weight.

7. The high-voltage user power consumption state health assessment system based on time series mutation point detection according to claim 6, characterized in that: The causal discovery unit (31) is specifically configured to: At least one of a constraint-based PC algorithm, a score-based GES algorithm, and a nearest neighbor-based CCI algorithm is used to calculate the conditional independence between variables and construct a causal graph; In the causal graph, identify the direct cause parent node and the indirect cause node that cause the mutation of the electricity consumption state; The physical variable corresponding to the direct cause parent node is output to the candidate root cause set. 8.The high-voltage user power consumption state health assessment system based on time series mutation point detection of claim 1, wherein: The mutation point detection engine (2) further integrates a deep learning model, and the deep learning model is a self-encoder, an LSTM network, and a Transformer architecture based on time series data. The running steps of the mutation point detection engine (2) are: The multi-source time series data set is input into the trained deep learning model to directly obtain the health degree anomaly probability and reconstruction error of the electricity consumption sequence; The reconstruction error sequence and the anomaly probability sequence are input into the multi-algorithm detection unit (21) for mutation point detection as new detection dimensions, so as to realize a hybrid detection strategy of traditional algorithms and deep learning models.

9. A high-voltage user power consumption state health assessment method based on time series mutation point detection, characterized in that, The high-voltage user electricity consumption state health assessment system based on time series mutation point detection according to any one of claims 1-8; comprising the following steps: Step 1: data acquisition and fusion: obtaining multi-dimensional time series electricity consumption data and external environment data of high-voltage users, and performing cleaning, alignment and fusion processing on the multi-dimensional time series electricity consumption data and external environment data to form a unified multi-source time series data set; Step 2: mutation point detection: using an integrated mutation point detection algorithm to identify and locate the electricity consumption state mutation point of the multi-source time series data set; Step 3: Health assessment and root cause analysis: According to the identified mutation point characteristics, combined with the power grid topology and user profile knowledge graph, the health score of the high-voltage user's power consumption state is calculated, and the root cause of the mutation point is inferred; Step 4: Visualization display and report generation: The health score, mutation point alarm information and root cause analysis conclusion are displayed in a visual form, and an automatic evaluation report is generated.

Images