Method and system for improving electric power metering verification detection service capability

Abstract

The invention provides an electric power metrological verification detection service capability improving method and system, and belongs to the technical field of electric power metrological verification detection services, and the method comprises the steps: obtaining multi-source data under electric power metrological verification of a power transmission line: equipment operation data, environment parameters and detection process data; based on the equipment operation data and the environmental parameters, obtaining an optimal working time sequence and task allocation scheme of the power transmission line electric power metering verification detection equipment, the environmental control equipment and the data acquisition equipment through the equipment cooperation model, and executing an electric power metering verification detection task to obtain actual process execution data; comparing the actual process execution data with the predicted process effect data of the optimal working time sequence and task allocation scheme to obtain an effect deviation and a bottleneck link; in response to the fact that the effect deviation is larger than a preset deviation threshold value, parameter adjustment is conducted on the equipment cooperation model through an optimization algorithm, optimized model parameters are obtained, the preset equipment cooperation model is updated, and a process optimization strategy for the bottleneck link is generated.

Description

Technical Field

[0001] This application relates to the field of power metering verification and testing service technology, and in particular to a method and system for improving power metering verification and testing service capabilities. Background Technology

[0002] The methods used for power metering verification and testing include manual inspection and traditional data analysis. Manual inspection involves professionals periodically visiting the site to check and test power metering equipment, recording its operating status and related data. Traditional data analysis involves simple statistical analysis of the collected data, judging whether the equipment is operating normally based on preset rules and experience. Manual inspection is inefficient, easily affected by human factors, and cannot accurately capture the equipment's operating status. Traditional data analysis methods can only handle simple data relationships and cannot comprehensively consider the connections between multiple sources of data, such as equipment operating data, environmental parameters, and testing process data. Fixed models and algorithms cannot be adjusted according to actual conditions to obtain the optimal working sequence and task allocation scheme for the equipment, resulting in the power metering verification and testing service capacity failing to meet demand.

[0003] Therefore, there is an urgent need for a method and system to improve the power metering verification and testing service capabilities. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides a method and system for improving the capability of power metering verification and testing services.

[0005] A first aspect of this application provides a method for improving the capability of electricity metering verification and testing services, including: Acquire multi-source data for power transmission line metering verification; the multi-source data includes equipment operation data, environmental parameters, and testing process data. Based on the equipment operation data and the environmental parameters, the optimal working sequence and task allocation scheme of the power transmission line power metering verification and testing equipment, environmental control equipment and data acquisition equipment are obtained through a preset equipment collaboration model. Based on the optimal work sequence and task allocation scheme, the power metering verification and testing task is executed to obtain the actual process execution data; The actual process execution data is compared and analyzed with the predicted process effect data corresponding to the optimal work sequence and task allocation scheme to obtain the effect deviation and bottleneck links. In response to the effect deviation being greater than a preset deviation threshold, the parameters of the preset device collaboration model are adjusted using an optimization algorithm to obtain optimized model parameters; The preset device collaboration model is updated based on the optimized model parameters, and a process optimization strategy for the bottleneck link is generated.

[0006] A second aspect of this application provides a system for improving the capability of electricity metering verification and testing services, including: The data acquisition module is used to acquire multi-source data under the power transmission line metering verification; the multi-source data includes equipment operation data, environmental parameters, and testing process data. The collaborative optimization module is used to obtain the optimal working sequence and task allocation scheme of the power transmission line power metering verification and testing equipment, environmental control equipment and data acquisition equipment based on the equipment operation data and the environmental parameters through a preset equipment collaboration model; The task execution module is used to execute power metering verification and testing tasks based on the optimal working sequence and task allocation scheme, and obtain actual process execution data. The performance analysis module is used to compare and analyze the actual process execution data with the predicted process effect data corresponding to the optimal work sequence and task allocation scheme to obtain the effect deviation and bottleneck links. The parameter adjustment module is used to adjust the parameters of the preset device collaboration model through an optimization algorithm in response to the effect deviation being greater than a preset deviation threshold, so as to obtain the optimized model parameters. The update and optimization module is used to update the preset device collaboration model based on the optimized model parameters and generate a process optimization strategy for the bottleneck link.

[0007] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for improving the capability of electricity metering verification and testing services.

[0008] In a fourth aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for improving the capability of power metering verification and testing services.

[0009] The beneficial effects of the method and system for improving the power metering verification and testing service capabilities provided in this application are as follows: By acquiring multi-source data under power transmission line metering verification and using the equipment collaboration model to obtain the optimal working sequence and task allocation scheme, the application can achieve efficient collaborative work of equipment; by executing testing tasks and comparing and analyzing the actual and predicted process effect data, it can identify effect deviations and bottlenecks; by adjusting the parameters of the equipment collaboration model and updating the model according to the effect deviations, it can generate process optimization strategies, thereby improving the power metering verification and testing service capabilities. Attached Figure Description

[0010] Figure 1 A flowchart illustrating a method for enhancing power metering verification and testing service capabilities according to an embodiment of this application; Figure 2 A structural block diagram of a power metering verification and testing service capability enhancement system provided in an embodiment of this application; Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0011] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0012] To make the purpose, technical solution, and advantages of this application clearer, the following will be described in conjunction with the appendix. Figure 1-3 The following is an explanation using specific examples.

[0013] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a method for improving the capability of electricity metering verification and testing services according to an embodiment of this application. The method includes: S101: Obtain multi-source data for power transmission line metering verification; multi-source data includes equipment operation data, environmental parameters, and testing process data.

[0014] In this embodiment, power transmission line metering verification is a professional technical activity involving the accuracy testing, performance verification, and compliance determination of equipment used for metering electrical energy in power transmission line systems, such as electricity meters and instrument transformers, according to national or industry standards, to ensure accurate and fair electricity metering. Multi-source data refers to a collection of structured / unstructured data collected from multiple different sources and dimensions related to the power transmission line metering verification scenario. Its characteristics include diverse data types and comprehensive coverage of various scenarios.

[0015] In this embodiment, the equipment operation data comes directly from the real-time or historical operating status data of the power metering verification equipment, indicating whether the equipment is working normally and how well it performs. Examples include: the detection error value, runtime, energy consumption, and fault codes of the verification equipment; the transmission rate and acquisition accuracy of the data acquisition equipment; and the temperature and humidity control accuracy of the environmental control equipment. Environmental parameters are external environmental data that affect the verification accuracy and equipment operation, indicating whether the verification environment is stable and compliant. Examples include: the temperature (required to be 18-25℃), humidity (40%-60%), air pressure, and electromagnetic interference intensity of the verification laboratory; and the wind speed and light intensity of outdoor verification scenarios. The testing process data records the entire verification process, indicating how the verification task is executed and its efficiency. Examples include: the time consumed for each verification step (equipment preheating, parameter calibration, data verification), task allocation records (which equipment / person is responsible for which task), and the pass rate of process nodes (such as the pass rate of the parameter calibration stage).

[0016] Specifically, as a further optimization of this embodiment, before inputting multi-source data into the preset device collaboration model, data preprocessing is required, including: aligning the timestamps of data from different sampling devices based on a unified network clock protocol; filling fields with missing values ​​using linear interpolation of data from previous and subsequent times; and performing Z-score standardization on continuous numerical features such as device operation data and environmental parameters to eliminate the influence of dimensions.

[0017] S102: Based on equipment operation data and environmental parameters, the optimal working sequence and task allocation scheme of the power transmission line metering and verification testing equipment, environmental control equipment, and data acquisition equipment are obtained through a preset equipment coordination model. In this embodiment, the preset equipment coordination model adopts a Long Short-Term Memory (LSTM) network model. This model is an intelligent decision-making model built based on machine learning, operations research, and other algorithms. In this embodiment, the LSTM network model is used to integrate equipment operation data and environmental parameters, simulate the effects of different equipment combinations, and finally output the optimal solution that balances efficiency, accuracy, and energy consumption, thus solving the problems of timing conflicts and task mismatches among multiple devices.

