A method and system for automatic testing and management of electric energy meters and environments

Abstract

The present invention discloses a method and system for automated testing and management of electric energy meters and environments. The method comprises: collecting electric energy meter data and laboratory environment data, dividing the standardized data into multiple levels for processing, and performing multi-layer weighted fusion, and obtaining a final fusion result through an adaptive weighting and cross-validation mechanism; utilizing a multi-layer hybrid neural network optimization algorithm to map the final fusion result and environmental data into a high-dimensional space, extracting spatiotemporal features through a spatiotemporal convolution layer, and performing feature integration and final prediction through a dynamic feedback loop layer and an adaptive weight-controlled fully connected layer to generate an electric energy meter performance evaluation result; based on a multi-objective adaptive scheduling algorithm, combining experimental requirements and environmental parameters, as well as the electric energy meter performance evaluation result, optimizing the scheduling strategy through a gradient descent method, and applying the updated scheduling strategy to the test process and experimental conditions to form a closed-loop control to ensure continuous optimization and performance improvement of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of automatic test management of electric energy meters and environments, and in particular to an automatic test management method and system for electric energy meters and environments. Background Art

[0002] Energy meters are not only essential equipment for electricity billing but also a crucial component of the smart grid, carrying out multiple functions such as energy measurement, data collection, and information transmission. To ensure the accuracy, reliability, and long-term stability of energy meters, comprehensive and systematic testing and management are crucial.

[0003] In practical applications, electricity meters operate in complex and diverse environments, encompassing not only conventional indoor environments but also harsh outdoor environments such as high temperature, high humidity, and high noise. The performance of electricity meters can vary significantly under these diverse environmental conditions, making environmental adaptability testing crucial. Furthermore, with the widespread adoption of IoT technology and smart devices, an increasing number of electricity meters are being integrated into smart grid systems for remote monitoring and management. This requires test systems that not only efficiently and accurately collect and process data, but also provide real-time analysis and remote communication capabilities.

[0004] The existing technology has at least the following technical problems: lack of detailed processing of data characteristics and importance, relatively simple performance evaluation methods, resulting in incomplete and inaccurate evaluation results; mostly adopting fixed or semi-fixed scheduling strategies, unable to automatically adjust the test process and experimental conditions according to experimental requirements and environmental changes, resulting in low test efficiency and difficulty in coping with complex experimental requirements, and low efficiency of automated test management. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the defects of the above-mentioned existing technologies and provide a method and system for automatic testing management of electricity meters and environments to achieve closed-loop control from scheduling strategy updates to experiment execution, data collection and performance evaluation, ensuring continuous optimization and performance improvement of the system.

[0006] To this end, the present invention adopts the following technical solutions.

[0007] In a first aspect, the present invention provides an electric energy meter and environment automated test management method, comprising:

[0008] S1. Collect electricity meter data and laboratory environment data, divide the standardized data into multiple levels for processing, and perform multi-layer weighted fusion. Through adaptive weighting and cross-validation mechanisms, ensure the accuracy of the fused data and obtain the final fusion result;

[0009] S2. Utilize a multi-layer hybrid neural network optimization algorithm to map the final fusion results and environmental data into a high-dimensional space. Use a spatiotemporal convolution layer to extract spatiotemporal features. Then, use a dynamic feedback loop layer and an adaptive weight-controlled fully connected layer to perform feature integration and final prediction, generating the performance evaluation results of the electricity meter.

[0010] S3. Based on a multi-objective adaptive scheduling algorithm, combined with experimental requirements and environmental parameters, as well as the performance evaluation results of the electricity meter, the scheduling strategy is optimized through the gradient descent method. The updated scheduling strategy is applied to the test process and experimental conditions to form a closed-loop control to ensure continuous optimization and performance improvement of the system.

[0011] Furthermore, in S1, the standardized data is divided into multiple levels for processing, and multi-layer weighted fusion is performed, and the weight of each data point is dynamically adjusted according to its volatility, so that the weighted fusion data is more accurate; to ensure the accuracy of weighted fusion, the weighted results of each level are verified and adjusted through a cross-validation mechanism, and a global adaptive adjustment is performed based on the cross-validation error of each level to obtain the final fusion result.

