Ammeter connector health state management method and storage medium

Abstract

The invention discloses an ammeter connector health state management method and a storage medium. The method comprises the following steps: constructing and binding an ammeter connector digital twin of a power grid topological relation; multi-source operation data such as temperature of a key connection point, electrical data flowing through a loop and environmental data are collected in real time through a sensor network; inputting the data into a health state evaluation model to output a health state grade, fusing preliminary analysis based on a preset rule base and deep analysis based on a data-driven model in the evaluation process of the model, and generating a final grade by adopting a dynamic weight fusion mechanism according to the uncertainty of the results of the preliminary analysis and the deep analysis; generating and triggering a differentiated maintenance decision instruction according to the grade; and optimizing and updating the evaluation rule base and / or the data driving model based on result data fed back by field maintenance. According to the invention, comprehensive perception, accurate evaluation, intelligent decision and closed-loop optimization of the health state of the electric meter connector are realized, and the intelligent level of state management and the reliability of power supply are improved.

Description

Technical Field

[0001] This invention relates to the field of power system condition monitoring technology, specifically to a method and storage medium for managing the health status of electricity meter connectors. Background Technology

[0002] As a key node in electricity metering and transmission, the reliability of electricity meter connectors directly affects metering accuracy and electricity safety. With the continuous advancement of smart grid construction and the surge in the number of electricity meters installed, the operating environment of connectors is becoming increasingly complex. Poor contact, corrosion, oxidation and aging can lead to progressive hidden dangers such as abnormal heating and increased resistance. If not dealt with in time, these problems may cause serious accidents such as metering deviations, power outages, or even fires, directly threatening the stable operation of the power system and the safety of users' electricity consumption.

[0003] Currently, the status management of electricity meter connectors still relies primarily on traditional manual inspections, supplemented by a few simple online monitoring methods. This existing management approach has several insurmountable drawbacks: First, the monitoring data is limited, often focusing only on local temperature or individual electrical parameters, lacking integrated analysis of multi-source information such as the temperature of key connection points, electrical data flowing through circuits, and the ambient temperature and humidity, making it difficult to comprehensively reflect the true health status of the connectors. Second, there is a lack of digital modeling methods; the correlation between connectors and the power grid topology has not been established, resulting in isolated status assessments and an inability to quickly locate the specific location and impact range of faults within the power grid, making fault location difficult. Third, the health assessment methods are rudimentary; existing mechanistic models lack accuracy under complex operating conditions, while... The driving model is limited by the scarcity of samples and poor interpretability, and fails to effectively integrate the advantages of both. It relies heavily on fixed thresholds for judgment, cannot identify trend-based degradation characteristics, and has low assessment accuracy. In addition, the management process lacks a closed-loop optimization mechanism, and the experience data from on-site maintenance feedback is not effectively used to optimize the assessment model, resulting in rigid assessment standards that are difficult to adapt to the usage needs of different regions and operating conditions. Finally, due to insufficient assessment accuracy, maintenance decisions are often delayed or crude, and it is impossible to implement preventive differentiated operation and maintenance based on the actual health status of connectors, which increases operation and maintenance costs and makes it difficult to guarantee power supply reliability.

[0004] In summary, existing technologies are insufficient to meet the needs of smart grids for accurate, intelligent, and preventative management of massive numbers of meter connectors. Therefore, there is an urgent need for an intelligent management method for the health status of meter connectors that can integrate multi-source data, construct digital associations, coordinate mechanisms and data models, and possess closed-loop optimization capabilities. Summary of the Invention

[0005] The purpose of this invention is to provide a method and storage medium for managing the health status of electricity meter connectors, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for managing the health status of an electricity meter connector, comprising: Construct and run a digital twin corresponding to the meter connector, wherein the digital twin is bound to the static attribute information of the connector and its association in the power grid topology; The meter connectors are collected in real time using a sensor network. The multi-source operating data includes at least the temperature data of their key connection points, the electrical data of the circuits flowing through them, and the environmental data of the surrounding environment. The multi-source operating data is input into the health status assessment model to output the health status level of the meter connector. The health status level includes at least normal, attention, warning and alarm. The output of the health status level includes: The multi-source operational data is preliminarily analyzed based on a preset evaluation rule base; The multi-source operational data is analyzed in depth based on a data-driven model. Based on the uncertainty of the preliminary analysis results and the in-depth analysis results, a dynamic weight fusion mechanism is used to fuse the two results to generate the health status level; Based on the health status level, generate and trigger corresponding maintenance decision instructions; Receive the results data from on-site maintenance, and optimize and update the evaluation rule base and / or the data-driven model based on the results data.

[0007] Preferably, the construction and operation of the digital twin corresponding to the meter connector includes: Assign a unique digital identity to each physical meter connector or its integrated unit; The digital identity identifier is associated and bound with the corresponding static attribute information to form a static attribute file. The static attribute information includes at least asset information, installation location information, and branch line information. The static attribute files are stored on the management platform, forming the basic data layer of the digital twin.

[0008] Preferably, the real-time acquisition of multi-source operating data of the meter connector via a sensor network includes: Real-time temperature data of the part is collected by a temperature sensing device installed at the key conductive connection part of the meter connector. The electrical data flowing through the circuit is obtained by an energy metering device connected to an electricity meter connector. The electrical data includes at least current, voltage and power data. Environmental data is collected by an environmental sensing device deployed in the installation environment of the electricity meter connector. The environmental data includes at least ambient temperature and ambient humidity data.

[0009] Preferably, the preset evaluation rule base includes multi-level threshold judgment rules based on temperature parameters, including: Level 1 rule: When the temperature difference between the critical connection point and the ambient temperature exceeds a first set threshold and the difference remains for a first preset duration, the status is determined to be "attention". Second-level rule: When the difference exceeds the second set threshold, the second set threshold is greater than the first set threshold, and the maximum temperature difference between different phase connection points in the same connector unit exceeds the third set threshold, the status is determined to be a warning. Level 3 rule: When the temperature value of a critical connection point exceeds the safety limit, or its temperature rise rate exceeds the fourth set threshold, the status is determined to be an alarm.

[0010] Preferably, the in-depth analysis of the multi-source operational data based on the data-driven model includes: The data-driven model is trained based on historical normal operation data to learn the operating characteristics of the meter connector under health conditions; Based on real-time collected electrical and temperature data, the contact resistance characteristic value or its changing trend of the meter connector is estimated. The contact resistance characteristic value or its changing trend is compared with the baseline trend predicted by the data-driven model. When the deviation exceeds the set tolerance and the deviation is maintained for a second preset time, the status is determined to be either "attention" or "warning".

[0011] Preferably, the method of fusing the two results using a dynamic weight fusion mechanism includes: Based on the uncertainty assessment results of the preliminary analysis results and the in-depth analysis results, initial fusion weights are assigned to the two; The initial fusion weights are dynamically adjusted using real-time multi-source operational data as feedback. An optimization algorithm is used to perform global optimization on the adjusted weight combination to determine the final fusion weight, and a health status level is generated based on the final fusion weight.

[0012] Preferably, the step of generating and triggering corresponding maintenance decision instructions based on the health status level includes: When the health status level is "concerned", a planned inspection task instruction is generated. When the health status level is warning, a work order for on-site verification within a specified period is generated and pushed to the maintenance terminal; When the health status level is alarm, an emergency response work order is generated, and a related early warning notification is sent to the power distribution management system simultaneously.

[0013] Preferably, the step of receiving the result data from on-site maintenance feedback and optimizing and updating based on the result data includes: Receive and record the on-site handling results and retest data reported by maintenance personnel; The on-site processing results and retest data are correlated with the multi-source operational data corresponding to the time the maintenance decision command is triggered to form case data; Based on accumulated case data, the threshold parameters in the evaluation rule base are adjusted, and / or the data-driven model is retrained.

[0014] Preferably, the method further includes: On the monitoring interface, based on the power grid topology map, the real-time health status level of each meter connector digital twin is displayed in a differentiated visual form. It provides a historical data query function, and the historical data includes at least temperature change curves, contact resistance trend curves, and status level change records.

[0015] This application also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for managing the health status of the meter connector.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: by constructing a digital twin to bind static attributes and power grid correlation, multi-source operation data is collected in real time, and a health status level is output based on the evaluation model. The results of preliminary analysis and in-depth analysis are integrated to generate maintenance decision instructions and optimize and update the model. This solves the problems of single data, low evaluation accuracy and lack of closed-loop optimization in existing technologies. It has the advantages of being able to comprehensively and accurately evaluate the health status of meter connectors, realize intelligent maintenance decisions, and improve system adaptability and power supply reliability through closed-loop optimization. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the health status management method according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the process of constructing and running a digital twin according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the multi-level threshold judgment rule process according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1-3A method for managing the health status of electricity meter connectors is disclosed. This method constructs and runs a digital twin of the electricity meter connector, binding its static attribute information and its correlation within the power grid topology. Multi-source operational data of the meter connector is collected in real time via a sensor network. This data includes at least temperature data of its key connection points, electrical data flowing through the circuits, and environmental data of the surrounding environment. The multi-source operational data is input into a health status assessment model to output a health status level for the electricity meter connector, which includes at least four levels: normal, watch out, warning, and alarm. The output health status level includes preliminary analysis of the multi-source operational data based on a preset assessment rule base. In-depth analysis of the multi-source operational data is performed based on a data-driven model. Due to the uncertainty of the preliminary and in-depth analysis results, a dynamic weighted fusion mechanism is used to fuse the results to generate a health status level. Based on the health status level, corresponding maintenance decision instructions are generated and triggered. Feedback data from on-site maintenance is received, and the assessment rule base and / or data-driven model are optimized and updated based on this feedback data.