[0018] The Long Short-Term Memory (LSTM) network model comprises a multi-layer encoder, with LSTM units at its core. Each unit includes an input gate, a forget gate, an output gate, and a cell state. Input data: The LSTM network model receives a multi-dimensional time series as input, composed of the following components: Equipment operating status sequence: e.g., voltage, current, power, and self-test error values ​​of key testing equipment over the past N hours. Environmental parameter sequence: e.g., temperature, humidity, and vibration amplitude in the laboratory over the past N hours. Task features: the type of testing task to be performed (e.g., electricity meter accuracy testing, transformer error testing), the preset stringency level, etc. (These features are replicated across the entire sequence length to align with the time series). Input normalization: All input features are normalized before being input into the model to accelerate the convergence of the LSTM network model.

[0019] The output data is as follows: the output layer of the last time step is connected to a fully connected layer, and the output of the fully connected layer is the optimal working sequence and task allocation scheme. This is a structured vector, including: the equipment startup sequence, for example: start the constant temperature chamber first, start the standard source 30 minutes later, and start the error calculation device 5 minutes later; the task allocation matrix, for example: prioritize the allocation of Class A verification tasks to the testing equipment numbered X, which has a higher health level; the prediction of the expected time consumption of each step; and the predicted process effect data, such as the overall verification uncertainty and the expected completion time.

[0020] In this embodiment, the power transmission line metering verification and testing equipment is the device that directly performs accuracy testing and performance verification of power metering equipment. It is the main body responsible for executing the verification task, such as fully automatic energy meter verification devices, transformer calibrators, and high-voltage metering devices. Environmental control equipment is auxiliary equipment used to adjust environmental parameters in the verification scenario. It is the fundamental support for ensuring verification accuracy, maintaining environmental stability to ensure verification results meet standards. Examples include constant temperature and humidity units, electromagnetic shielding covers, and exhaust and heat dissipation equipment. Data acquisition equipment is responsible for collecting various types of data in real time during the verification process. It is the carrier of data flow, transmitting data such as the output of the verification equipment and environmental status to the system. Examples include smart sensors, data acquisition terminals, and industrial IoT gateways.

[0021] In this embodiment, the optimal working sequence is one of the outputs of the equipment collaboration model. It is the planned start / run / stop time sequence for the power transmission line metering and testing equipment, environmental control equipment, and data acquisition equipment. It clarifies the start interval and runtime allocation between the equipment, avoiding energy consumption peaks or process delays caused by simultaneous equipment startup, and achieving optimization in the time dimension. The task allocation scheme is another output of the equipment collaboration model. It refers to the specific work content assigned to the power transmission line metering and testing equipment, environmental control equipment, and data acquisition equipment in the spatial and functional dimensions. For example, it specifies that a certain testing equipment is responsible for the testing of 110kV instrument transformers, a certain data acquisition equipment is responsible for tracking the data of the testing equipment, and a certain environmental control equipment is responsible for the temperature and humidity regulation of the area, achieving precise matching in the task dimension.

[0022] S103: Based on the optimal work sequence and task allocation scheme, execute the power metering verification and testing task to obtain the actual process execution data.

[0023] In this embodiment, the power metering verification and testing task is a complete process of compliance testing of the metering performance of power metering instruments (electricity meters, transformers, etc.), including visual inspection, accuracy calibration, functional verification, and data recording, which must comply with standards such as the "Power Metering Verification Regulations". The actual process execution data is the real operating data collected during the task execution, which is different from the predicted data. It includes equipment operating parameters, such as the output voltage and operating temperature of the verification equipment, task time (time for verifying a single electricity meter), process results (error detection values), and abnormal records (such as equipment lag time).

[0024] S104: Compare and analyze the actual process execution data with the predicted process effect data corresponding to the optimal work sequence and task allocation scheme to obtain the effect deviation and bottleneck links.

[0025] In this embodiment, the predicted process effect data corresponding to the optimal working sequence and task allocation scheme is the expected effect data generated synchronously by the equipment collaboration model when outputting the optimal scheme. It is a pre-judgment of task execution and corresponds one-to-one with the actual data. For example, it predicts that the verification time of a single electricity meter will be 4 minutes, the equipment will have no abnormalities, and the error value will be ≤0.03%. The comparative analysis is a process of multi-dimensional matching and verification between the actual process execution data and the predicted process effect data, including numerical difference calculation (e.g., time difference), pass rate comparison (e.g., percentage of qualified electricity meters), and anomaly type matching (e.g., no anomalies predicted but actual lag occurs).

[0026] In this embodiment, the performance deviation is the difference result obtained after comparative analysis, representing the degree of deviation between the actual execution effect and the predicted target. It is divided into positive deviation (actual performance is better than predicted, e.g., the time taken is 3.8 minutes less than the predicted 4 minutes) and negative deviation (actual performance is worse than predicted, e.g., the error is 0.05% greater than the predicted 0.03%). Whether optimization is needed needs to be determined based on a preset deviation threshold. The preset deviation threshold is set with reference to the "JJG596-2012 Verification Procedure for Electronic Energy Meters," and is adjusted quarterly based on the equipment aging rate (aging rate = number of failures / running time). The bottleneck is the process blockage point traced back to by the negative deviation; it is the specific link that causes the execution effect to fall short of expectations, including: equipment problems (e.g., verification equipment malfunction), personnel problems (e.g., excessively long material loading time), and timing problems (e.g., unreasonable equipment start-up intervals). For the identified negative deviation, a method based on SHAP value analysis can be used to trace and quantify the contribution of each process link (equipment preheating, parameter calibration, data verification) to the overall deviation, and the link with the highest contribution is identified as the bottleneck.

[0027] S105: In response to the effect deviation being greater than the preset deviation threshold, the parameters of the preset device collaboration model are adjusted through the optimization algorithm to obtain the optimized model parameters.

[0028] In this embodiment, the preset deviation threshold is a critical value for deviation set in advance based on power metering verification standards and business efficiency requirements. It serves as the basis for determining whether model optimization is needed. For example, if the deviation in verification time for a single device is greater than or equal to 0.3 minutes, or the meter error deviation is greater than or equal to 0.02%, parameter adjustment is triggered. The optimization algorithm is a method used to adjust the parameters of the equipment collaborative model. In this embodiment, the particle swarm optimization algorithm is used in the power metering scenario. It finds the optimal parameter combination that makes the model output more realistic through iterative calculation, thereby improving the prediction accuracy of the scheme.

[0029] In this embodiment, model parameters are configuration items of the device collaboration model, such as device performance weight coefficients (accuracy weight of verification equipment, response speed weight of environmental equipment), timing planning coefficients (device start-up interval coefficient, task switching time coefficient), etc. The parameter values ​​determine the calculation logic and output results of the device collaboration model. The optimized model parameters are new parameter combinations adjusted by optimization algorithms, making the prediction effect of the device collaboration model closer to the actual execution situation, which can improve the accuracy of the scheme and reduce deviation.

[0030] S106: Update the preset device collaboration model based on the optimized model parameters, and generate process optimization strategies for bottleneck links.

[0031] In this embodiment, the optimized model parameters are the equipment coordination model configuration items obtained by adjusting the optimization algorithm (particle swarm optimization algorithm). These parameters represent a more accurate configuration that makes the equipment coordination model's predictions more realistic. For example, increasing the weight of equipment maintenance status and adjusting the influence coefficients of temperature and humidity directly determines the rationality of the equipment coordination model's output scheme. The preset equipment coordination model is a decision-making model initially used to generate the optimal work sequence and task allocation scheme. After parameter optimization, iterative upgrades are completed by updating the parameters, enabling subsequent output scheduling schemes to avoid historical deviations. The update process involves replacing the original parameters of the equipment coordination model with the optimized parameters. This corrects the model's calculation logic, allowing the model to focus more on historical bottleneck-related influencing factors, such as equipment maintenance status and environmental fluctuations, when receiving equipment and environmental data later, thereby improving the accuracy of scheme predictions.

[0032] In this embodiment, the optimization algorithm is the particle swarm optimization algorithm, which is an intelligent optimization algorithm that simulates the group behavior of birds foraging. It iteratively searches the solution space by particles (representing potential parameter solutions), adjusts the flight direction and speed according to its own optimal solution and the group optimal solution, and finally finds the globally optimal parameter combination, which is suitable for the parameter optimization requirements of the equipment collaborative model.

