A dispatch optimization method based on low-pressure metering box automatic detection

Abstract

The application provides a scheduling optimization method based on low-voltage metering box automatic detection, and belongs to the technical field of smart power grids, and comprises the following steps: S1: acquiring detection data, pre-processing the detection data, and performing feature extraction based on the pre-processed detection data to obtain key feature data; S2: performing enhancement on the key feature data based on a dynamic weight analysis enhancement algorithm, and optimizing the representation of data features by dynamically adjusting the weight; S3: identifying the current state of the low-voltage metering box based on an intelligent fusion scheduling algorithm, constructing an optimization distribution model, and intelligently distributing detection tasks; and S4: adaptively optimizing the distribution of the detection tasks based on a multi-dimensional situation response scheduling algorithm to obtain the distribution of optimal detection tasks. The application can ensure that the skills of technical personnel match the task requirements, effectively improve the efficiency of low-voltage metering box automatic detection, ensure that the task distribution is reasonable and efficient, and simultaneously reduce cost and time waste.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart grids, and in particular relates to a scheduling optimization method based on automatic detection of low-voltage meter boxes. Background Art

[0002] In automated inspection systems, especially those for monitoring and managing low-voltage metering boxes, effectively processing and utilizing large amounts of data while ensuring the accuracy and efficiency of inspection tasks has always been a major challenge in the industry. Traditional inspection systems typically rely on simple data processing techniques and static task scheduling methods, which often cannot fully adapt to rapidly changing environmental conditions or complex operational requirements.

[0003] Furthermore, task scheduling in traditional systems is often pre-defined and lacks flexibility. This static scheduling strategy fails to consider dynamic information such as technician availability, task urgency, and the current state of equipment, leading to uneven resource utilization, long response times, and high costs.

[0004] Furthermore, existing systems often exhibit low adaptability and robustness when faced with rapidly changing equipment states or fluctuations in the external environment. These scheduling algorithms fail to consider multi-dimensional influencing factors, such as changing environmental conditions, differences in technician skills, and historical equipment failure data. This results in poor adaptability to environmental changes and low resource utilization and task scheduling efficiency. Summary of the Invention

[0005] In order to solve the above problems in the prior art, a scheduling optimization method based on automatic detection of low-voltage metering boxes is provided.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] This technical solution proposes a scheduling optimization method based on automatic detection of low-voltage meter boxes, including the following steps:

[0008] S1: Based on the goals and / or functional requirements of the low-voltage meter box automated detection, obtain detection data, pre-process the detection data, and extract features based on the pre-processed detection data to obtain key feature data;

[0009] S2: Based on the dynamic weight analysis enhancement algorithm, key feature data is enhanced. By dynamically adjusting the weights, the representation of data features is optimized to obtain intelligently processed data.

[0010] S3: Based on the intelligent fusion scheduling algorithm, it identifies the current status of the low-voltage meter box, builds an optimized allocation model, and intelligently allocates detection tasks;

[0011] S4: Based on the multi-dimensional situation response scheduling algorithm, the allocation of detection tasks is adaptively optimized to obtain the optimal allocation of detection tasks.

[0012] Preferably, in said S1, the preprocessing of the detection data includes data cleaning, missing value processing, noise smoothing and data normalization, and the key feature data includes statistical features, trend features and frequency domain features.

[0013] Preferably, in S2, the dynamic weight analysis enhancement algorithm includes:

[0014] Initialize the weight matrix: Assign an initial weight to each data feature. The weight matrix is ​​represented by W and the formula is as follows:

[0015]

[0016] Where w i is the initial weight of the i-th feature, and n is the number of features;

[0017] Calculate weighted eigenvalues: Use the weight matrix to weight the feature data X, each eigenvalue x i The corresponding weight w i Multiplying, the formula is as follows:

[0018]

[0019] Where Z is the weighted eigenvector, which represents the comprehensive index of the eigenvalue after weight processing, x i is the original value of the i-th feature.