[0020] For ease of understanding, the following explains some key terms in this embodiment: Digital twin: refers to a virtual mapping of a physical entity in digital space, enabling state monitoring, behavior simulation, fault diagnosis, and prediction of the physical entity through real-time data connections. In this method, the digital twin is used to characterize the real-time operating status and historical evolution of the meter connector.

[0021] Sensor networks are distributed network systems composed of multiple sensor nodes used to collect various data from the physical world in real time. In this method, the sensor network is responsible for acquiring operational data from the meter connector and its environment.

[0022] Multi-source operational data: refers to a collection of various data points collected from different types of sensors or data sources that reflect the operating status of the meter connector. These data may include temperature, current, voltage, humidity, etc.

[0023] Health status assessment model: refers to an algorithm or system used to comprehensively analyze multi-source operational data and output the current health status of the meter connector. This model aims to quantify the health level of the connector and classify it into different levels.

[0024] Evaluation rule base: refers to a set of pre-defined logical judgment rules or thresholds used to perform preliminary, experience-based or mechanism-based analysis of operational data.

[0025] Data-driven models refer to predictive or classification models built by learning patterns and regularities from historical data, such as machine learning models or deep learning models. These models are able to discover potential correlations from complex data.

[0026] Dynamic weight fusion mechanism: This refers to a strategy that adaptively adjusts the weights of different evaluation methods, such as rule-based analysis and data-driven model analysis, based on the uncertainty of real-time data or evaluation results. This mechanism aims to improve the accuracy and robustness of the final evaluation results.

[0027] Health status level: refers to the classification or grading of the current health status of the meter connector, which usually includes different levels of risk warnings such as normal, attention, warning and alarm.

[0028] Maintenance decision instructions: These are tasks or notices that are automatically generated and issued based on the health status level of the meter connectors, guiding on-site maintenance personnel to perform corresponding operations.

[0029] This embodiment provides a method for managing the health status of electricity meter connectors.

[0030] First, a digital twin corresponding to the meter connector is constructed and run. This digital twin binds the static attribute information of the connector and its association with the power grid topology. Specifically, a data record can be created in the management system for each physical meter connector. This record contains basic information such as the connector model, manufacturer, and installation date, and the line number or substation name it is connected to is manually entered to establish its initial association with the power grid.

[0031] Secondly, multi-source operational data from the meter connectors is collected in real time via a sensor network. This multi-source operational data includes at least temperature data at key connection points, electrical data flowing through the circuit, and environmental data of the surrounding environment. For example, a general-purpose temperature sensor can be placed near the meter connector to measure its surface temperature; electrical data can be manually read or recorded using ammeters and voltmeters connected to the circuit; and an ambient temperature and humidity meter can be placed in the connector installation area to obtain environmental data. This data is transmitted to the data processing center via wired or wireless means.

[0032] Furthermore, multi-source operational data is input into the health status assessment model to output the health status level of the meter connector. The health status levels include at least Normal, Attention, Warning, and Alarm. Outputting the health status level includes the following steps: The system performs preliminary analysis of multi-source operational data based on a pre-defined evaluation rule base. This rule base can include a series of simple threshold judgment rules. For example, when the collected connector temperature exceeds a fixed upper limit, the system determines its status as "concerned"; when the current value exceeds a preset safety range, it is determined as a warning. These rules are set based on industry experience or recommended values ​​in equipment manuals.

[0033] In-depth analysis of multi-source operational data is performed based on a data-driven model. This model can be a simple statistical model; for example, by analyzing historical normal operation data, a linear relationship between connector temperature and ambient temperature can be established. When there is a significant deviation between the real-time collected temperature data and the predicted value based on this linear relationship, the model outputs an anomaly index.

[0034] Based on the uncertainty of the preliminary and in-depth analysis results, a dynamic weight fusion mechanism is used to fuse the two results to generate a health status level. For example, a fixed weight can be preset for the preliminary and in-depth analysis results, such as 50% each. When the two analysis results are inconsistent, a decision can be made according to preset priority rules. For example, if the preliminary analysis result is an alarm, the final level will prioritize the alarm. Specifically, the uncertainty can be quantified by a confidence score. For the preliminary analysis, the confidence score is evaluated based on the degree of matching between the rule triggering conditions and historical cases; for the in-depth analysis, it can be evaluated based on the probability distribution of the model output or the difference between the distribution of real-time data and training data. The initial weight allocation can use the normalized reciprocal method to give higher weights to results with lower uncertainty. The optimization algorithm can be particle swarm optimization or Bayesian optimization, with the objective function being to minimize the error between the fused evaluation result and the actual state in historical cases, and the final fusion weights are determined through iterative optimization.

[0035] Based on the health status level, the system generates and triggers corresponding maintenance decision instructions. For example, when the health status level is "Concern," the system can generate a prompt message in the background, suggesting that maintenance personnel perform additional checks on the connector during the next routine inspection; when the level is "Warning," the system can send an SMS notification to the designated maintenance personnel, requiring them to conduct on-site verification within a specified time; when the level is "Alarm," the system sends an emergency notification, requiring immediate action.

[0036] Finally, the system receives feedback data from on-site maintenance and optimizes and updates the evaluation rule base and / or data-driven model based on this data. After completing on-site procedures, maintenance personnel can manually input the processing results and simple retest data into the management system. The system periodically conducts manual reviews of this feedback data and, based on the review results, manually adjusts the threshold parameters in the evaluation rule base or retrains the data-driven model.

[0037] This method constructs a digital twin to link connectors with the power grid topology, integrates multi-source operational data for comprehensive perception, and combines a rule base and data-driven model for intelligent assessment. The dynamically fused results output an accurate health status level. This enables early identification and differentiated maintenance of potential hazards in meter connectors, effectively avoiding problems such as data isolation, inaccurate assessments, and delayed decision-making in traditional management models, thereby improving the reliability of power grid operation and the level of electricity safety.

[0038] In some of the embodiments described above in this application, a digital twin corresponding to a meter connector is proposed to bind static attribute information and its association with the power grid topology, thereby supporting health status assessment. However, in its implementation, the lack of assigning a unique digital identity to each physical meter connector may lead to identity confusion, making it impossible to accurately identify and locate specific meter connectors; failure to associate and bind the identity with attribute information to form a file may result in scattered and incomplete static information, affecting data integration; and failure to store it on a management platform to form a basic data layer may leave the digital twin without systematic basic support, reducing the reliability and efficiency of subsequent status assessments.

[0039] In response, this application further proposes to construct and operate a digital twin corresponding to an electricity meter connector, including: assigning a unique digital identity to each physical electricity meter connector or its integrated unit; associating and binding the digital identity with the corresponding static attribute information to form a static attribute file, wherein the static attribute information includes at least asset information, installation location information and branch line information; and storing the static attribute file in a management platform to constitute the basic data layer of the digital twin.

[0040] Specifically, each physical meter connector or its integrated unit is assigned a unique digital identity. This aims to ensure that each physical meter connector has a unique digital representation in the digital twin system, thereby avoiding identity confusion and enabling accurate identification and location of individual meter connectors. For example, a Globally Unique Identifier (GUID) or a Universally Unique Identifier (UUID) mechanism can be used to automatically generate and write the identifier into the internal memory or external tag of the meter connector during production or installation. Alternatively, in conjunction with power grid asset coding rules, each meter connector or its integrated unit can be assigned a code with a hierarchical structure and business meaning, such as including area code, substation code, line code, tower code, and equipment serial number, ensuring its uniqueness throughout the entire power grid system.

[0041] Subsequently, the digital identity is associated and bound with the corresponding static attribute information to form a static attribute file. This step aims to establish a logical connection between the digital identity and the invariant characteristics of the physical entity, constructing a comprehensive and structured data record. The static attribute information includes at least asset information, installation location information, and branch line information, which are crucial for understanding the background, operating environment, and role of the meter connector in the power grid. For example, a master table can be created in the management platform's database, using the digital identity as the primary key and containing asset information such as model, manufacturer, commissioning date, installation location information, and branch line information, achieving one-to-one association storage. Alternatively, an object-oriented data model can be used, with the digital identity of each meter connector as the object ID and its static attribute information as the object's attribute fields. This information can be persisted to a relational database or NoSQL database using Object Relational Mapping (ORM) technology, forming a structured static attribute file.