[0033] The specific configuration of the particle swarm optimization algorithm is as follows: Particle encoding: Real number encoding is used, and each particle corresponds to a set of parameters of the equipment collaboration model (such as equipment performance weights and time-series planning coefficients). The parameter values ​​are limited to a physically reasonable range (such as temperature and humidity influence coefficients of 0.1-0.3). Fitness function: defined as fitness value = 1 / (1 + effect bias), where the effect bias is the weighted sum of time deviation and accuracy deviation (accuracy deviation weight 0.7, time deviation weight 0.3). The larger the fitness value, the better the parameters. Hyperparameter combination: including individual learning factors and social learning factors, the optimal combination is searched using Bayesian optimization method, the search range is limited to 1.5-2.5 (based on power scenario algorithm tuning experience), the number of search iterations is greater than or equal to 50, and the optimization objective is to minimize the effect deviation.

[0034] In this embodiment, the bottleneck is the specific obstruction point that causes process deviation, such as sensor delay in device 3, dust accumulation on the cooling fan of device 2, or lack of pre-processing of the meter wiring terminals. These are the targets for process optimization. The process optimization strategy is a feasible improvement plan for the bottleneck. It needs to identify the causes of the bottleneck (equipment problems / process deficiencies / personnel operation), clarify the improvement goals, specific actions, responsible parties, and verification standards, so that the bottleneck problem can be solved.

[0035] As can be seen from the above, this application can achieve efficient collaborative work of equipment by acquiring multi-source data under power transmission line metering verification and using the equipment collaboration model to obtain the optimal working sequence and task allocation scheme; by executing the testing task and comparing and analyzing the actual and predicted process effect data, it can identify effect deviations and bottleneck links; by adjusting the parameters of the equipment collaboration model and updating the model according to the effect deviations, it can generate process optimization strategies, thereby improving the power metering verification and testing service capabilities.

[0036] In one embodiment of this application, before adjusting the parameters of a preset device collaboration model using an optimization algorithm to obtain optimized model parameters in response to an effect deviation exceeding a preset deviation threshold, the method further includes: Accuracy testing and condition assessment of power transmission line metering and verification equipment based on equipment operation data; The control and verification equipment performs standard measurements on the built-in reference source and generates standard measurement values. The standard measurement value is compared with the known standard value of the built-in reference source to obtain the equipment accuracy deviation; If the equipment accuracy deviation is greater than the preset equipment accuracy threshold, it is determined that the detection equipment is inaccurate, an equipment calibration alarm is generated and the equipment calibration process is triggered. If the equipment accuracy deviation is less than the preset equipment accuracy threshold, a stability assessment is performed to obtain the overall equipment health score. If the overall health score of the equipment is less than the preset health threshold, a warning of equipment performance degradation will be generated.

[0037] In this embodiment, the equipment operation data refers to the data collected during the operation of the verification and testing equipment, such as output voltage, operating temperature, response time, and historical fault records, which serve as the basis for equipment evaluation. Transmission line power metering verification and testing equipment is specialized equipment used for performance testing of transmission line power metering instruments (meters, transformers, etc.), such as the JDJ-3 verification device and three-phase energy meter calibrator. Accuracy testing and condition assessment are systematic verifications of the performance (measurement accuracy) and overall operating status (stability, health) of the verification and testing equipment, and are prerequisites for ensuring the reliability of verification results.

[0038] In this embodiment, the built-in reference source is a standard reference unit integrated within the calibration equipment, with known and stable values, such as a high-precision voltage reference source or current reference source, used to calibrate the equipment's own measurement accuracy. The standard measured value is the result generated by the calibration equipment after measuring the built-in reference source; it needs to be determined through multiple repeated measurements and error analysis, representing the current actual measurement capability of the calibration equipment. The known standard value is the standard value calibrated at the factory by the built-in reference source and certified by an authoritative institution. For example, the known standard value of a 10V voltage reference source is 10.0000V, serving as a benchmark for judging the accuracy of the calibration equipment.

[0039] In this embodiment, the equipment accuracy deviation is the difference between the standard measured value and the known standard value. For example, the deviation of a measured value of 10.0002V from the standard value of 10.0000V is 0.0002V, indicating the accuracy of the measurement by the testing equipment. The preset equipment accuracy threshold is a critical value for equipment accuracy set according to the verification standard, such as a voltage measurement accuracy threshold of ±0.0003V, used to determine whether the testing equipment is inaccurate. Equipment calibration alarms are prompt signals triggered when the testing equipment is inaccurate, such as audible and visual alarms or system pop-ups, used to notify maintenance personnel to handle the issue. The equipment calibration process is a standardized operation to restore the accuracy of inaccurate equipment, such as adjusting the internal potentiometers of the testing equipment, updating calibration coefficients, and replacing aging components; it must be performed by qualified personnel according to procedures.

[0040] In this embodiment, stability assessment is an evaluation of the long-term operating status of the calibration and testing equipment. It determines whether the performance of the calibration and testing equipment drifts over time. Common indicators include operating temperature fluctuations, audio anomalies, and changes in response speed. The comprehensive equipment health score is a quantitative value of equipment health calculated using multi-dimensional indicators (accuracy, stability, historical faults, and maintenance records), for example, a score of 0-100, which comprehensively represents the operating status of the calibration and testing equipment.

[0041] In this embodiment, the preset health threshold is a critical value used to determine whether the testing equipment needs an early warning, such as 60 points. A score below the preset health threshold indicates a risk of equipment performance degradation. Equipment performance degradation warnings are alerts generated when the testing equipment's health fails to meet standards, such as system messages or SMS notifications, to proactively prevent equipment failure.

[0042] As can be seen from the above, this embodiment acquires multi-source data under power transmission line metering verification, obtains the optimal working sequence and task allocation scheme of the equipment based on equipment operation data and environmental parameters, and executes the testing task to obtain actual process execution data. Comparison and analysis with predicted process effect data reveals effect deviations and bottlenecks. Based on this, accuracy testing and status assessment are performed on the power transmission line metering verification and testing equipment. This accurately determines whether the verification and testing equipment is inaccurate, promptly generates calibration alarms and triggers the calibration process, ensuring equipment accuracy. Furthermore, stability assessment of the verification and testing equipment yields a comprehensive health score, enabling early detection of equipment performance degradation and issuing warnings, preventing equipment failures from affecting testing tasks and improving the power metering verification and testing service capabilities.

[0043] In one embodiment of this application, controlling the calibration and testing equipment to perform standard measurements on a built-in reference source and generate standard measurement values ​​includes: Under conditions where the calibration and testing equipment is idle and the environmental parameters are stable, at least two built-in reference sources with different measurement levels are repeatedly measured to obtain the measurement results. For multiple repeated measurements at each measurement level, the average value and standard deviation are calculated respectively, and the overall uncertainty is synthesized based on the principle of measurement uncertainty propagation. Standard measurements are generated based on the average, standard deviation, and overall uncertainty of all measurement levels.

[0044] In this embodiment, the built-in reference source is a standard component integrated within the verification and testing equipment, with stable values ​​and authoritative calibration. It provides a reference for the measurement accuracy of the verification and testing equipment itself. Common types include voltage reference sources and current reference sources, each corresponding to a fixed known standard value. The standard measurement value is an authoritative measurement result obtained by the verification and testing equipment after measuring the built-in reference source and processing the data. It serves as the basis for judging the accuracy of the verification and testing equipment itself, and it is necessary to balance measurement accuracy and reliability.

[0045] In this embodiment, the idle state refers to a state where the verification and testing equipment is not performing verification tasks, has no load input, and its core components are in a stable standby state. At this time, the verification and testing equipment experiences no additional interference, resulting in more accurate measurement results. Stable environmental parameters mean that environmental conditions affecting verification accuracy (temperature, humidity, voltage, electromagnetic interference, etc.) are within a preset range without drastic fluctuations, such as a temperature of 25℃±0.5℃ and humidity of 45%±5%, avoiding measurement errors introduced by environmental factors. Different measurement levels refer to different standard values ​​of the same type of reference source. For example, a voltage reference source can be set to a low value of 10V and a high value of 100V, covering the commonly used verification range of the verification and testing equipment, making the accuracy controllable across the entire range. Multiple repeated measurements involve performing at least three consecutive measurements on the same measurement level reference source. Multiple data points offset random errors, improving the reliability of the results. In power scenarios, 5-10 repeated measurements are used.