[0020] Preferably, the weight is adjusted dynamically: the weight of each feature value is updated according to its utility, and the formula is as follows:

[0021]

[0022] Where, is the updated weight, w i is the current weight of the i-th feature, exp is the exponential function used to calculate the update scale of the weight, σ i is the standard deviation of the i-th feature, which adjusts the sensitivity of weight update, s i and s j are the utility scores of the i-th and j-th features respectively, and n is the total number of features;

[0023] Based on the data fusion algorithm, the enhanced feature data and the preprocessed data are fused to obtain intelligently processed data.

[0024] Preferably, in S3, the optimization allocation model construction method includes:

[0025] Define decision variables: Define decision variables y ijt It is expressed as whether the i-th technician is assigned to the j-th task at time t, y ijt =1 means assigned, y ijt =0 means not allocated;

[0026] Define the objective function; construct the objective function to maximize the weighted completion rate of task priorities and minimize the total cost, as follows:

[0027]

[0028] Where F is the objective function, p j is the priority weight of task j, c ijt is the cost of the i-th technician performing task j at time t, N is the total number of technicians, M is the total number of tasks that need to be assigned, and T is the time period.

[0029] Preferably, the constraints are: each task must be assigned at least once within the specified time, and each technician can only perform one task at any time. The formula is as follows:

[0030]

[0031] Preferably, in S4, the multi-dimensional situation response scheduling algorithm includes:

[0032] Define multi-dimensional influencing factors, define influencing factors and quantification methods of influencing factors, and construct matrix representation of quantitative variables. jt is the influencing factor vector of task j at time t, including environmental factors, operator skill level, and historical failure rate of equipment;

[0033] Weight adjustment of influencing factors: For each task, the weight is dynamically adjusted according to the task influencing factor vector. The formula is as follows:

[0034]

[0035] Where w jt is the weight of the task at time, α k is the weight coefficient of the kth influencing factor, v jtk is the quantitative value of the kth influencing factor of the jth task at time t, β l is the weight coefficient of other influencing factors, u jtl is the quantitative value of other influencing factors.

[0036] Preferably, the objective function is optimized, and the formula is as follows:

[0037]

[0038] Where F' is the optimized objective function;

[0039] Constraint expansion: Add dynamic availability constraints, skill matching constraints, task completion requirements, continuous working time limits, and multi-task fairness constraints;

[0040] After optimizing the objective function and constraints, a heuristic algorithm combined with linear programming is used for optimization, a genetic algorithm is used to generate an initial solution, and then the initial solution is fine-tuned through linear programming to obtain the global optimal solution.

[0041] Preferably, based on the precision enhancement adjustment algorithm, the genetic algorithm is optimized to find a better solution in the global search space. The precision enhancement adjustment algorithm includes:

[0042] Introduction of local search: By introducing local search, the accuracy and ability of the adjustment algorithm to escape when encountering a local optimal solution are enhanced;

[0043] Implement local perturbation: In each generation of the genetic algorithm, a specified proportion of individuals are selected for local perturbation. The perturbation is in the form of applying periodic sinusoidal waveform changes to the gene r of the individual. The formula is as follows:

[0044]

[0045] Where r old is the original gene value, δ is the disturbance amplitude, which controls the size of the change, λ is the periodicity of the disturbance, which determines the frequency of the disturbance, and r new is the new gene value after perturbation adjustment, and φ is the phase offset.

[0046] Preferably, the maintenance of genetic diversity: calculate the difference between individuals in the population, use the Hamming distance as a measure of difference, and realize diversity assessment;

[0047] When the population diversity is lower than the preset threshold, the diversity is increased by adding randomly generated new individuals or producing more diverse offspring through crossover and mutation, as shown below:

[0048]

[0049] Where, H(r i ,r avg ) is individual r i and the population average r avg Hamming distance, γ is a positive adjustment factor used to adjust the impact of diversity, G new is the diversity index of the new generation population, r i is the new gene value of the i-th individual, and Num is the number of individuals in the population.