[0042] Finally, the static attribute files are stored on the management platform, forming the foundational data layer of the digital twin. This step aims to centrally manage and organize all static information, enabling efficient access and utilization by the various functional modules of the digital twin, laying a solid foundation for subsequent dynamic data analysis and health status assessment. For example, the management platform can be a cloud-based distributed database system, such as Hadoop or Cassandra, used to store massive amounts of static attribute files and provide data access services through API interfaces, ensuring high availability and scalability of the data. Alternatively, an Enterprise Asset Management System (EAM) or Geographic Information System (GIS) can be used as part of the management platform, integrating static attribute files with the power grid topology layer to achieve visualized data management and spatial querying, thereby forming the foundational data layer of the digital twin.

[0043] By assigning a unique digital identity to each physical meter connector or its integrated unit, the problem of meter connector identity confusion is effectively solved. This enables the system to accurately identify and locate each specific meter connector, providing a clear foundation for subsequent health status assessments and fault location. Furthermore, the digital identity is associated with corresponding static attribute information to form a structured static attribute file, which includes asset information, installation location information, and branch line information. This not only ensures the integrity and consistency of static information but also establishes a strong association between the meter connector and the power grid topology, allowing the digital twin to accurately reflect the spatial location and network relationships of the physical entity. Finally, the static attribute file is stored on a management platform, forming a solid foundational data layer for the digital twin. This ensures centralized management, efficient access, and reliability of all static information, providing indispensable data support for the stable operation of the digital twin and subsequent health status assessment models. This systematic construction approach gives the digital twin of the meter connector a high degree of accuracy and traceability, thereby improving the reliability of health status assessments and the accuracy of maintenance decisions.

[0044] In some of the solutions mentioned above in this application, multi-source operating data of the meter connectors are collected in real time through sensor networks to input into the health status assessment model. However, in this process, there are shortcomings in how to specifically deploy sensors to fully cover the temperature of key connection points, electrical data flowing through the circuit, and environmental data. This results in incomplete data collection, which cannot accurately reflect the true operating status of the connectors and affects the accuracy of subsequent health status assessments.

[0045] In response, this application further proposes to collect multi-source operating data of the meter connector in real time through a sensor network, specifically including the following aspects: First, a temperature sensing device is installed at the critical conductive connection point of the meter connector to collect real-time temperature data. This temperature sensing device is a device capable of sensing and measuring temperature. Its function is to directly monitor the temperature of the critical parts of the meter connector most prone to poor contact or abnormal overheating, thereby avoiding the omission of potential faults due to incomplete data collection. Specifically, the temperature sensing device can be a thermocouple sensor. By directly contacting or tightly attaching the measuring end of the thermocouple to the critical conductive connection point of the meter connector, the temperature signal is converted into an electrical signal for acquisition using the principle of thermoelectric electromotive force. Alternatively, the temperature sensing device can also be a thermistor sensor, installed near the critical conductive connection point. Utilizing the characteristic that its resistance changes with temperature, real-time temperature data is obtained by measuring the resistance change.

[0046] Secondly, a power metering device connected to the meter connector acquires real-time electrical data flowing through the circuit. This electrical data includes at least current, voltage, and power data. The power metering device is used to measure and record power consumption or transmission, and typically includes current transformers, voltage transformers, and metering chips. Its function is to provide core electrical operating parameters of the circuit flowing through the meter connector, helping to identify resistance changes or potential electrical faults caused by connector deterioration. Specifically, the metering module built into the smart meter can be used. This module typically integrates high-precision current and voltage sampling circuits and a power calculation unit, and can directly read real-time current, voltage, and power data through its communication interface. Alternatively, independent current and voltage sensors can be connected in series or parallel in the meter connector circuit, and the output signals of these sensors can be connected to the data acquisition unit for digital processing and power calculation.

[0047] Furthermore, environmental data is collected by an environmental sensing device deployed in the installation environment of the electricity meter connector. This environmental data includes at least ambient temperature and humidity data. The environmental sensing device is used to monitor surrounding environmental parameters, such as temperature and humidity sensors. Its function is to account for the impact of external environmental factors on the performance of the electricity meter connector, making data collection more comprehensive and avoiding evaluation biases caused by environmental interference. Specifically, an integrated temperature and humidity sensor module can be used, installed inside the meter box or near the connector, directly outputting ambient temperature and relative humidity data through a digital interface. Alternatively, separate temperature and humidity sensors can be deployed, transmitting data to the data acquisition unit via analog or digital interfaces.

[0048] Through the above technical solutions, this application effectively solves the problems of incomplete data collection and inaccurate reflection of the actual operating status of connectors in existing technologies. By installing temperature sensors at key conductive connection points of the meter connector, the parts most prone to abnormal heating can be directly and accurately monitored, and potential overheating risks can be detected in a timely manner. By acquiring real-time electrical data flowing through the circuit through an energy metering device connected to the meter connector, core operating parameters such as current, voltage, and power can be comprehensively grasped, providing a basis for judging changes in the electrical performance of the connector. At the same time, by deploying environmental sensors to collect ambient temperature and humidity data, the impact of the external environment on the connector performance can be fully considered, avoiding misjudgments caused by environmental factors. This multi-source, comprehensive, and targeted operating data provides high-quality input for subsequent health status assessment models, enabling the assessment models to more accurately identify the actual health status of connectors, including progressive degradation trends, thereby improving the accuracy and reliability of health status assessment, laying a solid foundation for timely generation and triggering of maintenance decision commands, and effectively ensuring the stability and safety of power grid operation.

[0049] In some of the embodiments described above in this application, a preset evaluation rule base is proposed for preliminary analysis of multi-source operating data. However, in its implementation, the existing rule base may only rely on a single threshold judgment mechanism and cannot perform fine-grained classification evaluation based on temperature parameters such as temperature difference, temperature rise rate, and temperature difference. This results in inaccurate health status level determination, difficulty in distinguishing different levels of temperature anomaly risks, and thus inability to effectively identify progressive degradation characteristics, affecting the timeliness and pertinence of maintenance decisions.

[0050] In response, this application further proposes a preset evaluation rule base containing multi-level threshold judgment rules based on temperature parameters, including: Level 1 rule: when the difference between the temperature of a critical connection point and the ambient temperature exceeds a first preset threshold, and the difference is maintained for a first preset duration, the judgment status is "attention"; Level 2 rule: when the difference exceeds a second preset threshold, the second preset threshold is greater than the first preset threshold, and the maximum temperature difference between different phase connection points within the same connector unit exceeds a third preset threshold, the judgment status is "warning"; Level 3 rule: when the temperature value of a critical connection point exceeds a safety limit, or its temperature rise rate exceeds a fourth preset threshold, the judgment status is "alarm".

[0051] The preset evaluation rule base is a knowledge set used to store and execute a series of predefined logical conditions and judgment criteria. Its function is to quickly and initially filter and classify multi-source operational data collected in real time to identify potential anomalies. This rule base can be implemented in various ways. For example, it can be a rule table based on a relational database, where each row records a rule condition and its corresponding health status level; or it can be a condition judgment module implemented in a programming language, including Python and Java, executing rule logic through a series of IF-THEN statements. By introducing multi-level threshold judgment rules based on temperature parameters, this rule base can overcome the limitations of single threshold judgments and achieve refined hierarchical evaluation of health status.

[0052] The first-level rule aims to identify early, minor abnormal heating of the meter connector. When the difference between the temperature of a key connection point, collected in real-time by the sensor network, and the ambient temperature continuously exceeds a preset first threshold, and this over-limit state persists for a first preset duration, the system classifies the health status of the meter connector as "of concern." For example, the system can continuously monitor the temperature difference; once it exceeds the first threshold, a timer is started. If the temperature difference returns to normal before the timer reaches the first preset duration, the "of concern" status is not triggered; otherwise, it is. Another implementation is that the system can calculate the frequency or cumulative duration of the temperature difference exceeding the first threshold over a period of time, and classify it as "of concern" when specific conditions are met. This mechanism can effectively filter out instantaneous or occasional temperature fluctuations, avoid false alarms, and thus more accurately capture gradual deterioration trends.

[0053] The second-level rule is used to identify more significant anomalies that may indicate local imbalances. When the temperature difference between a critical connection point and the ambient temperature exceeds not only the first set threshold but also the second set threshold, and simultaneously, within the same meter connector unit, the maximum temperature difference between different phase connection points also exceeds the third set threshold, the system determines the health status as "warning." For example, the system can monitor the temperature differences of multiple critical connection points in parallel and calculate the difference between the maximum and minimum temperature differences for each phase within the same unit. A warning state is triggered when both conditions—temperature difference exceeding the second set threshold and the maximum inter-phase temperature difference exceeding the third set threshold—are simultaneously met. This rule can be implemented using a complex logic judgment module capable of processing and comparing multiple temperature data streams simultaneously. By combining absolute and relative temperature differences, this rule can more comprehensively assess the health status of the connector, and is particularly suitable for detecting localized overheating or unbalanced heating caused by poor contact, loose bolts, etc.