[0046] In this embodiment, the average value is the arithmetic mean of multiple repeated measurements, calculated as: sum of all measurements ÷ number of measurements. This is used to mitigate random errors in a single measurement. The standard deviation represents the dispersion of multiple measurement results from the average value. A smaller value indicates more concentrated measurement data and less random error, making it a key indicator for evaluating measurement stability. The measurement uncertainty propagation principle synthesizes the uncertainties of a single measurement (e.g., inherent equipment errors, environmental fluctuation errors) according to mathematical laws to obtain the overall uncertainty of the final measurement result, representing the confidence range of the measurement result. The overall uncertainty is an indicator of the degree of doubt surrounding the standard measurement value. For example, 10.0001V ± 0.0001V, where ± represents the overall uncertainty. A smaller value indicates a more reliable measurement result.

[0047] The standard measurement value generation method involves performing at least five repeated measurements on at least two built-in reference sources of different measurement levels under conditions where the verification and testing equipment is idle and environmental parameters are stable, to obtain measurement results; calculating the average value and standard deviation of the repeated measurement results for each measurement level, and synthesizing the overall uncertainty based on the principle of measurement uncertainty propagation; and generating standard measurement values ​​based on the average value, standard deviation, and overall uncertainty of all measurement levels.

[0048] As can be seen from the above, this embodiment improves the accuracy and reliability of measurement data by repeatedly measuring at least two built-in reference sources of different magnitudes when the calibration and testing equipment is idle and the environmental parameters are stable. Calculating the average value and standard deviation of the repeated measurement results for each magnitude level and synthesizing the overall uncertainty allows for a more scientific assessment of the accuracy and reliability of the measurement. Generating standard measurement values ​​based on the average value, standard deviation, and overall uncertainty yields more accurate and reliable standard measurement values, providing a more precise data foundation for equipment accuracy testing and condition assessment.

[0049] In one embodiment of this application, a stability assessment is performed to obtain a comprehensive device health score, including: The system collects audio signals and external infrared thermal imaging signals from the power metering and testing equipment for power transmission lines during operation. Extracting voiceprint features from audio signals and extracting temperature distribution and thermal anomaly features from infrared thermal imaging signals; The acoustic signature features, thermal anomaly features, and stability feature indicators extracted from equipment operation data are fused together to generate an equipment feature vector. The device feature vector is input into a preset multimodal health assessment model to obtain a comprehensive health score for the device.

[0050] In this embodiment, stability assessment is a comprehensive evaluation of the power transmission line metering and testing equipment from the perspective of the continuity and consistency of its operating status. It determines whether the equipment performance shows a trend of drift, aging, or other deterioration over time. The comprehensive equipment health score integrates multi-dimensional status characteristics of the equipment and outputs a comprehensive health status value ranging from 0 to 100 points. A higher score indicates more stable operation and a lower risk of failure for the testing equipment. The audio signal is the sound signal generated by the vibration and friction of various components (such as fans, gears, and circuit modules) during the operation of the testing equipment. The audio characteristics differ significantly between normal operation and fault conditions; for example, bearing wear will produce high-frequency noise.

[0051] In this embodiment, the external infrared thermal imaging signal is a temperature distribution image signal of the equipment surface and internal core components acquired by an infrared thermal imager. It can indicate whether the testing equipment has abnormalities such as localized overheating; for example, a short circuit in the power module will cause hot spots. Acoustic signature features are characteristic parameters extracted from audio signals that characterize the operating status of the testing equipment, such as the frequency distribution, amplitude variation, and spectral entropy of the audio signal. For example, the acoustic signature frequency of a normal fan is concentrated and stable, while a fault will result in frequency abrupt changes. Temperature distribution features are characteristics of the overall temperature field of the testing equipment in the infrared thermal imaging signal, such as average temperature, maximum temperature, and temperature gradient (temperature difference between different areas), used to determine whether the heat dissipation of the testing equipment is normal. Thermal anomaly features are localized features in the infrared thermal imaging that deviate from the normal temperature range of the testing equipment, such as the location of hot spots, the temperature difference between the hot spots and the surrounding area, and the area of ​​the thermal anomaly region. These are important early warning indicators for testing equipment faults.

[0052] In this embodiment, equipment operation data refers to the electrical parameters and status data collected during the operation of the testing equipment, such as output voltage / current fluctuations, response time, task completion efficiency, and historical fault records. This data forms the basis for extracting stability indicators. Stability characteristic indicators are those selected from the equipment operation data that represent operational stability, such as voltage fluctuation amplitude, response time coefficient of variation, and continuous fault-free operation duration. Multi-source feature fusion integrates features from three different sources—audio soundprints, infrared thermal features, and operational data features—into a unified feature set through mathematical algorithms (e.g., feature concatenation, weighted fusion), achieving complementary individual features and a comprehensive overall state representation.

[0053] In this embodiment, the device feature vector is a numerical vector formed by fusing multi-source features and used as input to a preset multimodal health assessment model. Each vector dimension corresponds to a feature parameter, such as [peak frequency, highest temperature, voltage fluctuation value, fault-free duration]. The preset multimodal health assessment model is a machine learning model capable of simultaneously processing three different modal inputs: audio, infrared, and operational data. This embodiment uses a hybrid model based on CNN+LSTM, which learns the correlation between features and device health status through training and outputs a health score.

[0054] Specifically, this embodiment adopts a three-stage architecture: CNN feature extraction layer - LSTM temporal modeling layer - fully connected decision layer. Each layer transmits data unidirectionally and works collaboratively. The specific connection relationship is as follows: The data input layer receives standardized multi-dimensional monitoring data, reconstructs it into 4-dimensional values ​​(number of samples × time step × feature dimension × number of channels) according to the time window + spatial dimension, and inputs it to the CNN feature extraction layer; The CNN feature extraction layer captures spatial features in the data through convolutional kernels, such as the correlation between data from different sensors of the device. After dimensionality reduction by the pooling layer, it outputs a temporal feature matrix (number of samples × time step × LSTM temporal modeling layer). The stride (multiplied by the number of convolutional features) is passed to the LSTM temporal modeling layer. The LSTM temporal modeling layer expands the temporal feature matrix output by the CNN along the time dimension, captures long-term temporal dependencies (e.g., signal change trends before device malfunction) through gating units (input gate, forget gate, output gate), and outputs the temporal global features (number of samples × number of LSTM hidden layer features), which are then passed to the fully connected decision layer. The decision output layer maps the temporal global features to device status labels through a fully connected network, outputting the classification results and confidence scores for normal / mildly abnormal / severely abnormal conditions, and simultaneously outputting the fault warning time window. The multimodal health assessment model uses a CNN+LSTM hybrid model, with the following specific configuration: Training data: The sample size is greater than or equal to 5000 groups (including three states: normal equipment, slightly deteriorated equipment, and severely deteriorated equipment). The health score is calculated by weighting the accuracy deviation by 30% + stability by 20% + historical failure by 50%. Model hyperparameters: The CNN uses 3×3 convolutional kernels, with a number of 32 / 64; the LSTM has 128 hidden layer neurons. Validation metric: F1 score (harmonic average of precision and recall) is used, which must be greater than or equal to 0.9.

[0055] As can be seen from the above, this embodiment can obtain the operating information of the power transmission line metering verification and testing equipment from different dimensions by collecting audio signals and external infrared thermal imaging signals during operation; extracting acoustic features, temperature distribution, and thermal anomaly features enriches the characteristic information of the verification and testing equipment; fusing these features with stability characteristic indicators in the equipment operating data to generate equipment feature vectors can more comprehensively and accurately represent the status of the verification and testing equipment; inputting the equipment feature vectors into a preset multimodal health assessment model to obtain a comprehensive health score of the equipment can scientifically assess the stability of the equipment, provide a basis for determining whether the performance of the verification and testing equipment has deteriorated, help to discover potential problems of the verification and testing equipment in a timely manner, and ensure the accuracy and reliability of power metering verification and testing work.

[0056] In one embodiment of this application, the optimization algorithm is a particle swarm optimization algorithm; before adjusting the parameters of a preset device collaboration model to obtain optimized model parameters in response to an effect deviation greater than a preset deviation threshold, the method further includes: Based on the deviation between the actual process execution data and the predicted process effect data corresponding to the optimal work sequence and task allocation scheme, the hyperparameter combination of the particle swarm optimization algorithm is determined.