[0050] Preferably, the perturbation amplitude is dynamically adjusted: the perturbation amplitude is dynamically adjusted according to the progress of the search process, and the perturbation amplitude δ is adjusted according to the changes in the optimal solution in consecutive generations;

[0051] If the optimal solution is not improved for a time exceeding the set threshold, increase the perturbation amplitude and expand the exploration area. If the optimal solution is not improved for a time less than the set threshold, reduce the perturbation amplitude and refine the search in the current area. The formula is as follows:

[0052]

[0053] Where, δ old is the perturbation amplitude used in the previous generation, δ new is the newly calculated perturbation amplitude, which is used for the local search of the next generation. Δr is the difference in the optimal fitness of the previous and next generations, which indicates the speed of the algorithm optimization progress. Δt is the algebraic change, which indicates the progress of time. ρ is the adjustment coefficient, which controls the parameter of the perturbation adjustment rate to adapt to the dynamically changing optimization environment.

[0054] Compared with the prior art, the present invention has the following advantages:

[0055] 1. This application effectively optimizes the representation of feature data and improves the effect of data processing by applying a dynamic weight analysis enhancement algorithm, making the data more reflective of the actual device status, thereby accelerating the speed of status identification and decision-making.

[0056] 2. This application applies an intelligent fusion scheduling algorithm to dynamically allocate detection tasks based on the identified meter box status, thereby improving resource utilization efficiency and the timeliness of task execution. The optimized allocation model takes into account multiple factors, such as the urgency of the task and the availability of technical personnel, to ensure that task allocation is reasonable and efficient, while reducing costs and time waste.

[0057] 3. By introducing multi-dimensional influencing factors, such as environmental conditions, technicians' skill levels, and historical equipment failure rates, this application's multi-dimensional situation response scheduling algorithm can more accurately match task requirements and site conditions, optimize resource allocation, and the dynamic weight adjustment mechanism ensures that changes in various influencing factors are reflected in real time during the task allocation process, making scheduling decisions more flexible and adaptable.

[0058] 4. This application adjusts task allocation by applying an optimized allocation model to ensure that the skills of technicians match task requirements and improve work efficiency. At the same time, it takes into account labor costs and resource allocation to reduce overall operating costs. Dynamic availability and multi-task fairness constraints ensure that technicians will not be overworked and maintain work quality and efficiency. The introduced precision enhancement adjustment algorithm enhances the adaptability to environmental changes and potential errors through periodic sinusoidal waveform changes and genetic diversity maintenance mechanisms. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which:

[0060] Figure 1 It is the overall flow chart of the present invention. DETAILED DESCRIPTION

[0061] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0062] like Figure 1 As shown, this embodiment proposes a scheduling optimization method based on automatic detection of low-voltage meter boxes, including the following steps:

[0063] S1: Based on the goals and / or functional requirements of the low-voltage meter box automated detection, obtain detection data, pre-process the detection data, and extract features based on the pre-processed detection data to obtain key feature data;

[0064] S2: Based on the dynamic weight analysis enhancement algorithm, key feature data is enhanced. By dynamically adjusting the weights, the representation of data features is optimized to obtain intelligently processed data.

[0065] S3: Based on the intelligent fusion scheduling algorithm, it identifies the current status of the low-voltage meter box, builds an optimized allocation model, and intelligently allocates detection tasks;

[0066] S4: Based on the multi-dimensional situation response scheduling algorithm, the allocation of detection tasks is adaptively optimized to obtain the optimal allocation of detection tasks.