[0054] The third-level rule addresses emergency anomalies that could lead to serious consequences. When the temperature of a critical connection point directly exceeds a preset safety limit, or its rate of temperature increase exceeds a fourth preset threshold, the system determines the health status as an alarm. For example, the system can continuously compare the real-time collected temperature of the critical connection point with the preset safety limit; once exceeded, an alarm is triggered. Simultaneously, the system can also perform differential or regression analysis on temperature data at continuous time points to calculate the rate of temperature increase and compare it with the fourth preset threshold. Once either condition is met, an alarm is immediately triggered. Implementing this rule requires high-frequency temperature data acquisition and rapid computational response capabilities to ensure timely alerts in cases of extreme high temperatures or rapid temperature increases, providing valuable time for emergency response.

[0055] Through the above technical solution, this application effectively solves the problem that existing assessment rule bases cannot provide refined and graded assessments of temperature-related health status in preliminary analysis. By introducing multi-level threshold judgment rules based on temperature parameters, a refined and graded assessment of the health status of meter connectors is achieved. The first-level rule monitors the temperature difference between key connection points and the ambient temperature and its duration, enabling timely detection of slight, gradual abnormal heating, avoiding misjudgments caused by instantaneous fluctuations, and thus classifying the health status as concerning, providing a basis for early intervention. The second-level rule, based on a higher temperature difference, further incorporates the maximum temperature difference between different phase connection points within the same connector unit. This allows the system to identify more serious degradation that may be caused by local imbalances or poor contact, thus classifying the health status as a warning and prompting a time-limited inspection. The third-level rule directly addresses extreme high temperatures or rapid temperature rises. By judging whether the temperature of key connection points exceeds safety limits or whether its rate of increase is too fast, it can quickly identify emergency risks that may lead to serious accidents and classify them as alarms, ensuring rapid response and emergency handling in critical situations. This multi-level judgment mechanism comprehensively considers multiple dimensions such as temperature difference, relative temperature difference, absolute temperature value, and rate of change, making the preliminary analysis results more accurate and comprehensive. It can effectively distinguish different levels of health risks, provide more precise guidance for subsequent maintenance decisions, and improve the intelligence and preventiveness of meter connector health status management.

[0056] In some of the solutions mentioned above in this application, a data-driven model is proposed to perform in-depth analysis of multi-source operating data to evaluate the health status of meter connectors. However, in this process, the data-driven model may be unable to accurately learn the operating characteristics under the health status due to the scarcity of samples, and it is difficult to estimate the characteristic value of contact resistance or its changing trend, resulting in inaccurate in-depth analysis results and inability to effectively identify progressive degradation characteristics, thereby affecting the reliability of health status determination.

[0057] To address this, this application further proposes a method for in-depth analysis of the multi-source operational data based on a data-driven model. Specifically, this includes training the data-driven model based on historical normal operation data. As a preferred implementation, the data-driven model can employ a Long Short-Term Memory (LSTM) network. The input is a normalized sequence of multi-source operational data, including temperature, current, voltage, ambient temperature and humidity, and their statistical characteristics; the output is a probability distribution of health status levels or a contact resistance regression value. The model uses a cross-entropy or mean squared error loss function and is trained through backpropagation of errors expanded over time steps. Contact resistance estimation can be based on the electrothermal coupling principle: power is calculated based on real-time electrical data, and combined with temperature rise, ambient temperature, and thermal resistance parameters, the heating power is inferred from the thermal balance equation, thereby estimating the contact resistance. The estimated value is compared with the baseline trend predicted by the model. The deviation is calculated as a percentage of relative error and combined with the duration to determine the status, so as to learn the operating characteristics of the meter connector in a healthy state. Based on the real-time collected electrical and temperature data, the contact resistance characteristic value or its changing trend of the meter connector is estimated. The contact resistance characteristic value or its changing trend is compared with the baseline trend predicted by the data-driven model. When the deviation exceeds the set tolerance and the deviation status is maintained for a second preset time, the status is determined to be "attention" or "warning".

[0058] The data-driven model is trained based on historical normal operation data to learn the operational characteristics of the meter connector under healthy conditions. This historical normal operation data refers to a long-term set of operational data for the meter connector under fault-free, stable performance, and compliance with design specifications. This data forms the basis for the model to learn normal behavior patterns. For example, it can be constructed by continuously collecting multi-source operational data, such as temperature, electrical parameters, and environmental data, for a period of time during the initial operation of the meter connector or after a comprehensive overhaul confirming no faults. Alternatively, it can be obtained by filtering and organizing historical operational records of healthy meter connectors from the same batch, model, and operating environment. The data-driven model is a mathematical model that learns patterns and rules from data to make predictions, classifications, or decisions. For example, it can be a model based on machine learning algorithms, such as Support Vector Machine (SVM), Random Forest, or Gradient Boosting Tree (GBDT), used to identify complex relationships in multidimensional data; or it can be a deep learning model, such as Recurrent Neural Network (RNN) or Long Short-Term Memory (LSTM), particularly suitable for processing time-series data and capturing dynamic changes in operational characteristics. The aforementioned learning of the operating characteristics of the meter connector under healthy conditions refers to the data-driven model establishing an internal representation that describes the interrelationships, variation patterns, and fluctuation ranges of various parameters of the meter connector under healthy conditions, such as temperature, current, voltage, and contact resistance, by analyzing historical normal operating data. For example, the model can learn the normal temperature rise relationship between the critical connection point temperature and the ambient temperature under different loads and ambient temperatures, as well as the normal fluctuation range of electrical parameters; or, the model can also learn the baseline value of contact resistance under healthy conditions and its small, predictable variation trends with time, load, environment, and other factors.

[0059] Furthermore, based on real-time collected electrical and temperature data, the characteristic value of the contact resistance of the meter connector or its changing trend is estimated. The real-time collected electrical and temperature data refers to electrical parameters such as current, voltage, and power, as well as data such as the temperature of key connection points and ambient temperature, continuously acquired through a sensor network and reflecting the current operating status of the meter connector. For example, electrical data can be directly acquired through an energy metering device connected to the meter connector, and temperature data can be acquired through temperature sensors installed at key conductive connection points; alternatively, this data can be transmitted in real-time to the management platform through wireless sensor networks (WSN) or Internet of Things (IoT) devices. Estimating the characteristic value of the contact resistance of the meter connector or its changing trend refers to using known physical laws and / or through data-driven models, combined with the real-time collected electrical and temperature data, to deduce the current value of the internal contact resistance of the meter connector or its variation over time. Contact resistance is a key indicator reflecting the connection quality of the connector; its increase usually indicates deterioration. For example, the contact resistance can be directly calculated by measuring the voltage drop and current flowing through the connector and applying Ohm's law; alternatively, a thermo-electric coupling model can be established, and the contact resistance value causing the temperature rise can be calculated using temperature and electrical data. Furthermore, by performing time series analysis on continuously estimated contact resistance values, the trend of its increase, decrease, or stabilization can be obtained.

[0060] Based on this, the contact resistance characteristic value or its changing trend is compared with the baseline trend predicted by the data-driven model. When the deviation exceeds a set tolerance and the deviation persists for a second preset duration, the status is determined to be "attention" or "warning". The baseline trend predicted by the data-driven model refers to the theoretical value of the contact resistance under healthy conditions or its normal trajectory over time, which the data-driven model can predict based on the current operating conditions after learning the operating characteristics under healthy conditions. For example, the model can output a prediction range, representing the reasonable fluctuation range of the contact resistance under healthy conditions; or, the model can predict the expected value of the contact resistance or its normal change curve over a future period under specific operating conditions. The comparison refers to the quantitative comparison of the contact resistance characteristic value or its changing trend estimated in real time with the baseline trend predicted by the data-driven model. For example, the absolute difference or relative percentage difference between the real-time value and the predicted baseline value can be calculated; or, statistical methods, such as calculating the mean square error or correlation coefficient, can be used to measure the degree of deviation between the two. The deviation refers to the degree of difference between the real-time estimated value and the model's predicted baseline trend. For example, it could be the difference between the real-time contact resistance value and the reference value; or it could be the difference between the slope of the real-time contact resistance change trend and the slope of the reference trend. The set tolerance refers to the maximum acceptable range of difference between the real-time estimated value and the reference trend. For example, it could be a fixed value, such as the contact resistance deviating from the reference value by more than 0.1 milliohms; or it could be a dynamic percentage, such as a deviation of 10% from the reference value. The second preset duration refers to the length of time the deviation continues to exceed the set tolerance. For example, it could be from several minutes to several hours, such as a continuous deviation of 15 minutes; or it could be set based on historical experience or expert knowledge, such as a continuous deviation for 3 consecutive sampling periods. Determining the status as "attention" or "warning" means raising the health status level of the meter connector to the "attention" or "warning" level based on the comparison results and duration. For example, when the deviation first exceeds the set tolerance and remains for the second preset duration, it is determined to be "attention"; if the deviation further increases or the duration is longer, it is determined to be "warning"; or, based on the specific value and duration of the deviation, it can be directly determined to be "attention" or "warning" through preset logical rules or decision tree models.