[0012] Furthermore, in S1, the calculation formula of the cross-validation error is:

[0013]

[0014] in, It is The cross validation error of the layer, is the number of cross-validation folds, is the current fold of cross validation, It is The data of the i-th electric energy meter in the layer Discount value, It is The jth environment data of the layer is in Discount value, It is Tier The final fusion result after folding, is the number of normalized electric energy meter data, is the number of normalized environmental data.

[0015] Furthermore, in S1, the final fusion result is:

[0016]

[0017] in, is the final fusion result. For the The global weight of the layer, is the total number of data stratification layers;

[0018] The overall weight is dynamically adjusted according to the accuracy of the fusion results at each level, so that the final fusion result can integrate the data characteristics and accuracy of each level to achieve the optimal fusion effect.

[0019] Furthermore, the S2 specifically includes: after the data fusion is completed, the final fusion result and the environmental data are mapped to a high-dimensional space, and the spatiotemporal features of the high-dimensional data are extracted using a spatiotemporal convolution layer. The spatiotemporal convolution layer combines the time and space features and captures the complex relationship of the data in the spatiotemporal domain through a three-dimensional convolution kernel.

[0020] Furthermore, the S2 specifically also includes: combining the outputs of all convolution kernels into a spatiotemporal convolution feature matrix as the input of the dynamic feedback loop layer, the dynamic feedback loop layer adopts an adaptive feedback mechanism to capture the cyclic dependency characteristics of the data; the dynamic feedback loop layer dynamically adjusts the state of the loop layer through the adaptive feedback mechanism, thereby enhancing the multi-layer hybrid neural network model's ability to capture time dependency.

[0021] Furthermore, the S2 specifically includes: the output of the dynamic feedback loop layer is passed to the adaptive weight control fully connected layer to integrate features and make final predictions; the adaptive weight control fully connected layer dynamically adjusts weight parameters according to input features through an adaptive weight control mechanism to enhance the generalization ability of the multi-layer hybrid neural network model.

[0022] Furthermore, the S3 specifically includes: determining the multi-objective optimization target according to the experimental requirements, using the performance evaluation results of the electric energy meter output by the multi-layer hybrid neural network as the performance index input, and calculating the adaptability of the scheduling strategy to the experimental requirements and performance indicators and the adaptability of the scheduling strategy to the environmental data and performance indicators.

[0023] Furthermore, the S3 specifically also includes: calculating the gradient of the adaptability of the scheduling strategy to experimental requirements and performance indicators and the adaptability of the scheduling strategy to environmental data and performance indicators, summarizing all gradients to obtain the gradient of the optimization target, using the gradient descent method to update the scheduling strategy, and applying the updated scheduling strategy to the test process and experimental conditions.

[0024] In a second aspect, the present invention provides an electric energy meter and environment automated test management system, which includes an electric energy meter detection unit, an environmental monitoring unit, a data acquisition module, a preprocessing module, a multi-layer fusion module, a cross-validation module, a performance evaluation module, and a scheduling optimization module;

[0025] The electric energy meter detection unit collects the current, voltage and power of the electric energy meter through the current sensor and the voltage sensor, and transmits the collected electric energy meter data to the data acquisition module;

[0026] The environmental monitoring unit collects temperature, humidity and noise data inside and outside the laboratory through temperature and humidity sensors and noise sensors, and transmits the collected environmental data to the data acquisition module;

[0027] The data acquisition module performs preliminary processing on the energy meter data and the environmental data, and transmits the processed energy meter data and the environmental data to the pre-processing module;

[0028] The pre-processing module performs standardization on the received electric energy meter data and environmental data, eliminates the dimensional differences between the data, and transmits the standardized data to the multi-layer fusion module;

[0029] The multi-layer fusion module divides the standardized data into multiple layers for processing and performs multi-layer weighted fusion to finally obtain a fusion result, which is then sent to the cross-validation module. Global adaptive adjustment is performed based on the optimized fusion result fed back by the cross-validation module, and the final fusion result is obtained and sent to the performance evaluation module.