[0067] In S1, the goals and functional requirements for automated testing of low-voltage meter boxes are determined based on experience. These goals include improving testing efficiency, reducing human error, and enhancing data management and utilization. Functional requirements include automatically identifying meter box status, automatically recording test data, and automatically diagnosing faults. Based on these goals and functional requirements, sensors and data acquisition equipment are configured to acquire test data in real time. This data includes operational data on voltage, current, and power, environmental data on temperature and humidity, and equipment status data.

[0068] The preprocessing of detection data includes data cleaning, missing value processing, noise smoothing and data normalization. The preprocessed data is subjected to feature extraction using feature engineering algorithms to obtain key feature data, which include statistical features, trend features and frequency domain features.

[0069] In S2, the dynamic weight analysis enhancement algorithm includes:

[0070] Initialize the weight matrix: assign an initial weight to each data feature, and the weight matrix is denoted as W. The initialization of the weight can be pre-set according to the experience method combined with the importance of the feature, and the formula is as follows:

[0071]

[0072] In the formula, w i is the initial weight of the i-th feature, and n is the number of features;

[0073] Calculate the weighted feature value: use the weight matrix to weight the feature data X to enhance the influence of important features, and each feature value x i is multiplied by its corresponding weight w i , and the formula is as follows:

[0074]

[0075] In the formula, Z is the weighted feature vector, which represents the comprehensive index of the feature value after weight processing, x i is the original value of the i-th feature.

[0076] Dynamically adjust the weight: update the weight of each feature value according to its utility, and the formula is as follows:

[0077]

[0078] In the formula, is the updated weight, w i is the current weight of the i-th feature, exp is the exponential function, which is used to calculate the update size of the weight, σ i is the standard deviation of the i-th feature, which adjusts the sensitivity of weight update, s i and s j are the utility scores of the i-th and j-th features, respectively, which reflect the contribution of each feature to the model, and n is the total number of features;

[0079] Based on the data fusion algorithm, the enhanced feature data and the preprocessed data are fused to obtain intelligent processed data.

[0080] In S3, a machine learning model such as support vector machine is used to identify the state of the metering box according to the intelligent processed data, and the state includes normal, needs maintenance, and failure. An optimized allocation model is constructed according to the identification state of the metering box to intelligently allocate detection tasks.

[0081] The optimization allocation model can dynamically adjust task allocation based on the status of the meter box, the urgency of the task, and the availability of technicians. The optimization allocation model construction method includes:

[0082] Define decision variables: Define decision variables y ijt It is expressed as whether the i-th technician is assigned to the j-th task at time t, y ijt =1 means assigned, y ijt =0 means not allocated;

[0083] Define the objective function; construct the objective function to maximize the weighted completion rate of task priorities and minimize the total cost, as follows:

[0084]

[0085] Where F is the objective function, p j is the priority weight of task j, which reflects the urgency and importance of the task. The priority is set based on the impact scope, risk level or criticality of the task according to the empirical method. ijt is the cost of the i-th technician performing task j at time t, which may include labor cost, travel time, cost of required resources, etc. N is the total number of technicians, M is the total number of tasks that need to be assigned, and T is the time period, reflecting the time dimension of task scheduling.

[0086] Constraints: Each task must be assigned at least once within the specified time, and each technician can only perform one task at any time. The formula is as follows:

[0087]

[0088] By applying the dynamic weight analysis enhancement algorithm, the representation of feature data is effectively optimized, the effect of data processing is improved, and the data is made to better reflect the actual device status, thereby accelerating the speed of status identification and decision making.

[0089] By applying an intelligent fusion scheduling algorithm, inspection tasks are dynamically allocated based on the identified meter box status, improving resource utilization efficiency and the timeliness of task execution. The optimized allocation model takes into account multiple factors, such as the urgency of the task and the availability of technicians, ensuring that task allocation is reasonable and efficient while reducing costs and time waste.