[0061] Through the aforementioned technical solution, this application trains a data-driven model based on historical normal operation data, ensuring that the model can accurately learn the operating characteristics of the meter connector in a healthy state, thereby establishing a reliable benchmark and effectively avoiding model bias caused by scarce samples. Simultaneously, based on real-time collected electrical and temperature data, it can accurately estimate the contact resistance characteristic value or its changing trend of the meter connector, directly quantifying the core physical quantity of connector degradation and enhancing the ability to identify progressive degradation characteristics. Furthermore, by comparing the estimation results with the benchmark trend predicted by the data-driven model, and combining this with the condition that the deviation exceeds a set tolerance and the deviation remains for a second preset duration, the accuracy and stability of health status determination are effectively improved by introducing tolerance and duration thresholds, avoiding misjudgments and making the in-depth analysis results more accurate and reliable, thus providing solid technical support for the health status assessment of meter connectors.

[0062] In some of the solutions mentioned above in this application, the preliminary analysis results and the in-depth analysis results are integrated to generate a health status level. However, in the process of implementation, due to the uncertainty of the preliminary analysis results and the in-depth analysis results, the fixed weight fusion mechanism may lead to inaccurate evaluation results and fail to dynamically adapt to real-time data changes, thereby affecting the reliability and accuracy of health status assessment.

[0063] In response, this application further proposes a dynamic weight fusion mechanism to fuse the results of the two analyses, including: assigning initial fusion weights to the two analyses based on the uncertainty assessment results of the preliminary analysis results and the in-depth analysis results; dynamically adjusting the initial fusion weights using real-time collected multi-source operating data as feedback; using an optimization algorithm to perform global optimization on the adjusted weight combination to determine the final fusion weights, and generating the health status level based on the final fusion weights.

[0064] The initial fusion weights, assigned based on the uncertainty assessments of the preliminary and in-depth analysis results, quantify the reliability, confidence level, or error range of each result. The preliminary analysis may be based on a pre-defined evaluation rule base, and its uncertainty may stem from the universality of the rules or the precision of the threshold settings. The in-depth analysis may be based on a data-driven model, and its uncertainty may arise from the completeness of the model training data, the model's generalization ability, or real-time data noise. Assigning initial fusion weights aims to give different levels of confidence to each result in the initial fusion phase, based on their inherent uncertainty, to avoid the bias of a single result having an excessive impact on the final evaluation. Specifically, statistical methods can be used, such as calculating the variance between the preliminary analysis results and historical actual states, or the confidence interval of the data-driven model's prediction results, to use these statistics as uncertainty assessment results. Then, based on the magnitude of the variance or the width of the confidence interval, the initial weights are determined using an inverse proportionality or normalization function; that is, the smaller the uncertainty, the larger the weight. Alternatively, fuzzy logic or expert systems can be used. Based on preset fuzzy rules and the output features of preliminary and in-depth analysis, a fuzzy inference system trained by expert experience or historical data can be used to assess its uncertainty level and assign initial weights accordingly.

[0065] Using real-time collected multi-source operational data as feedback to dynamically adjust the initial fusion weights is to enable the fusion mechanism to adapt to real-time changes in the operating conditions of the meter connector. Real-time collected multi-source operational data, such as temperature data at key connection points, electrical data flowing through circuits, and environmental data of the surrounding environment, can reflect the current state and environmental conditions of the connector. This data may reveal the limitations or advantages of preliminary or in-depth analysis under specific operating conditions. By using this real-time data as feedback, the initial weights can be corrected in a timely manner, ensuring the real-time nature and accuracy of the fusion results. Specifically, this can be achieved by monitoring the deviation between real-time multi-source operational data and model prediction results. For example, when there is a large deviation between the real-time data and the predicted value of the data-driven model, it may indicate that the model is performing poorly under the current operating conditions. In this case, the weight of the data-driven model can be appropriately reduced, and the weight of the preliminary analysis can be increased, and vice versa. Alternatively, adjustments can be made based on changes in system operating conditions reflected by real-time data. For example, when the ambient temperature rises sharply or the flowing current fluctuates significantly, it may be necessary to adjust the confidence level of the temperature threshold rule or contact resistance estimation model to adapt to these external disturbances.

[0066] An optimization algorithm is used to globally optimize the adjusted weight combination to determine the final fusion weight, and the health status level is generated based on the final fusion weight. Even after adjustments based on real-time data feedback, the weight combination may still exhibit local optima or suboptimal results. The global optimization algorithm aims to search the entire weight space to find the weight combination that most accurately reflects the connector's true health status, thus determining the final fusion weight. The final fusion weight is the optimal quantification of the contribution of the preliminary and in-depth analysis results to the health status level under the current operating conditions. Specifically, heuristic optimization algorithms, such as genetic algorithms or particle swarm optimization, can be used. These algorithms simulate natural selection or swarm intelligence behavior, searching in a multi-dimensional weight space to find the optimal weight combination; their objective function can be minimizing the error between the fusion result and the actual health status. Alternatively, gradient-based numerical optimization algorithms, such as gradient descent or Newton's method, can be used. By calculating the gradient of the objective function with respect to the weights, the weights are iteratively adjusted to approach the optimal solution along the gradient descent direction.

[0067] Through the above technical solution, this application effectively solves the uncertainty problem in fusing preliminary and in-depth analysis results, improving the accuracy and adaptability of health status assessment. Specifically, initial fusion weights are assigned based on the uncertainty assessment results of the preliminary and in-depth analysis results. By quantifying the degree of uncertainty, the initial weight setting is ensured to be more objective and reasonable, avoiding unscientific fusion starting points caused by subjective bias. The initial fusion weights are dynamically adjusted using real-time collected multi-source operational data as feedback. Continuous feedback is provided by real-time data streams, enabling the weights to adaptively optimize with changes in operational status, enhancing the real-time responsiveness of the fusion process. An optimization algorithm is used to globally optimize the adjusted weight combination to determine the final fusion weights. The global search algorithm avoids local optima, ensuring that the weight combination reaches overall optimality, improving the stability and reliability of the fusion results. A health status level is generated based on the final fusion weights. The optimized weights are applied to the status level calculation, ensuring that the final assessment result more accurately reflects the actual health status of the connector. This dynamic and adaptive fusion mechanism enables the health status assessment of meter connectors to no longer rely on fixed and potentially inaccurate weights, but to make intelligent adjustments based on real-time operating conditions and data characteristics. This improves the reliability and accuracy of health status assessment, provides a more solid foundation for subsequent maintenance decisions, effectively avoids misjudgments or omissions caused by inaccurate assessments, and ensures the stable operation of the power grid and the safety of users' electricity consumption.

[0068] In some of the solutions mentioned above in this application, maintenance decision instructions are generated and triggered based on health status levels to achieve maintenance response. However, in this process, maintenance decisions may lack differentiated processing for different health status levels, resulting in low maintenance efficiency and untimely response.

[0069] In response, this application further proposes to generate and trigger corresponding maintenance decision instructions based on the health status level, specifically including: when the health status level is "attention", generating a planned inspection task instruction; when the health status level is "warning", generating a time-limited on-site verification work order instruction and pushing it to the maintenance terminal; when the health status level is "alarm", generating an emergency response work order instruction and simultaneously sending a related early warning notification to the power distribution management system.

[0070] The generation and triggering of corresponding maintenance decision instructions based on health status levels refers to the system automatically or semi-automatically creating and issuing specific maintenance operation instructions based on the health status assessment results of the meter connectors. Its function is to transform abstract health status assessment results into executable actions, ensuring the timeliness and targeted nature of maintenance work. This can be achieved through a pre-defined rule engine or expert system. When a specific health status level is received, the system queries a predefined rule base, matches the corresponding maintenance decision template, fills in relevant information such as connector ID, location, and status details, and then generates the instruction. Alternatively, it can be implemented through a machine learning-based decision support system. After receiving the health status level, the system combines historical maintenance data and expert experience, uses a trained model to predict the most appropriate maintenance decision, and then generates the instruction.

[0071] When the health status level is "Attention," a planned inspection task instruction is generated. This is designed for meter connectors that are in a "Attention" state—meaning there is a potential risk or minor abnormality, but it has not yet reached the level of emergency handling. The system generates a non-urgent, pre-schedulable maintenance instruction. This helps to conduct preventative checks before problems escalate, avoiding wasted resources. For example, the system can automatically create a new planned inspection task in the maintenance management system, specifying the task type, target connectors, recommended inspection items, and estimated completion time, and assign it to the appropriate inspection team or personnel. Alternatively, the system generates an electronic inspection checklist, including a list of connectors requiring attention and recommended inspection items, and sends it to the relevant personnel via email or internal messaging system for them to arrange the specific inspection plan.