[0057] In this embodiment, the Particle Swarm Optimization (PSO) algorithm is an intelligent optimization algorithm that simulates the group behavior of birds foraging. It iteratively searches the solution space using particles (representing potential parameter solutions), adjusting their flight direction and speed based on their own optimal solution and the group's optimal solution, ultimately finding the globally optimal parameter combination. In the power metering scenario, its fast convergence speed and ease of implementation make it suitable for the parameter optimization needs of equipment collaborative models. For power metering verification scenarios, optimizing the parameters of the equipment collaborative model minimizes the deviation in the verification process, solving the problems of traditional parameter tuning relying on experience, slow convergence, and susceptibility to local optima.

[0058] Specifically, a closed-loop architecture of initialization-iterative search-convergence output is adopted, including a particle initialization module, a fitness calculation module, a velocity / position update module, and an optimal solution storage module. The connection relationship is as follows: The initialization module generates a particle population (parameter solution set), initializes particle positions (initial parameter values) and velocities (parameter adjustment ranges), and outputs to the velocity / position update module; the update module receives population data and individual / group optimal solutions from the optimal solution storage module, calculates new velocities and positions, and passes them to the fitness calculation module; the fitness module maps particle positions to equipment collaborative model parameters, calculates the effect deviation (fitness value) based on power metering scenario data, and feeds it back to the optimal solution storage module; the storage module compares the current fitness value with historical optimal solutions, updates and stores individual and group optimal solutions, and triggers iterative or convergence output.

[0059] In this embodiment, the preset deviation threshold is a pre-set critical value based on power metering business standards (such as the "Power Metering Verification Regulations") and operation and maintenance efficiency requirements. For example, if the deviation in verification time for a single device is greater than or equal to 0.3 minutes, or the deviation in meter error detection is greater than or equal to 0.02%, optimization needs to be initiated if the deviation exceeds the preset deviation threshold. The actual process execution data is real operational data collected from the verification site, including equipment operating parameters (such as verification equipment temperature and response time), task execution results (such as the number of qualified meters and the time spent handling anomalies), and personnel operation data (such as material loading speed). This data serves as the baseline for deviation calculation. The optimal work sequence and task allocation scheme is a scheduling scheme output by the equipment collaboration model. It includes the start / stop time nodes of each verification device and environmental control device, as well as the allocation rules for meter verification tasks among devices and personnel. For example, device 1 is responsible for meters 1-45. The predicted process effect data is the expected result data synchronously output when the equipment collaborative model generates the scheduling scheme. It is completely matched with the actual data dimensions. For example, it predicts that the verification time of a single electricity meter will be 4 minutes, the equipment will have no abnormalities, and the error pass rate will be 100%. It serves as a reference benchmark for deviation comparison.

[0060] In this embodiment, the hyperparameter combination refers to the configuration parameters of the particle swarm optimization algorithm itself. These parameters are not automatically adjusted with algorithm iterations and need to be set in advance. They include individual learning factors and social learning factors, and their combination directly affects the algorithm's search efficiency and optimization accuracy. Determining the hyperparameter combination involves adjusting the hyperparameter values ​​of the particle swarm optimization algorithm based on the type, magnitude, and distribution characteristics of the performance deviation, so that the algorithm adapts to the current deviation optimization requirements. For example, when the deviation is small, the learning factor is reduced to stabilize convergence.

[0061] As can be seen from the above, this embodiment first acquires multi-source data including equipment operation data, environmental parameters, and testing process data under power transmission line metering verification. Based on the equipment operation data and environmental parameters, the optimal working sequence and task allocation scheme are obtained through a preset equipment collaboration model. The actual process execution data is obtained by executing the tasks. The effect deviation and bottleneck links are obtained by comparing the actual and predicted process effect data. When the effect deviation is greater than the preset deviation threshold, the hyperparameter combination of the particle swarm optimization algorithm determined according to the effect deviation between the actual and predicted process effect data is used to adjust the parameters of the preset equipment collaboration model, making the parameter adjustment more targeted. This allows for more accurate updating of the equipment collaboration model and generation of process optimization strategies for bottleneck links, effectively improving the power metering verification and testing service capabilities.

[0062] In one embodiment of this application, the hyperparameter combination includes individual learning factors and social learning factors; Based on the performance deviation between actual process execution data and predicted process performance data corresponding to the optimal work sequence and task allocation scheme, the hyperparameter combination of the particle swarm optimization algorithm is determined, including: With minimizing effect bias as the optimization objective, a Bayesian optimization method is used to search for and determine the optimal combination of individual learning factors and social learning factors.

[0063] In this embodiment, the individual learning factor is a parameter in the particle swarm optimization algorithm that guides particles to review their own historical best solutions, with a value ranging from 0 to 4. A larger value indicates that the particle relies more on its past experience, making it easier to discover local optima; a smaller value indicates that the particle lacks autonomous search ability and misses precise parameters. The social learning factor is a parameter that guides particles towards the global optimum of the swarm, with a value range similar to the individual learning factor. A larger value indicates faster particle convergence, making it easier to quickly lock onto the global trend; a larger value indicates that the particle blindly follows the swarm, ignoring better local solutions, leading to insufficient optimization accuracy.

[0064] In this embodiment, minimizing the effect deviation is the optimization objective of the particle swarm optimization algorithm. That is, by adjusting the hyperparameter combination, the actual deviation of the verification process after the equipment collaboration model parameters are optimized is made infinitely close to 0, making the solution output by the equipment collaboration model more closely match the actual situation on site. Bayesian optimization is a highly efficient optimization algorithm based on a probabilistic model. By continuously learning the mapping relationship between the deviation optimization effect of hyperparameter combinations, it constructs a probabilistic model (e.g., a Gaussian process model) to predict and find the hyperparameter region of the optimal solution, reducing blind searches. It is suitable for the needs of power metering scenarios where hyperparameter influences are complex and trial-and-error costs are high.

[0065] Specifically, the Bayesian optimization algorithm adopts a closed-loop architecture of data input - probabilistic modeling - optimization decision - effect feedback, including four units: data preprocessing module, Gaussian process model module, data acquisition function module, and hyperparameter output module. The modules have clear hierarchical structure and unidirectional dependencies, and the specific connection relationships are as follows: The data input layer consists of a data preprocessing module that receives historical and real-time deviation data from the power metering scenario, performs data cleaning and feature extraction, and outputs a standardized hyperparameter combination-effect deviation sample set, which is then passed to the Gaussian process model module. The probabilistic modeling layer involves the Gaussian process model module constructing a probabilistic model based on the sample set, outputting the predicted effect deviation value and prediction uncertainty corresponding to the hyperparameter combination, and synchronously passing it to the acquisition function module. The optimization decision layer involves the acquisition function module combining the predicted value and uncertainty to calculate the acquired values ​​for each hyperparameter combination, selecting the most valuable hyperparameter combination, and passing it to the hyperparameter output module. The effect feedback layer involves the hyperparameter output module sending the optimal hyperparameter combination to the particle swarm optimization algorithm. After the particle swarm optimization algorithm completes the optimization of the equipment collaborative model parameters, it collects new effect deviation data and feeds it back to the data preprocessing module, achieving iterative optimization.

[0066] As can be seen from the above, this embodiment acquires multi-source data, including equipment operation data, environmental parameters, and testing process data, under the power transmission line metering verification. Based on the equipment operation data and environmental parameters, it obtains the optimal working sequence and task allocation scheme through a preset equipment collaboration model. It executes the power metering verification and testing tasks to obtain actual process execution data. By comparing and analyzing the data, it obtains the effect deviation and bottleneck links. When the optimization algorithm is particle swarm optimization, it uses Bayesian optimization method or regression model trained based on historical optimization data to search and determine the optimal combination of individual learning factors and social learning factors, based on the effect deviation as the objective. This allows for more accurate parameter adjustment of the preset equipment collaboration model, resulting in better model parameters. Consequently, the equipment collaboration model is updated, and a more effective process optimization strategy is generated, thereby improving the power metering verification and testing service capabilities.