[0090] In S4, the multi-dimensional situation response scheduling algorithm includes:

[0091] Define multi-dimensional influencing factors, define influencing factors and quantification methods of influencing factors, and construct matrix representation of quantitative variables. jtis the influencing factor vector of task j at time t, including environmental factors, operator skill level, and historical failure rate of equipment;

[0092] Weight adjustment of influencing factors: For each task, the weight is dynamically adjusted according to the task influencing factor vector. The formula is as follows:

[0093]

[0094] Where w jt is the weight of the task at time, α k is the weight coefficient of the kth influencing factor, v jtk is the quantitative value of the kth influencing factor of the jth task at time t, β l is the weight coefficient of other influencing factors, u jtl is the quantitative value of other influencing factors.

[0095] Optimize the objective function, the formula is as follows:

[0096]

[0097] Where F' is the optimized objective function;

[0098] Constraint extensions: Dynamic availability constraints are added to ensure that technicians are assigned to appropriate tasks based on their real-time availability and qualifications; skill matching constraints are added to ensure that technicians assigned to specific tasks have the skill level required to perform the task; task completion requirements are added to ensure that each task is assigned a sufficient number of times within its required time window to meet its completion criteria; continuous working time restrictions are added to limit the continuous working time of technicians to prevent overwork; and multi-task fairness constraints are added to ensure that task allocation is as fair as possible among technicians.

[0099] After optimizing the objective function and constraints, a heuristic algorithm combined with linear programming was used for optimization. A genetic algorithm was used to generate an initial solution, which was then fine-tuned through linear programming to ensure global optimality. This optimal solution enabled optimal inspection task allocation and optimized scheduling for automated inspection of low-voltage meter boxes.

[0100] In the implementation of the genetic algorithm, in order to avoid the problem of local optimal solutions instead of global optimal solutions, a precision enhancement adjustment algorithm is introduced to optimize the genetic algorithm to ensure that the algorithm can effectively avoid local optimal solutions and find a better solution in the global search space. The precision enhancement adjustment algorithm includes:

[0101] Introduction of local search: By introducing local search, the accuracy and ability of the adjustment algorithm to escape from local optimal solutions are enhanced, thereby exploring a larger search space;

[0102] Local perturbation is implemented: in each generation of genetic algorithm, a specified proportion of individuals are selected for local perturbation according to the experience method, the form of perturbation is to apply a periodic sinusoidal wave change on the gene r of the individual, to increase the randomness and coverage of the search, the formula is as follows:

[0103]

[0104] In the formula, r old is the original gene value, δ is the perturbation amplitude, which controls the size of the change, λ is the control of the periodicity of the perturbation, which determines the perturbation frequency, r new is the new gene value after perturbation adjustment, φ is the phase shift, which is introduced to increase the randomness of the perturbation.

[0105] Maintenance of genetic diversity: maintaining the genetic diversity of the population is a key measure to avoid premature convergence of genetic algorithm, which can enhance the global search ability, the difference between individuals in the population is calculated, the Hamming distance is used as the measure of difference, and the diversity evaluation is realized;

[0106] When the population diversity is lower than the preset threshold, the diversity is improved by increasing the randomly generated new individuals or by crossing and mutation to produce more diversified offspring, which is shown as follows:

[0107]

[0108] In the formula, H(r i ,r avg ) is the Hamming distance between individual r i and the average r avg of the population, γ is a positive adjustment factor used to adjust the influence degree of diversity, G new is the diversity index of the new generation population, r i is the new gene value of the i-th individual, and Num is the number of individuals in the population.