[0072] When the health status level is at the warning level, a time-limited on-site verification work order is generated and pushed to the maintenance terminal. This is designed for meter connectors in a warning state, indicating a high level of risk requiring timely intervention to prevent further deterioration. The system generates an on-site verification instruction with a clear time limit, ensuring that the instruction reaches the personnel executing the task quickly. This emphasizes the urgency and directness of maintenance. Specifically, the system can create a time-limited on-site verification work order in the maintenance management platform, specifying the verification content, deadline, and responsible person, and push the work order details to the maintenance personnel's mobile terminal devices in real time via mobile applications or SMS. Alternatively, the system can generate an electronic work order containing detailed verification requirements and time limits, sending it directly to the on-site maintenance personnel's work terminal via a dedicated communication protocol or API interface, requiring the recipient to confirm to ensure information delivery.

[0073] When the health status level is alarmed, an emergency response work order is generated, and a related early warning notification is simultaneously sent to the power distribution management system. This is designed to address situations where a meter connector is in an alarm state, indicating a serious fault or impending fault requiring immediate action. The system generates a highest-priority maintenance instruction and simultaneously notifies higher-level management systems for rapid response and global coordination. This ensures rapid and coordinated handling of emergencies. For example, the system can immediately create an emergency response work order, mark it as highest priority, and automatically trigger SMS, phone, or in-app notifications, requiring maintenance personnel to immediately proceed to the site. Simultaneously, through data interfaces or message queues, a related early warning notification containing alarm information, connector location, and potential impact range is sent to the power distribution management system for collaborative decision-making such as load adjustments and isolation operations. Alternatively, the system generates an emergency work order and, through the interface of the SCADA monitoring and data acquisition system or EMS energy management system, synchronously transmits the alarm information and suggested handling measures to the power distribution management system, triggering its internal emergency response processes, such as automatically generating dispatch instructions or issuing alarms to dispatchers.

[0074] Through the above technical solution, this application addresses the problem of low maintenance efficiency and untimely response caused by the lack of differentiated processing for different health status levels in maintenance decisions, thereby achieving refined and differentiated maintenance decisions. Specifically, when the health status level is "Concern," a planned inspection task instruction is generated, avoiding overreaction to minor anomalies, saving maintenance resources, and realizing preventative maintenance. When the health status level is "Warning," a time-limited on-site verification work order instruction is generated and pushed to the maintenance terminal, ensuring timely intervention in potential risks, preventing further deterioration of problems, and improving response efficiency. When the health status level is "Alarm," an emergency handling work order instruction is generated, and a related early warning notification is simultaneously sent to the power distribution management system, realizing rapid response to serious faults and system-level collaborative processing, minimizing accident risks and impacts. Based on digital twins, multi-source operational data acquisition, and health status assessment model output of health status levels, this solution ensures that the generation of maintenance decision instructions is based on comprehensive and accurate health status assessment results, thus making maintenance decisions more scientific and effective. This differentiated maintenance strategy makes resource allocation more rational, optimizes maintenance costs, and improves the operational reliability of meter connectors and the security of the entire power system.

[0075] In some of the solutions mentioned above in this application, the results data of on-site maintenance are received and optimized and updated based on the results data to improve the adaptability of the model. However, in this process, the feedback data is not structured, correlated and accumulated, resulting in limited optimization effect and inability to continuously improve the evaluation accuracy.

[0076] To address this, this application further proposes a method for managing the health status of electricity meter connectors. This method involves receiving feedback data from on-site maintenance and optimizing and updating the data accordingly. This includes: receiving and recording on-site processing results and retest data from maintenance personnel; associating the on-site processing results and retest data with multi-source operational data corresponding to the triggered maintenance decision command to form case data; adjusting threshold parameters in the evaluation rule base based on the accumulated case data; and / or retraining the data-driven model. Specifically, during the optimization process, case data is structured and stored and analyzed according to dimensions such as environmental humidity and load rate. For the evaluation rule base, thresholds can be dynamically fine-tuned based on case statistics, for example, by introducing a humidity compensation coefficient. For the data-driven model, an incremental learning strategy can be adopted, adding validated new cases to the training set for local weight updates; or, after accumulating sufficient data, initiating full retraining to improve model adaptability and evaluation accuracy.

[0077] Specifically, the purpose of receiving and recording on-site processing results and retest data from maintenance personnel is to ensure that information generated during maintenance activities can be captured and stored by the system, providing a raw and authentic data foundation for subsequent analysis and model optimization. For example, the system can provide a maintenance work order feedback portal through a mobile application or web interface. After completing on-site work, maintenance personnel can enter the processing results and retest data through this portal, and the system will automatically timestamp and store them. Alternatively, maintenance personnel can use a handheld device to scan the QR code or NFC tag on the meter connector. The system will automatically retrieve the maintenance work order for that connector, and the maintenance personnel can fill in the processing results and retest data, and upload on-site photos or videos as supporting evidence. The data is uploaded to the management platform in real time via a wireless network for recording.

[0078] In linking on-site processing results and retest data with the multi-source operational data corresponding to the triggered maintenance decision command to form case data, the aim is to establish a logical connection between maintenance feedback and the original operational state that led to the maintenance, integrating discrete data points into meaningful "cases" to reveal the entire process of fault occurrence, development, and handling. For example, when generating a maintenance decision command, the system records multi-source operational data of the meter connector on which the command is based, including temperature data of key connection points, electrical data flowing through the circuit, and timestamps and data snapshots of environmental data of the surrounding environment. When maintenance feedback data is received, the system automatically matches and associates this data based on the unique identifier of the meter connector and the ID of the maintenance command to form a complete case record. As another example, the management platform can be designed with a case database, where each case contains a unique case ID. When a maintenance decision command is triggered, the system packages and stores the command content, the health status level at the time of triggering, and the corresponding multi-source operational data into the case database. On-site processing results and retest data reported by maintenance personnel are bound to this case ID to supplement and improve the case, ensuring that all relevant information is concentrated under a single case entry.

[0079] The aim of adjusting threshold parameters in the assessment rule base based on accumulated case data and / or retraining the data-driven model is to continuously optimize the accuracy and adaptability of health status assessments using actual maintenance experience and results, enabling the assessment system to learn and improve from practice. For example, the system can analyze case data periodically or after accumulating a certain amount. If a threshold is found to frequently cause false alarms (i.e., triggering a warning / alarm but the on-site verification result is normal), or to result in missed alarms (i.e., a problem is found on-site but the system does not trigger the corresponding level), then statistical analysis or machine learning algorithms, including genetic algorithms and particle swarm optimization algorithms, can be used to automatically or semi-automatically fine-tune the relevant threshold parameters in the assessment rule base to better reflect actual working conditions. Furthermore, accumulated case data, especially data containing multi-source operational data before and after a failure, as well as maintenance processing results, can serve as new training samples. The system can use these new, validated health / abnormal samples to incrementally learn or fully retrain the existing data-driven model. For example, this data can be used to update model weights, adjust model structure, or train new classifiers to improve the model's accuracy and generalization ability in recognizing different health status levels.

[0080] By receiving and recording on-site handling results and retest data from maintenance personnel, this method can obtain verified, real-world information about the health status of meter connectors. Furthermore, these on-site handling results and retest data are structurally correlated with the multi-source operational data corresponding to the triggered maintenance decision commands, thus forming case data encompassing the entire process of fault occurrence, development, assessment, and handling. This accumulated case data, as valuable experiential knowledge, can be used to continuously optimize the health status assessment model. Specifically, threshold parameters in the assessment rule base can be adjusted based on this case data to better reflect actual operating conditions and reduce false alarms or missed alarms. Simultaneously, this real, verified data can be used to retrain the data-driven model, thereby improving the model's accuracy and generalization ability in identifying the health status of meter connectors. This closed-loop optimization mechanism allows the meter connector health status management method to continuously learn and improve from actual operation and maintenance experience, enhancing the accuracy, adaptability, and robustness of the assessment, ensuring that the system can provide accurate health status assessments and maintenance decision support stably over the long term.

[0081] In some of the solutions mentioned above in this application, a method for managing the health status of meter connectors is proposed to achieve health assessment and maintenance decisions. However, in this process, users cannot intuitively monitor the real-time health status distribution of each connector in the power grid, nor can they query historical data to analyze the changing trends of temperature, contact resistance and status level, resulting in insufficient decision support, low fault location efficiency, and difficulty in identifying progressive degradation problems.

[0082] In this regard, this application further proposes that the method also includes displaying the real-time health status level of each meter connector digital twin in a differentiated visual form on the monitoring interface based on the power grid topology map; and providing a historical data query function, wherein the historical data includes at least temperature change curves, contact resistance trend curves and status level change records.