[0067] In one embodiment of this application, it further includes: Establish personnel skill profiles, which include personnel's basic skill level, types of historical operational errors, and degree of mastery of specialized skills; Match and analyze the technical requirements of bottleneck links with personnel skill profiles to identify skill gaps; Based on skills gaps, corresponding training courses and practical training content are pushed from the training resource library; Update personnel skill profiles and training content based on performance data from simulation training.

[0068] In this embodiment, the personnel skills profile is a standardized digital skills file constructed based on the practical data of power metering verification personnel. It serves as the basis for accurately matching tasks and conducting training, and its dimensions include basic capabilities, historical performance, and specialized strengths. The basic skills level indicates the personnel's mastery of basic verification operations, such as beginner / intermediate / advanced, corresponding to abilities including meter wiring, basic equipment operation, and data recording standards, determined by qualification examinations and daily assessments. Historical operational error types record specific errors made by personnel in their past work, such as wiring errors, parameter setting deviations, and missed abnormal data, used to pinpoint skill weaknesses; for example, a high-frequency wiring error indicates insufficient practical proficiency.

[0069] In this embodiment, the level of specialized skill mastery refers to the personnel's ability level in a specific verification scenario, such as high-voltage meter verification, instrument transformer calibration, and troubleshooting of abnormal equipment, expressed as mastery / proficiency / expertise or a quantitative score. The technical requirements of bottleneck links are the specific skills required at the process blockage points; for example, sensor calibration for equipment 3 requires high-precision instrument operation skills, and meter preprocessing requires terminal oxidation identification capabilities. Skill gaps are the differences between the personnel's existing skill profile and the technical requirements of bottleneck links; for example, a basic skill level of intermediate, but lacking specialized sensor calibration skills. The training resource library is a collection of standardized training content in the field of power metering, including theoretical courses (e.g., "Verification Procedures"), practical tutorials (e.g., sensor calibration videos), and simulation training systems (virtual verification scenarios). Simulation training performance data are the personnel's operational data in virtual / semi-physical simulation scenarios, such as calibration time, operational accuracy, and number of errors, serving as the basis for evaluating training effectiveness.

[0070] As can be seen from the above, this embodiment can comprehensively understand the skill status of personnel by establishing personnel skill profiles, accurately identify skill gaps by matching and analyzing the technical requirements of bottleneck links with personnel skill profiles, improve personnel skills by pushing corresponding training courses and practical training content based on skill gaps, and make training more in line with the actual skill improvement needs of personnel by updating personnel skill profiles and training content based on simulation training performance data, thereby improving personnel capabilities and enhancing the power metering verification and testing service capabilities.

[0071] In one embodiment of this application, based on equipment operating data and environmental parameters, an optimal working sequence and task allocation scheme for the power transmission line metering and verification testing equipment, environmental control equipment, and data acquisition equipment is obtained through a preset equipment coordination model. The scheme further includes: The evaluation results were correlated with the effect deviation, and statistical methods were used to calculate the correlation strength between personnel factors and effect deviation. The correlation strength, equipment operation data, and environmental parameters are input into a preset equipment collaboration model to ensure that the generated solution takes into account personnel capability factors.

[0072] In this embodiment, correlation analysis uses statistical methods to uncover the causal relationship between evaluation results (personnel / equipment status) and performance deviations, determining whether insufficient personnel capability is the primary cause of the deviation. Personnel factors are personnel-related variables affecting process effectiveness, such as operational proficiency, skill gaps, and historical error rates. The correlation strength value, ranging from 0 to 1, represents the degree of correlation between personnel factors and performance deviations; the closer to 1, the stronger the correlation. For example, a correlation strength value of 0.85 between wiring errors and meter error deviations indicates that personnel operation is the primary cause of the deviation.

[0073] In this embodiment, equipment operating data refers to the real-time operating parameters of the calibration equipment, such as temperature 38℃, response time 0.6 seconds, and historical fault records, representing the objective state of the equipment. Environmental parameters refer to environmental conditions affecting calibration accuracy, such as temperature and humidity at the calibration site (25℃±0.5℃), voltage stability, and electromagnetic interference intensity. The optimal work sequence and task allocation scheme is the output full-process scheduling plan, including equipment start-up and shutdown times (e.g., environmental equipment starts at 8:00), task allocation (e.g., Zhang San is responsible for loading equipment 1), and the upgraded plan also considers personnel capability matching.

[0074] As can be seen from the above, this embodiment performs correlation analysis on the evaluation results and effect deviations and calculates the correlation strength value between personnel factors and effect deviations. The correlation strength, along with equipment operation data and environmental parameters, is input into a preset equipment collaboration model, so that the generated optimal work sequence and task allocation scheme take into account personnel ability factors, thereby improving the power metering verification and testing service capabilities.

[0075] In one embodiment of this application, a correlation analysis is performed between the evaluation results and the effect deviation, and a statistical method is used to calculate the correlation strength value between personnel factors and the effect deviation, including: Construct a multiple linear regression model with the dimensional indicators of the personnel competence assessment matrix as independent variables and the effect deviation as the dependent variable; the dimensional indicators should at least include operation speed, error rate, emergency response time, and emergency response success rate. Historical task data was collected as training samples to train the multiple linear regression model and obtain the regression coefficients corresponding to each dimension index. Based on the magnitude and sign of the regression coefficients, the direction and intensity of the influence of each personnel's ability dimension on the effect deviation are determined, and the overall correlation strength value of personnel factors is calculated.

[0076] In this embodiment, the multiple linear regression model is a statistical model that analyzes the linear relationship between multiple independent variables and a dependent variable. It is used to explore the correlation between personnel indicators such as operation speed and error rate and performance deviation. The formula is: Performance Deviation = Constant Term + Operation Speed ​​× First Regression Coefficient + Error Rate × Second Regression Coefficient + ... + Error Term. The personnel capability assessment matrix is ​​a tabular tool that breaks down personnel capabilities into multiple quantifiable dimensions. These dimensions include operation speed (time spent loading and wiring a single meter), error rate (number of operational errors / total number of tasks), emergency response time (time from equipment malfunction to the start of handling), and emergency handling success rate (number of successfully resolved malfunctions / total number of malfunctions). It can also be expanded to include indicators such as verification accuracy compliance rate.

[0077] In this embodiment, the independent variables are those that influence the dependent variable in the multiple linear regression model; in this case, they are the dimensional indicators (operation speed, error rate, etc.) of the personnel competence assessment matrix, and are the causal variables in the analysis. The dependent variable is the variable that is affected; in this case, it is the effect bias, such as time consumption bias or accuracy bias, and is the outcome variable in the analysis.

[0078] In this embodiment, historical task data refers to the records of meter calibration tasks previously completed by the calibration center. This data must include both personnel operation indicator data and corresponding task performance deviation data, serving as the training sample basis. For example, calibration data from 1000 meters over the past 12 months. Training samples are valid data selected from the historical task data. For example, 800 sets of data after removing interfering data such as environmental anomalies and equipment malfunctions are divided into a training set (640 sets for training) and a validation set (160 sets for verifying accuracy).

[0079] In this embodiment, the regression coefficient is the output of the multiple linear regression model, representing the strength and direction of the influence of a single independent variable on the dependent variable. The larger the absolute value of the coefficient, the stronger the influence. A positive coefficient indicates that the dependent variable also increases when the independent variable increases, for example, the higher the error rate, the greater the accuracy deviation. A negative coefficient indicates that the dependent variable decreases when the independent variable increases, for example, the higher the success rate of emergency response, the smaller the time deviation.

[0080] As can be seen from the above, this embodiment obtains multi-source data under power transmission line metering verification, obtains the optimal working sequence and task allocation scheme based on equipment operation data and environmental parameters, and executes the testing task. By comparing the actual and predicted process effect data, the effect deviation and bottleneck links are obtained. A multiple linear regression model is constructed with the personnel capability assessment matrix dimension indicators as independent variables and the effect deviation as the dependent variable. The regression coefficients are obtained by training the model using historical task data. The influence direction and intensity of each personnel capability dimension on the effect deviation are determined and the correlation strength value is calculated. This allows the generated scheme to take into account personnel capability factors, thereby more accurately considering the impact of personnel factors on the power metering verification and testing service effect and improving service capabilities.