[0030] The cross-validation module verifies and adjusts the fusion results of the multi-layer weighted fusion through a cross-validation mechanism to ensure the accuracy of the weighted fusion and transmits the optimized fusion results back to the multi-layer fusion module;

[0031] The performance evaluation module uses a multi-layer hybrid neural network optimization algorithm to perform high-dimensional mapping, spatiotemporal convolution feature extraction, dynamic feedback loop, and adaptive weight control on the final fusion result and environmental data, and finally outputs the performance evaluation result of the electric energy meter, which is transmitted to the scheduling optimization module;

[0032] The scheduling optimization module uses a multi-objective adaptive scheduling algorithm to automatically adjust the test process and experimental conditions based on experimental requirements and environmental data, combined with the performance evaluation results of the electricity meter, to generate an optimized scheduling strategy, and applies the updated scheduling strategy to the test process and experimental conditions.

[0033] The present invention has the following beneficial effects:

[0034] 1. Divide the standardized data into multiple layers for processing and perform multi-layer weighted fusion. Through layered processing, the different characteristics of the data are processed more carefully, so that each layer of data fully reflects its importance and characteristics during weighted fusion. The weighted results of each layer are verified and adjusted through a cross-validation mechanism to ensure the high accuracy of the fusion results.

[0035] 2. A multi-layer hybrid neural network optimization algorithm is used to evaluate the performance of the electricity meter, and a comprehensive judgment is made in combination with environmental parameters. By mapping the final fusion results and environmental data to a high-dimensional space, the spatiotemporal features of the high-dimensional data are extracted using a spatiotemporal convolution layer to capture the complex relationships of the data in the spatiotemporal domain. The dynamic feedback loop layer captures the cyclic dependency characteristics of the data through an adaptive feedback mechanism, enhancing the model's ability to capture time dependencies. The adaptive weight control fully connected layer dynamically adjusts the weight parameters according to the input features through an adaptive weight control mechanism, enhancing the generalization ability of the model.

[0036] 3. Utilize a multi-objective adaptive scheduling algorithm and adopt the gradient descent method to update the scheduling strategy, achieving closed-loop control from scheduling strategy update to experiment execution, data collection, and performance evaluation, ensuring continuous optimization and performance improvement of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of an electric energy meter and environment automatic test management method of the present invention;

[0038] Figure 2 This is a structural diagram of an electric energy meter and environmental automatic test management system of the present invention. DETAILED DESCRIPTION

[0039] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0041] The specific scheme of an electric energy meter and environment automatic test management system and method provided by the present invention is described in detail below with reference to the accompanying drawings.

[0042] Refer to the attached Figure 1 , which shows a flow chart of an electric energy meter and environment automated test management method provided by one embodiment of the present invention, which adopts the above-mentioned electric energy meter and environment automated test management system, and the method includes the following steps:

[0043] S1. Collect electricity meter data and laboratory environment data, divide the standardized data into multiple levels for processing, and perform multi-layer weighted fusion. Through adaptive weighting and cross-validation mechanisms, ensure the accuracy of the fused data and obtain the final fusion result.

[0044] The electricity meter detection unit and the environmental monitoring unit collect electricity meter data and environmental data inside and outside the laboratory respectively. The current sensor, voltage sensor and power sensor of the electricity meter detection unit transmit the current, voltage, power and other parameters of the electricity meter to the data acquisition module. The data acquisition module performs preliminary processing on the electricity meter data, including data denoising, data smoothing, data calibration, data synchronization, etc., and then sends it to the preprocessing module; the temperature and humidity sensor and noise sensor of the environmental monitoring unit are responsible for collecting temperature, humidity and noise data inside and outside the laboratory, and the collected environmental data is sent to the preprocessing module.

[0045] The pre-processing module performs standardization on the received energy meter data and environmental data. The initial energy meter data is , the environmental data is In order to eliminate the dimensional differences between the data, the data is standardized so that its mean is 0 and its variance is 1. The standardization formula is:

[0046]

[0047] in, and They are the standardized electricity meter data and environmental data, and are the i-th electricity meter data and the j-th environmental data, and are the mean and standard deviation of the electric energy meter data, and are the mean and standard deviation of the environmental data, respectively. This ensures that data from different sources are processed on the same scale, avoiding processing errors caused by different data dimensions.