[0109] Dynamic adjustment of perturbation amplitude: according to the progress in the search process, the perturbation amplitude is dynamically adjusted to maintain the balance between exploration and development, according to the change of the optimal solution in several consecutive generations, the perturbation amplitude δ is adjusted; if the time of optimal solution improvement exceeds the set threshold, the perturbation amplitude is increased to enhance the exploration area, if the time of optimal solution improvement is less than the set threshold, the perturbation amplitude is reduced to refine the search of the current area, the formula is as follows:

[0110]

[0111] In the formula, δ old is the perturbation amplitude used in the last generation, δ newis the newly calculated perturbation amplitude, which is used for the local search of the next generation. Δr is the difference in the optimal fitness of the previous and next generations, which indicates the speed of the algorithm optimization progress. Δt is the algebraic change, which indicates the progress of time. ρ is the adjustment coefficient, which controls the parameter of the perturbation adjustment rate to adapt to the dynamically changing optimization environment.

[0112] The above process can effectively help the genetic algorithm escape from the local optimum while maintaining the effectiveness and efficiency of the algorithm in the global search process. It increases the probability of the algorithm finding the global optimal solution and enhances the accuracy and reliability of the solution.

[0113] By introducing multi-dimensional influencing factors such as environmental conditions, technician skill levels, and historical equipment failure rates, the multi-dimensional situation response scheduling algorithm can more accurately match task requirements and site conditions and optimize resource allocation. The dynamic weight adjustment mechanism ensures that changes in various influencing factors are reflected in real time during the task allocation process, making scheduling decisions more flexible and adaptable.

[0114] By applying an optimized allocation model and adjusting task allocation, we ensure that the skills of technicians match the task requirements and improve work efficiency. At the same time, we take into account labor costs and resource allocation and reduce overall operating costs. Dynamic availability and multi-task fairness constraints ensure that technicians will not be overworked and maintain work quality and efficiency. The introduced precision enhancement adjustment algorithm enhances the adaptability to environmental changes and potential errors through periodic sinusoidal waveform changes and genetic diversity maintenance mechanisms.

[0115] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A scheduling optimization method based on automatic detection of low-voltage metering boxes, characterized in that: The following steps are involved: S1: Based on the goals and / or functional requirements of the low-voltage meter box automated detection, obtain detection data, pre-process the detection data, and extract features based on the pre-processed detection data to obtain key feature data; S2: Based on the dynamic weight analysis enhancement algorithm, key feature data is enhanced. By dynamically adjusting the weights, the representation of data features is optimized to obtain intelligently processed data. S3: Based on the intelligent fusion scheduling algorithm, it identifies the current status of the low-voltage meter box, builds an optimized allocation model, and intelligently allocates detection tasks; S4: Based on the multi-dimensional situation response scheduling algorithm, the allocation of detection tasks is adaptively optimized to obtain the optimal detection task allocation; In S3, the optimization allocation model construction method includes: Define decision variables: Define decision variables Expressed as time t No. i Is the technician assigned to the j tasks, 1 means assigned, 0 means not allocated; Define the objective function; construct the objective function to maximize the weighted completion rate of task priorities and minimize the total cost, as follows: （4）； Where, F is the objective function, p j For the task j The priority weight, For the i technicians at a time t Execute the task j the cost, N is the total number of technical personnel, M is the total number of tasks that need to be assigned, T is the time period; In S4, the multi-dimensional situation response scheduling algorithm includes: Define multi-dimensional influencing factors, define influencing factors and quantification methods of influencing factors, and construct matrix representation of quantitative variables. For the task j In time t A vector of influencing factors, including environmental factors, operator skill level, and historical failure rate of equipment; Weight adjustment of influencing factors: For each task, the weight is dynamically adjusted according to the task influencing factor vector. The formula is as follows: （7）； Where, is the weight of the task in time, For the k The weight coefficient of each influencing factor, For the j Tasks in time t The next k The quantitative value of the influencing factors, is the weight coefficient of other influencing factors, is the quantitative value of other influencing factors.

2. A scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S1, the preprocessing of the detection data includes data cleaning, missing value processing, noise smoothing and data normalization, and the key feature data includes statistical features, trend features and frequency domain features.