[0081] In one embodiment of this application, it further includes: The device collaboration model updated based on the optimized model parameters is used as the local model and uploaded to the federated learning server. It also receives new global model parameters issued by the server after securely aggregating the local model updates uploaded from multiple nodes. The device collaboration model is updated based on the new global model parameters, and process optimization strategies for bottleneck links are generated.

[0082] In this embodiment, the local model is a device collaboration model optimized by a specific node in the power metering verification process (e.g., the verification center in City A) based on its own data. It only includes the model structure and parameters, without disclosing original business data such as local meter verification records and equipment operation data, thus ensuring data privacy. The federated learning server is the hub that coordinates the training of multiple node models. Its function is to receive local models from each node, securely aggregate parameters, and then distribute the global model without touching the original data of each node, resolving privacy and compliance issues related to cross-regional data sharing. A node is an independent power metering unit participating in federated learning, which can be a city-level verification center, a county-level maintenance station, etc. Each node has its own device collaboration model and business data, serving as the data and computational core for model training. Local model updates refer to the parameter changes generated after parameter optimization of the node's local model, such as the difference between adjusting the equipment maintenance weight from 0.2 to 0.5, rather than the complete model, thus reducing data transmission volume.

[0083] In this embodiment, the global model parameters are unified parameters generated by the server after aggregating the updates of multiple node models. They incorporate the business scenario experience of different nodes, such as the equipment aging scenario in City A and the high humidity environment scenario in City B, making them more versatile and robust than a single node local model.

[0084] In this embodiment, the bottleneck process optimization strategy is an improvement plan for process blockage points (such as excessively long equipment calibration time or omissions in meter preprocessing). The updated model generation strategy incorporates multi-node experience and is more universal.

[0085] As can be seen from the above, this embodiment uploads the updated device collaboration model as a local model to the federated learning server, receives the new global model parameters issued by the server and updates the device collaboration model, and can use multi-node data to optimize the model, improve the model's accuracy and generalization ability, thereby generating more effective bottleneck process optimization strategies and improving the power metering verification and testing service capabilities.

[0086] In one embodiment of this application, it further includes: Monitor the quality and contribution of local model updates uploaded by each node; quality is evaluated by the accuracy improvement rate of the local model on the validation set, and contribution is evaluated by the cosine similarity between the local model update and the global model parameters and the amount of data of the node. In response to any of the following triggering conditions, the key parameters of federated learning are dynamically adjusted using a federated optimization algorithm: Condition 1: In multiple consecutive training cycles, the performance improvement rate of the global model on the public test set is lower than a preset threshold; Condition 2: If the difference coefficient of model update contribution between nodes is found to be greater than the preset difference threshold, it indicates that the distribution of node data shows significant non-independent and identically distributed characteristics; Condition 3: The new node joins the federated learning network; Key parameters for federated learning include at least the aggregation weight allocation strategy, the number of local training rounds, and the sparsity of model pruning; Federated optimization algorithms are built upon reinforcement learning. They take the state of the federated learning environment as input and output adjustments to key parameters to maximize the long-term performance gains of the global model on a pre-defined validation set.

[0087] In this embodiment, quality is an indicator of the effectiveness of local model updates, evaluated by the accuracy improvement rate on the validation set, i.e., the improvement in accuracy of the updated local model on an independent validation set (e.g., calibration data from 100 standard electricity meters) compared to before the update; a higher improvement rate indicates better quality. Contribution is an indicator of the value of node model updates to the global model, based on two dimensions: first, cosine similarity, the degree of similarity between the local model update and the current global model parameters; moderate similarity indicates strong contribution uniqueness; second, the amount of node data; the larger the data volume, the stronger the sample representativeness, and the higher the contribution weight. Cosine similarity quantifies the directional consistency of two vectors (local model parameter vector and global model parameter vector), taking a value between 0 and 1 in federated learning. Values ​​that are too close indicate duplicate updates, while values ​​that are too far apart indicate abnormal updates, both of which reduce contribution.

[0088] In this embodiment, the training cycle is a complete iterative process of federated learning. Each node uploads the model → the server aggregates the data → global parameters are distributed, and N rounds of local training constitute one cycle. For example, 5 rounds of local optimization by a node in City A constitute one cycle. The public test set is a standard dataset shared by the federated learning consortium and free of privacy information, such as meter verification data calibrated by an authoritative institution. It is used to uniformly evaluate the performance of the global model and avoid bias caused by nodes using local data for evaluation. The performance improvement rate is the percentage difference in accuracy of the global model on the public test set between two adjacent training cycles. For example, if the accuracy is 92% in cycle 1 and 93% in cycle 2, the improvement rate is 1.09%, indicating the iterative effect of the global model. The contribution difference coefficient is a statistical indicator representing the dispersion of the contribution of each node. A larger value indicates that some nodes contribute too much and some contribute too little, indicating that the node data distribution is not independent. For example, City A has all new meter data, and City B has all old meter data. The non-independent and identically distributed characteristic means that the data distribution of each node is significantly different, such as different scenarios, device types, and error types. This is a common challenge in federated learning, causing the global model to be biased towards nodes with large amounts of data.

[0089] In this embodiment, the federated optimization algorithm is an intelligent algorithm used to dynamically adjust key parameters of federated learning. This scenario is based on reinforcement learning, and its core is to perceive the state of the federated environment → output parameter adjustment actions → pursue optimal long-term performance.

[0090] The federated optimization algorithm in this embodiment is an intelligent decision-making algorithm based on a reinforcement learning framework, used to dynamically adjust key parameters of a federated learning system for power metering verification. The goal is to maximize the long-term performance gains of the global model in power metering verification tasks by sensing the real-time state of the federated learning environment, such as global model accuracy and node contribution distribution, and outputting optimal adjustments to aggregate weight allocation, local training rounds, and model pruning sparsity. This results in improvements in meter error detection accuracy and equipment scheduling efficiency, while simultaneously addressing issues in federated learning such as non-independent and identically distributed node data, imbalanced contributions of some nodes, and difficulties in adapting new nodes. The algorithm comprises four core units: a state awareness module, a decision output module, a reward calculation module, and a parameter memory module. The module hierarchy is: data input layer → feature processing layer → decision execution layer → feedback optimization layer, with the specific connections as follows: The data input layer consists of a state-aware module that receives raw data (global model accuracy, node contribution, etc.) from the federated learning system, providing raw input for feature processing. The feature processing layer involves the state-aware module standardizing and dimensionally fusing the raw data, outputting a structured federated environment state vector, which is then passed to the decision output module. The decision execution layer involves the decision output module generating parameters and adjusting actions based on the state vector, which are then sent to the federated learning server for execution and to the reward calculation module. The feedback optimization layer involves the reward calculation module combining the adjusted global model performance changes to generate reward values, which are then synchronously fed back to the parameter memory module and the decision output module to achieve iterative optimization of algorithm parameters.

[0091] The state of a federated learning environment is a set of metrics describing the current operation of the federated learning process, such as global model accuracy, contribution of each node, and differences in data distribution. Long-term performance gains are the goal of federated optimization algorithms; they do not aim for optimal performance in a single cycle, but rather to continuously improve the global model over multiple iterations, for example, increasing accuracy from 90% to 98% after 10 cycles.

[0092] As can be seen from the above, this embodiment can accurately evaluate the role of each node in federated learning by monitoring the quality and contribution of the local model updates uploaded by each node. When specific triggering conditions are met, the federated optimization algorithm based on reinforcement learning can dynamically adjust key parameters such as the aggregation weight allocation strategy, the number of local training rounds, and the sparsity of model pruning, which can maximize the long-term performance gains of the global model on the preset validation set and improve the performance and adaptability of the model in the method for improving the power metering verification and testing service capabilities.

[0093] Corresponding to the method for improving the power metering verification and testing service capabilities in the above embodiment, Figure 2This is a structural block diagram of a power metering verification and testing service capability enhancement system provided according to an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The power metering verification and testing service capability enhancement system 20 includes: a data acquisition module 21, a collaborative optimization module 22, a task execution module 23, an efficiency analysis module 24, a parameter adjustment module 25, and an update and optimization module 26.