[0048] The multi-layer fusion module divides the standardized data into multiple layers for processing and performs multi-layer weighted fusion. The layered data is set as and (No. layer), the initial weight is and The purpose of layered processing is to process the different characteristics of the data in a more detailed manner, so that each layer of data can fully reflect its importance and characteristics during weighted fusion. The multi-layer weighted fusion formula is:

[0049]

[0050] in, It is The data results after layer fusion, is the number of normalized electric energy meter data, is the number of standardized environmental data, and The adaptive adjustment formula is:

[0051]

[0052]

[0053] in, and Respectively The standard deviation of the i-th electric energy meter data and the j-th environmental data, and To adaptively adjust parameters and control the sensitivity of weight adjustment, the weight of each data point is dynamically adjusted according to its volatility, making the fused data more accurate.

[0054] In order to ensure the accuracy of weighted fusion, the cross-validation module verifies and adjusts the weighted results of each level through the cross-validation mechanism. The error formula of cross-validation is:

[0055]

[0056] in, It is The cross validation error of the layer, is the number of cross-validation folds, is the current fold of cross validation, It is The data of the i-th electric energy meter in the layer Discount value, It is The jth environment data of the layer is in Discount value, It is Tier Through the cross-validation mechanism, the fusion results can be evaluated, and the weights of each layer can be dynamically adjusted according to the error feedback to further optimize the data fusion effect.

[0057] The multi-layer fusion module performs global adaptive adjustment based on the cross-validation errors at each level to obtain the final fusion result:

[0058]

[0059] in, is the final fusion result. For the The global weight of the layer, is the total number of data layers. The overall weight is dynamically adjusted based on the accuracy of the fusion results at each layer, so that the final fusion result can integrate the data characteristics and accuracy of each layer to achieve the optimal fusion effect.

[0060] S2. Utilize a multi-layer hybrid neural network optimization algorithm to map the final fusion results and environmental data into a high-dimensional space. Use the spatiotemporal convolution layer to extract spatiotemporal features. Then, use the dynamic feedback loop layer and the adaptive weight-controlled fully connected layer to perform feature integration and final prediction, generating the performance evaluation results of the electricity meter.

[0061] After the data fusion is completed, the performance evaluation module uses a multi-layer hybrid neural network optimization algorithm to evaluate the performance of the electricity meter and makes a comprehensive judgment based on environmental parameters.

[0062] Specifically, the result after data fusion and environmental data Mapped to high-dimensional space, let the input data be ,in , the data after high-dimensional mapping is , the mapping formula is:

[0063]

[0064] in, 、 、 is the mapping parameter. Through high-dimensional mapping, the nonlinear characteristics of the data are fully displayed, and the expressive power of the multi-layer hybrid neural network model is enhanced.

[0065] Furthermore, the spatiotemporal convolution layer is used to extract the spatiotemporal features of high-dimensional data. The spatiotemporal convolution layer combines temporal and spatial features and captures the complex relationship of data in the spatiotemporal domain through a three-dimensional convolution kernel. The formula for the convolution operation is:

[0066]

[0067] in, is the output of the rth convolution kernel, is the weight of the rth convolution kernel at the sth input feature position, is the sth feature of the input data, is the bias of the rth convolution kernel, is the activation function. This extracts the local features of high-dimensional data and achieves preliminary aggregation of features through convolution operations.

[0068] The output of all convolution kernels Combined into a spatiotemporal convolution feature matrix : , spatiotemporal convolution feature matrix As the input of the dynamic feedback loop layer, the dynamic feedback loop layer adopts an adaptive feedback mechanism to capture the cyclic dependency characteristics of the data; the dynamic feedback loop layer dynamically adjusts the state of the loop layer through the adaptive feedback mechanism, enhancing the multi-layer hybrid neural network model's ability to capture time dependencies. The calculation formula of the dynamic feedback loop layer is:

[0069]

[0070] in, is the output of the dynamic feedback loop layer at the current moment, and are weight matrix and feedback weight matrix respectively, is the bias, is the activation function, is the adaptive feedback parameter, is the feedback function. By capturing the temporal dependencies in the data, the adaptive feedback mechanism improves the processing capability of the multi-layer hybrid neural network model for time series data.