[0083] Specifically, the monitoring interface refers to a graphical user interface (GUI) for human-computer interaction, displaying system status, and operational control. Its function is to provide a unified view, enabling users to understand the real-time operating status of the meter connectors. This monitoring interface can be a standalone desktop application, a web-based browser interface, or integrated into an existing SCADA system or Distribution Management System (DMS). The power grid topology map is a graphical representation of the physical connections and logical relationships of the power grid, typically including equipment such as substations, lines, switches, and meters, and their interconnections. Its function is to provide the geographical location and electrical connection context of the meter connectors within the entire power grid, facilitating users to quickly locate and understand their impact range. This power grid topology map can be generated based on Geographic Information System (GIS) data, mapping the digital twins of the meter connectors to the actual geographical location and power grid structure; alternatively, it can be a simplified logical topology map, displaying only key equipment and connections. By associating the real-time health status levels output by the aforementioned health status assessment model, such as normal, attention, warning, and alarm, with preset differentiated visual formats and rendering them on the power grid topology map, differentiated visual displays can be achieved. For example, color coding can be used, such as green for normal, yellow for attention, orange for warning, and red for alarm; icon changes can also be used, such as icons of different shapes or with exclamation marks; or animation effects, such as flashing or size changes, can be used to highlight connectors in abnormal states, thereby quickly distinguishing meter connectors of different health status levels through intuitive visual cues and improving information identification efficiency.

[0084] Furthermore, this method also provides a historical data query function, which aims to allow users to review the key operating parameters and status changes of the meter connector over a period of time, in order to conduct trend analysis, fault tracing, and maintenance strategy optimization. This function typically provides a time range selector and displays the query results in the form of charts or tables, supporting data export. The historical data includes at least a temperature change curve, which is based on temperature data of key connection points of the meter connector collected in real time through a sensor network, visually displaying the temperature change trend over time, helping to identify problems such as abnormal heating, overheating risk, or poor heat dissipation. The historical data also includes a contact resistance trend curve, which is based on the contact resistance characteristic value of the meter connector or its change trend estimated by a data-driven model, reflecting the degree of deterioration of the internal connection condition of the connector, and is an important indicator for judging progressive faults such as poor contact and oxidation corrosion. Simultaneously, the historical data also includes a status level change record, which records the history of the meter connector's health status level changing over time based on the health status level output by the health status assessment model, helping to analyze the process of status deterioration, verify the accuracy of the assessment model, and provide historical evidence for maintenance decisions.

[0085] Through the aforementioned technical solutions, users can clearly grasp the health status distribution of all meter connectors in the entire power grid, quickly identify abnormal areas and fault points, and greatly improve the efficiency and accuracy of fault location. This visualization method combines abstract health levels with specific physical locations and power grid structures, enabling maintenance personnel to respond quickly. Simultaneously, it provides historical data query functions, allowing maintenance personnel to deeply analyze the long-term trends and status evolution of key parameters of meter connectors. This helps identify gradual degradation characteristics that are not easily detected in real-time, such as a slow increase in contact resistance or continuous small fluctuations in temperature, thereby achieving earlier warnings and preventative maintenance, avoiding sudden failures. The combination of real-time monitoring and historical data analysis provides comprehensive, intuitive, and in-depth data support for maintenance decisions. Maintenance personnel can not only see the current health status but also trace its evolution history, thereby developing more accurate and forward-looking maintenance strategies, shifting from reactive emergency repairs to proactive prevention, and improving the maintenance management level of meter connectors and the reliability of power supply.

[0086] In some of the solutions mentioned above in this application, a method for managing the health status of electricity meter connectors is proposed to solve the problems of single monitoring data, lack of digital modeling, rudimentary health assessment methods, lack of closed-loop optimization in management processes, and lagging maintenance decisions in the existing technology. However, when implementing this method, a reliable storage medium is needed to ensure that the computer program can be stably executed by the processor, so as to efficiently deploy the method in practical applications and avoid execution obstacles caused by unstable program operation or low deployment efficiency.

[0087] In this regard, this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements a method for managing the health status of an electricity meter connector.

[0088] The computer-readable storage medium refers to a physical device capable of storing digital data and allowing a computer system to read and write data. This medium serves as a carrier of computer programs, ensuring persistent storage and reliable access to program code. Specifically, the computer-readable storage medium can be a non-volatile memory, such as a hard disk drive, solid-state drive, or flash memory, which retains data even after power is lost and is suitable for long-term storage. Alternatively, the computer-readable storage medium can also be an optical disc or a removable storage device, which facilitates program distribution, transmission, and installation.

[0089] The computer program refers to a collection of instructions designed to enable a computer to perform specific tasks or operations. This program encapsulates the logic and algorithm of the meter connector health status management method and is the core of its functionality. For example, the computer program can be a compiled program, written in high-level languages ​​such as C++ or Java and compiled into machine code or bytecode, allowing direct execution on the target system. Alternatively, the computer program can be an interpreted program, written in languages ​​such as Python or JavaScript, executed line by line by an interpreter at runtime.

[0090] The program being executed by a processor means that the processor, such as a central processing unit (CPU) or dedicated processing unit, operates according to the instruction sequence defined in the computer program to complete the predetermined calculation and control tasks. This process transforms static program code into dynamic, executable operations, driving the entire health status management method. Specifically, the program can be executed by a general-purpose processor, with the operating system responsible for loading the program into memory and scheduling its execution. Alternatively, the program can also be executed by a dedicated processor or microcontroller, which may be optimized for specific computational tasks to improve execution efficiency.

[0091] The method for managing the health status of electricity meter connectors refers to a computer program that, through its internal logic and algorithms, completely executes each step and function of the aforementioned electricity meter connector health status management method. This ensures that the management method can operate automatically, efficiently, and accurately, thereby achieving intelligent monitoring and preventative maintenance of electricity meter connectors. For example, the program can adopt a modular design, encapsulating functions such as digital twin construction, multi-source data acquisition, health status assessment, maintenance decision generation, and model optimization and updating into independent software modules, which then work collaboratively through interfaces. Alternatively, a service-oriented architecture (SOA) or microservice architecture can be adopted, deploying different functions in the management method as independent services that communicate and collaborate via a network to improve the system's scalability and flexibility.

[0092] Through the above technical solution, this application provides a reliable carrier and execution mechanism, effectively solving the reliability and efficiency problems in deploying and executing the meter connector health status management method. The computer-readable storage medium ensures persistent storage and reliable access to the computer program, avoiding data loss or access delays, thereby improving program stability. Simultaneously, the computer program stored on it defines the software code implementing the management method, facilitating distribution and updates, and supporting standardized implementation of the method. When the program is executed by a processor, it can automatically run the management process, achieving automated operation, reducing the need for manual intervention, and improving execution consistency and real-time performance. This enables the complex functions of the above-mentioned meter connector health status management method, such as constructing a digital twin, real-time collection of multi-source operating data, deep analysis based on an evaluation rule base and data-driven model, employing a dynamic weight fusion mechanism, and generating maintenance decision instructions, to be implemented and deployed in a stable and efficient environment. This effectively overcomes the execution obstacles caused by deployment difficulties, unstable program operation, or low deployment efficiency in existing technologies, ensuring the efficient and stable operation of the method in practical applications.

[0093] At the system implementation level, the sensor network can adopt LoRa or NB-IoT wireless networking, with data aggregated to the management platform via a gateway. The digital twin is represented using an object-oriented architecture, including identity identifiers, static attributes, real-time data stream pointers, and historical indexes. The health status assessment model provides a RESTful API interface in the form of microservices. Maintenance instructions are pushed to the production management system or mobile work terminals via message queues, following a standard work order model. The monitoring interface is based on WebGIS technology, rendering real-time health status on the power grid topology map using color- and shape-coded icons, supporting click-to-query details, historical curves, and status change records.

[0094] The following example will provide a more detailed explanation of the above technical solution: This method first constructs and runs a corresponding digital twin for the meter connector. Specifically, a unique digital identity, such as MTR-CONN-001, is assigned to the physical meter connector. This identity is associated with and bound to the connector's static attribute information to form a static attribute file, which includes its asset number, precise installation location information, and its branch information within the power grid topology. These static attribute files are stored on a management platform, forming the basic data layer of the digital twin. In this way, the connector's location and relationships within the power grid are digitally represented, solving the problem of difficult fault location caused by the lack of digital modeling in existing technologies.

[0095] While the digital twin is operating, multi-source operational data of the meter connector is collected in real time via a sensor network. Specifically, this includes: installing temperature sensors at key conductive connection points of the connector to collect real-time temperature data; acquiring real-time electrical data flowing through the circuit, such as current, voltage, and power, via an energy metering device connected to the meter connector; and deploying environmental sensors in the installation environment of the meter connector to collect ambient temperature and humidity data. This fusion of multi-source data overcomes the shortcomings of existing technologies that rely on single-source monitoring data and cannot comprehensively reflect the true health status of the connector.

[0096] Subsequently, this real-time collected multi-source operational data is input into the health status assessment model to output the health status level of the meter connector. This health status level includes at least four levels: Normal, Attention, Warning, and Alarm. The process of outputting the health status level includes three stages: In the first stage, preliminary analysis of multi-source operational data is performed based on a pre-defined evaluation rule base. For example, the evaluation rule base includes multi-level threshold judgment rules based on temperature parameters. These threshold parameters can be set according to the equipment's physical characteristics and operational statistics. For instance, the first preset threshold temperature difference initial alarm value can be 8-12°C, the second preset threshold temperature difference escalation value can be 15-20°C, the third preset threshold maximum temperature difference can be 5-8°C, and the fourth preset threshold temperature rise rate can be 1-2°C / minute. The first and second preset durations can be set to 30 minutes and 2 hours respectively to avoid instantaneous interference. The initial threshold values ​​can be derived from equipment technical specifications, industry standards, or historical statistical averages and are set as configurable parameters in the system.