3. The scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 1 is characterized in that: In S2, the dynamic weight analysis enhancement algorithm includes: Initialize the weight matrix: Assign an initial weight to each data feature. The weight matrix is ​​expressed as W , the formula is as follows: （1）； Where, w i For the i The initial weights of the features, n is the number of features; Calculate weighted eigenvalues: Use weight matrix to calculate the weighted eigenvalues ​​of feature data. X Weighted, each eigenvalue x i The corresponding weight w i Multiplying, the formula is as follows: （2）； Where, Z is the weighted eigenvector, which represents the comprehensive index of the eigenvalue after weight processing. x i For the i The original value of the feature.

4. A scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 3, characterized in that: Dynamically adjust weights: Update the weight of each feature value based on its utility. The formula is as follows: （3）； Where, is the updated weight, w i For the i The current weight of the feature, exp is an exponential function used to calculate the update scale of the weights, For the i The standard deviation of the features adjusts the sensitivity of weight updates. and Respectively i and j The utility score of each feature, n is the total number of features; Based on the data fusion algorithm, the enhanced feature data and the preprocessed data are fused to obtain intelligently processed data.

5. The scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 1 is characterized in that: Constraints: Each task must be assigned at least once within the specified time, and each technician can only perform one task at any time. The formula is as follows: （5）； （6）。 6. The scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 1 is characterized in that: Optimize the objective function, the formula is as follows: （ 8）； Where, is the optimized objective function; Constraint expansion: Add dynamic availability constraints, skill matching constraints, task completion requirements, continuous working time limits, and multi-task fairness constraints; After optimizing the objective function and constraints, a heuristic algorithm combined with linear programming is used for optimization, a genetic algorithm is used to generate an initial solution, and then the initial solution is fine-tuned through linear programming to obtain the global optimal solution.

7. A scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 6, characterized in that: Based on the precision enhancement adjustment algorithm, the genetic algorithm is optimized to find a better solution in the global search space. The precision enhancement adjustment algorithm includes: Introduction of local search: By introducing local search, the accuracy and ability of the adjustment algorithm to escape when encountering a local optimal solution are enhanced; Implement local perturbation: In each generation of the genetic algorithm, a specified proportion of individuals are selected for local perturbation. The perturbation is in the form of r A periodic sinusoidal waveform is applied to the （9）； Where, is the original gene value, is the disturbance amplitude, controlling the size of the change, To control the periodicity of the disturbance and determine the frequency of the disturbance, is the new gene value after perturbation adjustment, is the phase offset.

8. The scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 7 is characterized in that: Maintaining genetic diversity: Calculate the differences between individuals in a population and use the Hamming distance as a measure of differences to achieve diversity assessment; When the population diversity is lower than the preset threshold, the diversity is increased by adding randomly generated new individuals or producing more diverse offspring through crossover and mutation. The formula is as follows: （10）； Where, For individuals Average with the population The Hamming distance of is a positive regulatory factor used to adjust the impact of diversity. is the diversity index of the new generation population, For the i The new gene value of each individual, Num is the number of individuals in the population.

9. The scheduling optimization method based on automatic detection of low-voltage metering boxes according to claim 8 is characterized in that: Dynamically adjust the perturbation amplitude: dynamically adjust the perturbation amplitude according to the progress of the search process, and adjust the perturbation amplitude according to the changes in the optimal solution in consecutive generations ; If the optimal solution is not improved for a time exceeding the set threshold, increase the perturbation amplitude and expand the exploration area. If the optimal solution is not improved for a time less than the set threshold, reduce the perturbation amplitude and refine the search in the current area. The formula is as follows: （11）； Where, is the perturbation amplitude used in the previous generation, is the newly calculated perturbation amplitude used for the next generation of local search, is the difference between the optimal fitness of the previous and next generations, indicating the speed of algorithm optimization progress. is an algebraic variable, indicating the progress of time, To adjust the coefficient, control the parameters of the disturbance adjustment rate to adapt to the dynamically changing optimization environment.

Images