[0071] The output of the dynamic feedback loop layer is passed to the adaptive weight control fully connected layer to integrate features and make the final prediction. The adaptive weight control fully connected layer uses the adaptive weight control mechanism to dynamically adjust the weight parameters according to the input features, thereby enhancing the generalization ability of the multi-layer hybrid neural network model. The output calculation formula of the adaptive weight control fully connected layer is:

[0072]

[0073] in, is the output of the fully connected layer with adaptive weight control, i.e., the performance evaluation result of the electric energy meter. The weight matrix of the fully connected layer is adjusted for adaptive weights. is the bias, is the activation function, is the adaptive weight parameter, is the weight control function, which is based on the input features and the weight matrix Dynamically adjust the weight parameters to enhance the generalization ability of the multi-layer hybrid neural network model. The specific calculation process of the weight control function is as follows:

[0074] First calculate the weight matrix The row norm of and column norm :

[0075]

[0076]

[0077] in, is the weight matrix No. Rank Elements of the column, is the number of rows, is the number of columns.

[0078] Calculate eigenvectors Norm of :

[0079]

[0080] in, is the eigenvector No. elements, is the eigenvector The number of elements.

[0081] Calculate the adaptive control factor using the row and column norms of the weight matrix and the norm of the eigenvector and :

[0082]

[0083]

[0084] in, A constant to prevent division by zero.

[0085] Weight control function By combining adaptive regulatory factors and And the linear combination of the weight matrix and the eigenvector is:

[0086]

[0087] in, is the weight matrix The transpose of .

[0088] S3. Based on a multi-objective adaptive scheduling algorithm, combined with experimental requirements and environmental parameters, as well as the performance evaluation results of the electricity meter, the scheduling strategy is optimized through the gradient descent method. The updated scheduling strategy is applied to the test process and experimental conditions to form a closed-loop control to ensure continuous optimization and performance improvement of the system.

[0089] After the performance evaluation of the electricity meter is completed, the scheduling optimization module uses a multi-objective adaptive scheduling algorithm to automatically adjust the test process and experimental conditions according to the experimental requirements.

[0090] Specifically, the multi-objective optimization goal is determined according to the experimental requirements. Let the experimental requirements be , the environmental data is , the scheduling strategy is , the optimization goal is , the performance evaluation results of the electric energy meter output by the multi-layer hybrid neural network are used as the performance index input, and the multi-objective optimization formula is:

[0091]

[0092] in, Indicates the adaptability of the scheduling strategy to experimental requirements and performance indicators, Indicates the adaptability of the scheduling strategy to environmental data and performance indicators, is the weight of the experimental demand indicator q, is the weight of the environmental data indicator p, is the total number of experimental requirements, is the total number of environmental data.

[0093]

[0094]

[0095] in, Control experiment requirements Importance in the fitness function, is the qth experimental requirement, which indicates the specific demand indicators during the experiment. Is the impact of experimental needs and scheduling strategies The parameters of the relationship between is the part of the scheduling strategy that corresponds to the qth experimental requirement. is the performance index of the electric energy meter related to the qth experimental requirement, Is the control performance index Importance in fitness functions; Control environment data Importance in the fitness function, is the pth environmental data, Is the impact of environmental data and scheduling strategies The parameters of the relationship between It is the policy part corresponding to the pth environmental data in the scheduling policy. Control environment data Importance in the fitness function, It is the performance index of the electric energy meter related to the pth experimental requirement.

[0096] right Taking the derivative, we get The gradient of is:

[0097]

[0098] right Taking the derivative, we get The gradient of is:

[0099]

[0100] Summarize all gradients and optimize the target The gradient of is:

[0101]

[0102] Substituting the gradients of each part obtained previously, we get:

[0103]

[0104] Update the scheduling strategy using gradient descent:

[0105]

[0106] The specific update formula is:

[0107]

[0108] in, and They are the scheduling strategies before and after the update, is the learning rate, which controls the step size of the scheduling policy update.