[0094] Among them, the data acquisition module 21 is used to acquire multi-source data under the power transmission line metering verification; the multi-source data includes equipment operation data, environmental parameters and testing process data; The collaborative optimization module 22 is used to obtain the optimal working sequence and task allocation scheme of the power transmission line metering and testing equipment, environmental control equipment and data acquisition equipment based on equipment operation data and environmental parameters and through a preset equipment collaborative model. The task execution module 23 is used to execute power metering verification and testing tasks based on the optimal work sequence and task allocation scheme, and obtain actual process execution data. The performance analysis module 24 is used to compare and analyze the actual process execution data with the predicted process effect data corresponding to the optimal work sequence and task allocation scheme to obtain the effect deviation and bottleneck links. The parameter adjustment module 25 is used to adjust the parameters of the preset equipment collaboration model through an optimization algorithm in response to the effect deviation being greater than the preset deviation threshold, so as to obtain the optimized model parameters. The update and optimization module 26 is used to update the preset device collaboration model based on the optimized model parameters and generate process optimization strategies for bottleneck links.

[0095] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Figure 2 The functions of the data acquisition module 21, collaborative optimization module 22, task execution module 23, performance analysis module 24, parameter adjustment module 25, and update optimization module 26 are shown.

[0096] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0097] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0098] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0099] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation methods described in any embodiment of the power metering verification and testing service capability improvement method provided in the embodiments of this application, or they can execute the implementation methods of the electronic devices described in the embodiments of this application, which will not be repeated here.

[0100] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0101] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0102] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0103] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0104] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0105] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0106] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0107] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for improving the capability of electricity metering verification and testing services, characterized in that, include: Acquire multi-source data under the power metering verification of power transmission lines; The multi-source data includes equipment operation data, environmental parameters, and testing process data; Based on the equipment operation data and the environmental parameters, the optimal working sequence and task allocation scheme of the power transmission line metering and testing equipment, environmental control equipment and data acquisition equipment are obtained through a preset equipment collaboration model. Based on the optimal work sequence and task allocation scheme, the power metering verification and testing task is executed to obtain the actual process execution data; The actual process execution data is compared and analyzed with the predicted process effect data corresponding to the optimal work sequence and task allocation scheme to obtain the effect deviation and bottleneck links. In response to the effect deviation being greater than a preset deviation threshold, the parameters of the preset device collaboration model are adjusted using an optimization algorithm to obtain optimized model parameters; The preset device collaboration model is updated based on the optimized model parameters, and a process optimization strategy for the bottleneck link is generated.

2. The method for improving the capability of power metering verification and testing services according to claim 1, characterized in that, Before the step of responding to an effect deviation greater than a preset deviation threshold and adjusting the parameters of the preset device collaboration model using an optimization algorithm to obtain optimized model parameters, the method further includes: Based on the equipment operation data, the accuracy test and status assessment of the power transmission line metering and testing equipment are performed. The calibration and testing equipment is controlled to perform standard measurements on the built-in reference source and generate standard measurement values. The standard measurement value is compared with the known standard value of the built-in reference source to obtain the equipment accuracy deviation; If the accuracy deviation of the device is greater than the preset device accuracy threshold, it is determined that the detection device is inaccurate, a device calibration alarm is generated and the device calibration process is triggered. If the device accuracy deviation is less than the preset device accuracy threshold, a stability assessment is performed to obtain a comprehensive device health score. If the overall health score of the device is less than the preset health threshold, a device performance degradation warning will be generated.

3. The method for improving the capability of power metering verification and testing services according to claim 2, characterized in that, The control of the calibration and testing equipment to perform standard measurements on the built-in reference source and generate standard measurement values ​​includes: Under the condition that the calibration and testing equipment is idle and the environmental parameters are stable, the built-in reference sources with at least two different magnitude levels are repeatedly measured to obtain the measurement results. For multiple repeated measurements at each measurement level, the average value and standard deviation are calculated respectively, and the overall uncertainty is synthesized based on the principle of measurement uncertainty propagation. The standard measurement value is generated based on the average, standard deviation, and overall uncertainty of all measurement levels.

4. The method for improving the capability of power metering verification and testing services according to claim 2, characterized in that, The stability assessment is performed to obtain a comprehensive equipment health score, including: The audio signal and external infrared thermal imaging signal of the power transmission line metering and testing equipment during operation are collected; Extract voiceprint features from the audio signal, and extract temperature distribution and thermal anomaly features from the infrared thermal imaging signal; The acoustic signature features, thermal anomaly features, and stability feature indicators extracted from equipment operation data are fused using multi-source features to generate an equipment feature vector. The device feature vector is input into a preset multimodal health assessment model to obtain the device's overall health score.

5. The method for improving the capability of power metering verification and testing services according to claim 1, characterized in that, The optimization algorithm is a particle swarm optimization algorithm; before the step of adjusting the parameters of the preset device collaboration model through the optimization algorithm to obtain the optimized model parameters in response to the effect deviation being greater than a preset deviation threshold, the method further includes: Based on the effect deviation between the actual process execution data and the predicted process effect data corresponding to the optimal work sequence and task allocation scheme, the hyperparameter combination of the particle swarm optimization algorithm is determined.

6. The method for improving the capability of power metering verification and testing services according to claim 5, characterized in that, The hyperparameter combination includes individual learning factors and social learning factors; The step of determining the hyperparameter combination of the particle swarm optimization algorithm based on the effect deviation between the actual process execution data and the predicted process effect data corresponding to the optimal work sequence and task allocation scheme includes: With minimizing effect bias as the optimization objective, a Bayesian optimization method is used to search for and determine the optimal combination of the individual learning factor and the social learning factor.

7. The method for improving the capability of power metering verification and testing services according to claim 1, characterized in that, Also includes: Establish personnel skill profiles, which include personnel's basic skill level, types of historical operational errors, and degree of mastery of specialized skills; The technical requirements of the bottleneck links are matched and analyzed with personnel skill profiles to identify skill gaps; Based on the skill gaps, corresponding training courses and practical training content are pushed from the training resource library; The personnel skill profiles and training content are updated based on the performance data of personnel in the simulation training.

8. The method for improving the capability of power metering verification and testing services according to claim 7, characterized in that, The method of obtaining the optimal working sequence and task allocation scheme for the power transmission line metering and testing equipment, environmental control equipment, and data acquisition equipment based on the equipment operation data and environmental parameters through a preset equipment collaboration model also includes: The evaluation results were correlated with the effect deviation, and the correlation strength between personnel factors and effect deviation was calculated using statistical methods. The correlation strength, the equipment operation data, and the environmental parameters are input into the preset equipment collaboration model so that the generated solution takes into account personnel capability factors.

9. A method for improving the capability of power metering verification and testing services according to claim 8, characterized in that, The step of performing a correlation analysis between the evaluation results and the effect deviation, and using statistical methods to calculate the correlation strength value between personnel factors and effect deviation, includes: Construct a multiple linear regression model with the dimensional indicators of the personnel competence assessment matrix as independent variables and the effect deviation as the dependent variable; the dimensional indicators include at least operation speed, error rate, emergency response time, and emergency response success rate; Historical task data is collected as training samples to train the multiple linear regression model and obtain the regression coefficients corresponding to each dimension index. Based on the magnitude and sign of the regression coefficients, the direction and intensity of the influence of each personnel ability dimension on the effect deviation are determined, and the overall correlation strength value of the personnel factors is calculated.

10. A system for improving the capability of electricity metering verification and testing services, characterized in that, include: The data acquisition module is used to acquire multi-source data under the power transmission line metering verification. The multi-source data includes equipment operation data, environmental parameters, and testing process data; The collaborative optimization module is used to obtain the optimal working sequence and task allocation scheme of the power transmission line power metering verification and testing equipment, environmental control equipment and data acquisition equipment based on the equipment operation data and the environmental parameters through a preset equipment collaboration model; The task execution module is used to execute power metering verification and testing tasks based on the optimal working sequence and task allocation scheme, and obtain actual process execution data. The performance analysis module is used to compare and analyze the actual process execution data with the predicted process effect data corresponding to the optimal work sequence and task allocation scheme to obtain the effect deviation and bottleneck links. The parameter adjustment module is used to adjust the parameters of the preset device collaboration model through an optimization algorithm in response to the effect deviation being greater than a preset deviation threshold, so as to obtain the optimized model parameters. The update and optimization module is used to update the preset device collaboration model based on the optimized model parameters and generate a process optimization strategy for the bottleneck link.

Images