[0097] When the temperature difference between the critical connection point of connector L1 and the ambient temperature exceeds a first preset threshold (e.g., 8°C), and this difference remains for a first preset duration (e.g., 2 hours), the connector is initially determined to be in a state of concern. If the temperature difference further exceeds a second preset threshold (e.g., 15°C), and the maximum temperature difference between different phase connection points within the same connector unit exceeds a third preset threshold (e.g., 5°C), the initial determination is to issue a warning. When the temperature of the critical connection point exceeds a safety limit (e.g., 80°C), or its temperature rise rate exceeds a fourth preset threshold (e.g., 5°C / minute), the initial determination is to issue an alarm. This multi-level threshold judgment rule is more refined than the simple fixed threshold judgment in existing technologies.

[0098] The second stage involves in-depth analysis of multi-source operational data based on a data-driven model. This model has been trained on historical normal operation data, learning the operating characteristics of the meter connector in a healthy state. Through real-time collected electrical and temperature data, the model estimates the contact resistance characteristic value or its changing trend of the meter connector. For example, if the estimated contact resistance shows a continuous upward trend, and its deviation from the baseline trend predicted by the data-driven model exceeds a set tolerance limit (e.g., 5%), and remains there for a second preset duration (e.g., 4 hours), the data-driven model determines the status as "concerned" or "warning." This trend-based analysis solves the problem of existing technologies being unable to identify progressive degradation characteristics.

[0099] In the third stage, based on the uncertainties inherent in both the preliminary and in-depth analysis results, a dynamic weight fusion mechanism is employed to combine the two results and generate the final health status level. For example, if the rule base analysis result is of concern and has a high confidence level, while the data-driven model analysis result is also of concern but with a slightly lower confidence level, the system will dynamically adjust the fusion weights of the two based on real-time multi-source operational data as feedback. An optimization algorithm is used to globally optimize the adjusted weight combination to determine the final fusion weights, and the final health status level is generated based on these final fusion weights. This fusion mechanism effectively combines the advantages of mechanistic models and data-driven models, improving the accuracy of the assessment and overcoming the shortcomings of existing technologies, such as rudimentary assessment methods and the failure to effectively integrate the advantages of both.

[0100] Based on the generated health status level, the system generates and triggers corresponding maintenance decision instructions. For example, when the health status level is "Concern," the system generates a planned inspection task instruction, notifying maintenance personnel to conduct on-site inspections within a specified time. When the health status level is "Warning," the system generates a time-limited on-site verification work order instruction and pushes it to the maintenance terminal, requiring maintenance personnel to immediately go to the site for verification. When the health status level is "Alarm," the system generates an emergency response work order instruction and simultaneously sends a related early warning notification to the power distribution management system for rapid response. This differentiated maintenance decision instruction enables preventative management, avoiding the problems of delayed or inefficient maintenance decisions in existing technologies.

[0101] Finally, the system receives feedback data from on-site maintenance and optimizes and updates the assessment rule base and / or data-driven model based on this data. For example, after completing on-site verification, maintenance personnel will provide feedback on the on-site processing results and retest data. The system associates these on-site processing results and retest data with the multi-source operational data corresponding to the maintenance decision command, forming case data. Based on the accumulated case data, the system can adjust the threshold parameters in the assessment rule base and / or retrain the data-driven model to better adapt it to the usage needs of different regions and operating conditions. This closed-loop optimization mechanism solves the problems of lack of closed-loop optimization in management processes and rigid assessment standards in existing technologies, continuously improving the accuracy and reliability of health status assessment.

[0102] In addition, the monitoring interface displays the real-time health status level of each meter connector's digital twin in a differentiated visual format based on the power grid topology map, and provides historical data query functions, including temperature change curves, contact resistance trend curves, and status level change records, providing intuitive and comprehensive information support for operation and maintenance personnel.

[0103] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for managing the health status of an electricity meter connector, characterized in that, include: Construct and run a digital twin corresponding to the meter connector, wherein the digital twin is bound to the static attribute information of the connector and its association in the power grid topology; The meter connectors are collected in real time using a sensor network. The multi-source operating data includes at least the temperature data of their key connection points, the electrical data of the circuits flowing through them, and the environmental data of the surrounding environment. The multi-source operating data is input into the health status assessment model to output the health status level of the meter connector. The health status level includes at least normal, attention, warning and alarm. The output of the health status level includes: The multi-source operational data is preliminarily analyzed based on a preset evaluation rule base; The multi-source operational data is analyzed in depth based on a data-driven model. Based on the uncertainty of the preliminary analysis results and the in-depth analysis results, a dynamic weight fusion mechanism is used to fuse the two results to generate the health status level; Based on the health status level, generate and trigger corresponding maintenance decision instructions; Receive the results data from on-site maintenance, and optimize and update the evaluation rule base and / or the data-driven model based on the results data.

2. The method for managing the health status of an electricity meter connector according to claim 1, characterized in that, The construction and operation of the digital twin corresponding to the meter connector includes: Assign a unique digital identity to each physical meter connector or its integrated unit; The digital identity identifier is associated and bound with the corresponding static attribute information to form a static attribute file. The static attribute information includes at least asset information, installation location information, and branch line information. The static attribute files are stored on the management platform, forming the basic data layer of the digital twin.

3. The method for managing the health status of an electricity meter connector according to claim 1, characterized in that, The method of collecting multi-source operational data of the meter connector in real time through a sensor network includes: Real-time temperature data of the part is collected by a temperature sensing device installed at the key conductive connection part of the meter connector. The electrical data flowing through the circuit is obtained by an energy metering device connected to an electricity meter connector. The electrical data includes at least current, voltage and power data. Environmental data is collected by an environmental sensing device deployed in the installation environment of the electricity meter connector. The environmental data includes at least ambient temperature and ambient humidity data.

4. The method for managing the health status of an electricity meter connector according to claim 1 or 3, characterized in that, The preset evaluation rule base includes multi-level threshold judgment rules based on temperature parameters, including: Level 1 rule: When the temperature difference between the critical connection point and the ambient temperature exceeds a first set threshold and the difference remains for a first preset duration, the status is determined to be "attention". Second-level rule: When the difference exceeds the second set threshold, the second set threshold is greater than the first set threshold, and the maximum temperature difference between different phase connection points in the same connector unit exceeds the third set threshold, the status is determined to be a warning. Level 3 rule: When the temperature value of a critical connection point exceeds the safety limit, or its temperature rise rate exceeds the fourth set threshold, the status is determined to be an alarm.

5. The method for managing the health status of an electricity meter connector according to claim 1, characterized in that, The in-depth analysis of the multi-source operational data based on the data-driven model includes: The data-driven model is trained based on historical normal operation data to learn the operating characteristics of the meter connector under health conditions; Based on real-time collected electrical and temperature data, the contact resistance characteristic value or its changing trend of the meter connector is estimated. The contact resistance characteristic value or its changing trend is compared with the baseline trend predicted by the data-driven model. When the deviation exceeds the set tolerance and the deviation is maintained for a second preset time, the status is determined to be either "attention" or "warning".

6. The method for managing the health status of an electricity meter connector according to claim 1, characterized in that, The method of fusing the two results using a dynamic weight fusion mechanism includes: Based on the uncertainty assessment results of the preliminary analysis results and the in-depth analysis results, initial fusion weights are assigned to the two; The initial fusion weights are dynamically adjusted using real-time multi-source operational data as feedback. An optimization algorithm is used to perform global optimization on the adjusted weight combination to determine the final fusion weight, and a health status level is generated based on the final fusion weight.

7. The method for managing the health status of an electricity meter connector according to claim 1, characterized in that, The process of generating and triggering corresponding maintenance decision instructions based on health status levels includes: When the health status level is "concerned", a planned inspection task instruction is generated. When the health status level is warning, a work order for on-site verification within a specified period is generated and pushed to the maintenance terminal; When the health status level is alarm, an emergency response work order is generated, and a related early warning notification is sent to the power distribution management system simultaneously.

8. The method for managing the health status of an electricity meter connector according to claim 1, characterized in that, The process of receiving feedback data from on-site maintenance and optimizing and updating the data includes: Receive and record the on-site handling results and retest data reported by maintenance personnel; The on-site processing results and retest data are correlated with the multi-source operational data corresponding to the time the maintenance decision command is triggered to form case data; Based on accumulated case data, the threshold parameters in the evaluation rule base are adjusted, and / or the data-driven model is retrained.

9. The method for managing the health status of an electricity meter connector according to claim 1, characterized in that, The method further includes: On the monitoring interface, based on the power grid topology map, the real-time health status level of each meter connector digital twin is displayed in a differentiated visual form. It provides a historical data query function, and the historical data includes at least temperature change curves, contact resistance trend curves, and status level change records.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the method for managing the health status of the meter connector as described in any one of claims 1 to 9.

Images