[0109] The updated scheduling strategy Applied to test processes and experimental conditions. Specifically, the operating parameters of the test equipment and the control parameters of the experimental environment are adjusted to meet the requirements of the new scheduling strategy; experiments and tests are carried out according to the updated scheduling strategy. During the execution of experiments and tests, real-time data from the electricity meter and environmental data inside and outside the laboratory are continuously collected; the new experimental data is input into the multi-layer hybrid neural network model for performance evaluation and the generation of new performance indicators; the effect of the new scheduling strategy is evaluated to determine whether it has achieved the expected optimization goal; if the performance indicator has achieved the expected goal, the optimization process is terminated; otherwise, the scheduling strategy is optimized again based on the new performance indicator and the current experimental data, and the above steps are repeated until the optimization goal is achieved; thus, closed-loop control is achieved from scheduling strategy update to experiment execution, data collection and performance evaluation, ensuring continuous optimization and performance improvement of the system.

[0110] Refer to the attached Figure 2, which shows a structural diagram of an electric energy meter and environment automatic test management system provided by an embodiment of the present invention. The system is used to implement the above-mentioned electric energy meter and environment automatic test management method, and includes the following parts: an electric energy meter detection unit, an environmental monitoring unit, a data acquisition module, a preprocessing module, a multi-layer fusion module, a cross-validation module, a performance evaluation module and a scheduling optimization module.

[0111] The electric energy meter detection unit collects the current, voltage, power and other parameters of the electric energy meter through the current sensor and voltage sensor, and transmits the collected data to the data acquisition module;

[0112] The environmental monitoring unit collects temperature, humidity and noise data inside and outside the laboratory through temperature and humidity sensors and noise sensors, and transmits the collected data to the data acquisition module;

[0113] The data acquisition module performs preliminary processing on the electric energy meter data and environmental data, and transmits the processed data to the pre-processing module;

[0114] The pre-processing module standardizes the received electric energy meter data and environmental data, eliminates the dimensional differences between the data, and transmits the standardized data to the multi-layer fusion module;

[0115] The multi-layer fusion module divides the standardized data into multiple layers for processing and performs multi-layer weighted fusion to obtain the fusion result, which is then sent to the cross-validation module. Global adaptive adjustment is performed based on the optimized fusion result fed back by the cross-validation module, and the final fusion result is sent to the performance evaluation module.

[0116] The cross-validation module verifies and adjusts the fusion results of the multi-layer weighted fusion through the cross-validation mechanism to ensure the accuracy of the weighted fusion and transmits the optimized fusion results back to the multi-layer fusion module;

[0117] Performance evaluation module: uses a multi-layer hybrid neural network optimization algorithm to perform high-dimensional mapping, spatiotemporal convolution feature extraction, dynamic feedback loop, and adaptive weight control on the final fusion result and environmental data, and finally outputs the performance evaluation result of the electric energy meter, which is transmitted to the scheduling optimization module;

[0118] The scheduling optimization module uses a multi-objective adaptive scheduling algorithm to automatically adjust the test process and experimental conditions according to experimental requirements and environmental data, combined with performance evaluation results, to generate an optimized scheduling strategy, and then apply the updated scheduling strategy to the test process and experimental conditions.

[0119] The beneficial effects of this embodiment are:

[0120] 1. The standardized data is divided into multiple layers for processing and multi-layer weighted fusion is performed. The layered processing more carefully handles the different characteristics of the data, so that each layer of data fully reflects its importance and characteristics during weighted fusion. An adaptive adjustment formula is used to dynamically adjust the weight of each data point according to its volatility, making the fused data more accurate. The weighted results of each layer are verified and adjusted through a cross-validation mechanism to ensure the high accuracy of the fusion results.

[0121] 2. A multi-layer hybrid neural network optimization algorithm is used to evaluate the performance of electricity meters and make comprehensive judgments based on environmental parameters. By mapping the fusion results and environmental data into a high-dimensional space, a spatiotemporal convolution layer is used to extract the spatiotemporal features of the high-dimensional data and capture the complex relationships of the data in the spatiotemporal domain. The dynamic feedback loop layer captures the cyclic dependency characteristics of the data through an adaptive feedback mechanism, enhancing the model's ability to capture time dependencies. The adaptive weight control fully connected layer dynamically adjusts the weight parameters based on the input features through an adaptive weight control mechanism, enhancing the model's generalization ability.

[0122] 3. The scheduling optimization module uses a multi-objective adaptive scheduling algorithm to automatically adjust the test process and experimental conditions according to the experimental requirements. Through the multi-objective optimization formula, it comprehensively considers the experimental requirements, environmental parameters and performance indicators to dynamically adjust the scheduling strategy. The gradient descent method is used to update the scheduling strategy. By evaluating the effect of the new scheduling strategy, it is determined whether it has achieved the expected optimization goal. It then performs iterative optimization based on the feedback results to achieve closed-loop control from scheduling strategy update to experiment execution, data collection and performance evaluation, ensuring continuous optimization and performance improvement of the system.

[0123] The order in which the embodiments of the present invention are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0124] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0125] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for automatic test management of electric energy meters and environments, characterized in that: include: S1. Collect electricity meter data and laboratory environment data, divide the standardized data into multiple levels for processing, and perform multi-layer weighted fusion. Through adaptive weighting and cross-validation mechanism, the final fusion result is obtained; S2. Utilize a multi-layer hybrid neural network optimization algorithm to map the final fusion results and environmental data into a high-dimensional space. Use a spatiotemporal convolution layer to extract spatiotemporal features. Then, use a dynamic feedback loop layer and an adaptive weight-controlled fully connected layer to perform feature integration and final prediction, generating the performance evaluation results of the electricity meter. S3. Based on a multi-objective adaptive scheduling algorithm, combined with experimental requirements and environmental parameters, as well as the performance evaluation results of the electricity meter, the scheduling strategy is optimized through the gradient descent method. The updated scheduling strategy is applied to the test process and experimental conditions to form a closed-loop control; In S1, the specific contents of the adaptive weighting and cross-validation mechanism are as follows: dynamically adjusting the weight of each data point according to its volatility, verifying and adjusting the weighted results of each level through a cross-validation mechanism, and performing global adaptive adjustment based on the cross-validation error of each level; Said S2 specifically includes: after the data fusion is completed, mapping the final fusion result and the environmental data to a high-dimensional space, using the spatiotemporal convolution layer to extract the spatiotemporal features of the high-dimensional data, the spatiotemporal convolution layer combines the time and space features, and captures the complex relationship of the data in the spatiotemporal domain through the three-dimensional convolution kernel; combining the outputs of all convolution kernels into a spatiotemporal convolution feature matrix as the input of the dynamic feedback loop layer, the dynamic feedback loop layer adopts an adaptive feedback mechanism to capture the cyclic dependency characteristics of the data; the dynamic feedback loop layer dynamically adjusts the state of the loop layer through the adaptive feedback mechanism; the output of the dynamic feedback loop layer is passed to the adaptive weight control fully connected layer to integrate the features and make the final prediction; the adaptive weight control fully connected layer dynamically adjusts the weight parameters according to the input features through the adaptive weight control mechanism; In S3, the multi-objective optimization goal is determined according to the experimental requirements. Let the experimental requirements be , the environmental data is , the scheduling strategy is , the optimization goal is , the performance evaluation results of the electric energy meter output by the multi-layer hybrid neural network are used as the performance index input, and the multi-objective optimization formula is: ， in, Indicates the adaptability of the scheduling strategy to experimental requirements and performance indicators, Indicates the adaptability of the scheduling strategy to environmental data and performance indicators, is the weight of the experimental demand indicator q, is the weight of the environmental data indicator p, is the total number of experimental requirements, is the total number of environmental data; ， ， in, Control experiment requirements Importance in the fitness function, is the qth experimental requirement, which indicates the specific demand indicators during the experiment. Is the impact of experimental needs and scheduling strategies The parameters of the relationship between is the part of the scheduling strategy that corresponds to the qth experimental requirement. is the performance index of the electric energy meter related to the qth experimental requirement, Is the control performance index Importance in fitness functions; Control environment data Importance in the fitness function, is the pth environmental data, Is the impact of environmental data and scheduling strategies The parameters of the relationship between It is the policy part corresponding to the pth environmental data in the scheduling policy. Control environment data Importance in the fitness function, is the performance index of the electric energy meter related to the pth experimental requirement; The scheduling strategy is updated using the gradient descent method. The specific update formula is: ， in, and They are the scheduling strategies before and after the update, is the learning rate, which controls the step size of the scheduling policy update.

2. The method for automatic test management of electric energy meters and environments according to claim 1, characterized in that: In S1, the calculation formula of the cross-validation error is: ， in, It is The cross validation error of the layer, is the number of cross-validation folds, is the current fold of cross validation, It is The data of the i-th electric energy meter in the layer Discount value, It is The jth environment data of the layer is in Discount value, It is Tier The final fusion result after folding, is the number of normalized electric energy meter data, is the number of normalized environmental data.

3. The method for automatic test management of electric energy meters and environments according to claim 2, characterized in that: In S1, the final fusion result is: ， in, is the final fusion result. For the The global weight of the layer, is the total number of data stratification layers; The overall weight is dynamically adjusted according to the accuracy of the fusion results at each level, so that the final fusion result can integrate the data characteristics and accuracy of each level to achieve the optimal fusion effect.

4. The method for automatic test management of electric energy meters and environments according to claim 1, characterized in that: Said S3 specifically includes: determining the multi-objective optimization target according to the experimental requirements, using the performance evaluation results of the electric energy meter output by the multi-layer hybrid neural network as the performance index input, and calculating the adaptability of the scheduling strategy to the experimental requirements and performance indicators and the adaptability of the scheduling strategy to the environmental data and performance indicators.

5. The method for automatic test management of electric energy meters and environments according to claim 4, characterized in that: The S3 specifically also includes: calculating the gradients of the scheduling strategy's adaptability to experimental requirements and performance indicators and the scheduling strategy's adaptability to environmental data and performance indicators, aggregating all gradients to obtain the gradient of the optimization target, updating the scheduling strategy using the gradient descent method, and applying the updated scheduling strategy to the test process and experimental conditions.

6. An automatic test management system for electric energy meters and environments, used to implement the automatic test management method for electric energy meters and environments according to any one of claims 1 to 5, characterized in that: It includes an energy meter detection unit, an environmental monitoring unit, a data acquisition module, a preprocessing module, a multi-layer fusion module, a cross-validation module, a performance evaluation module, and a scheduling optimization module; The electric energy meter detection unit collects the current, voltage and power of the electric energy meter through the current sensor and the voltage sensor, and transmits the collected electric energy meter data to the data acquisition module; The environmental monitoring unit collects temperature, humidity and noise data inside and outside the laboratory through temperature and humidity sensors and noise sensors, and transmits the collected environmental data to the data acquisition module; The data acquisition module performs preliminary processing on the energy meter data and the environmental data, and transmits the processed energy meter data and the environmental data to the pre-processing module; The pre-processing module performs standardization on the received electric energy meter data and environmental data, eliminates the dimensional differences between the data, and transmits the standardized data to the multi-layer fusion module; The multi-layer fusion module divides the standardized data into multiple layers for processing, and performs multi-layer weighted fusion to finally obtain the fusion result, which is then transmitted to the cross-validation module; Perform global adaptive adjustment based on the optimized fusion results fed back by the cross-validation module, and obtain the final fusion results and transmit them to the performance evaluation module; The cross-validation module verifies and adjusts the fusion results of the multi-layer weighted fusion through a cross-validation mechanism to ensure the accuracy of the weighted fusion and transmits the optimized fusion results back to the multi-layer fusion module; The performance evaluation module uses a multi-layer hybrid neural network optimization algorithm to perform high-dimensional mapping, spatiotemporal convolution feature extraction, dynamic feedback loop, and adaptive weight control on the final fusion result and environmental data, and finally outputs the performance evaluation result of the electric energy meter, which is transmitted to the scheduling optimization module; The scheduling optimization module uses a multi-objective adaptive scheduling algorithm to automatically adjust the test process and experimental conditions based on experimental requirements and environmental data, combined with the performance evaluation results of the electricity meter, to generate an optimized scheduling strategy, and applies the updated scheduling strategy to the test process and experimental conditions.

Images