Method and system for dynamic planning of junction box maintenance path based on multi-objective optimization

Abstract

The application relates to the technical field of power distribution equipment maintenance management, and discloses a multi-target optimization-based dynamic planning method and system for the maintenance path of a branch box, which has the technical scheme as follows: an enhanced task list, a composite travel time matrix and a collaborative maintenance reward calculation function are obtained through temperature sensing equivalent service time modeling, communication blind area task closed loop construction and multi-loop collaborative maintenance reward; a non-dominated path scheme set is obtained through temperature communication sensing weighted clustering, cluster level dynamic planning and cluster internal dynamic planning screening through a Pareto frontier; an updated complete maintenance path scheme is obtained through reverse path re-inspection value evaluation and residual task re-planning; and an updated model parameter configuration file and system performance evaluation report are obtained through statistical fitting and parameter rolling update. The application unifies temperature, communication and collaborative maintenance constraints into a multi-target dynamic planning framework, realizes high-precision adaptation of the maintenance path to actual working conditions and improves operation and maintenance efficiency.

Description

Technical Field

[0001] This invention relates to the field of power distribution equipment maintenance management technology, specifically to a dynamic planning method and system for junction box maintenance paths based on multi-objective optimization. Background Technology

[0002] With the continuous expansion of power distribution networks and the widespread deployment of smart metering devices, the workload of daily inspections and emergency repairs of terminal power distribution equipment such as junction boxes and metering boxes has increased significantly. In the field of operation and maintenance scheduling, path optimization-based maintenance planning methods have been widely used. These methods typically aim to minimize travel distance or total time, combining geographic information systems and vehicle path planning models to generate maintenance sequences. However, existing technologies often treat the processing time of each maintenance task as a fixed parameter when planning paths, rarely considering the impact of differences in equipment operating environments on actual operation time, and failing to fully reflect the interference of on-site constraints such as additional walking and communication data synchronization by maintenance personnel in different work scenarios on the overall travel time chain. In addition, existing independent task scheduling methods lack effective quantitative evaluation and optimization methods for the correlation between multiple outgoing circuits within the same box and their potential impact on the economic efficiency of maintenance decisions. These factors lead to significant deviations between maintenance plans generated by existing methods and on-site execution during actual operation and maintenance. The overall efficiency of scheduling and power supply reliability still have room for improvement, thus existing technologies have shortcomings. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention aims to provide a dynamic planning method and system for junction box maintenance paths based on multi-objective optimization. This method obtains an enhanced task list and a composite travel time matrix through temperature-sensing equivalent service duration modeling, communication blind zone task closed-loop construction, and multi-loop collaborative maintenance rewards. It then obtains a set of non-dominated path solutions based on temperature-sensing weighted clustering, cluster-level dynamic programming, and nested intra-cluster dynamic programming. The complete maintenance path solution is updated based on reverse path re-inspection value assessment and remaining task re-planning. Finally, it achieves adaptive parameter optimization through statistical fitting and rolling parameter updates. This solves the problems of large deviations between maintenance plans and on-site execution, and low scheduling efficiency in existing maintenance and operation management technologies due to neglecting environmental temperature differences, on-site communication constraints, and the benefits of multi-loop collaborative maintenance. This improves the time window satisfaction rate and overall operation and maintenance economy of junction box maintenance path planning under actual high-temperature conditions.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a dynamic planning method for junction box maintenance paths based on multi-objective optimization, comprising: The basic data set is obtained, and the enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function are obtained through temperature sensing equivalent service duration modeling, communication blind zone task closed loop construction and multi-loop collaborative maintenance reward. Based on the enhanced task list, composite travel time matrix, and collaborative maintenance reward calculation function, a complete state transition model is obtained through multi-objective normalization and state space definition. The model is solved by nested solutions of temperature communication sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming. A set of non-dominated path schemes is obtained by Pareto front screening. Based on the set of non-dominated path schemes, combined with real-time status snapshots and a list of new tasks, an updated complete maintenance path scheme is obtained through reverse path re-inspection value assessment and remaining task replanning. Based on the updated complete maintenance path scheme, the updated model parameter configuration file and system performance evaluation report are obtained through statistical fitting and rolling parameter updates.

[0005] Furthermore, the acquisition of the basic data set, through temperature-sensing equivalent service duration modeling, communication blind spot task closed-loop construction, and multi-loop collaborative maintenance reward, yields an enhanced task list, a composite travel time matrix, and a collaborative maintenance reward calculation function, including: Based on a multi-source heterogeneous system, a basic data set is obtained through data extraction, field mapping and format unification rules, including a standardized task list, equipment ledger dictionary, electrical topology directed graph, circuit adjacency list, work group list, road network configuration parameter set, meteorological time series data and regional communication signal strength heat map; Based on the equipment ledger dictionary and meteorological time series data, the equivalent service duration function of temperature rise sensing for each junction box maintenance task is obtained through material thermal inertia coefficient mapping and surface temperature linear estimation model. Based on the standardized task list, regional communication signal strength heat map, and work group list, the maintenance task execution process and composite travel time matrix after communication constraint correction are obtained through signal threshold determination, nearest communication reachable point search, and walking segment accumulation method. Based on the aforementioned loop adjacency list and standardized task list, the collaborative maintenance reward calculation function is obtained through loop health weighted scoring, thermal coupling degradation coefficient calculation, and collaborative maintenance reward formula. Based on the temperature rise sensing equivalent service duration function and the maintenance task execution process modified by communication constraints, an enhanced task list is obtained.

[0006] Furthermore, the step of obtaining the equivalent service duration function for temperature rise sensing of each junction box maintenance task based on the equipment ledger dictionary and meteorological time-series data through material thermal inertia coefficient mapping and surface temperature linear estimation model includes: Based on the box material field in the equipment ledger dictionary, the list of equipment labeled with material thermal inertia coefficient is obtained through the material thermophysical property lookup table; Based on the latitude and longitude coordinates of the equipment and meteorological time series data, the orientation factor calculation rules for each piece of equipment are obtained through geographic information system shadow analysis and solar azimuth angle calculation. Based on the material thermal inertia coefficient labeling equipment list, orientation factor calculation rules, meteorological time series data, and preset safety operation thresholds, the temperature rise sensing equivalent service duration function is obtained through the surface temperature linear superposition formula and the simplified formula of Newton's law of cooling.

[0007] Furthermore, based on the standardized task list, regional communication signal strength heatmap, and work group list, the maintenance task execution process and composite travel time matrix after communication constraint correction are obtained through signal threshold determination, nearest communication reachable point search, and walking segment accumulation method, including: Based on the latitude and longitude coordinates of the devices in the standardized task list and the heat map of regional communication signal strength, a communication scenario classification label is obtained by determining the signal threshold. Based on the coordinates of devices in communication blind spots and the heat map of regional communication signal strength, the coordinates of the nearest reachable communication point for each device in a communication blind spot are obtained through signal satisfaction point search and open area priority rules. Based on the coordinates of the nearest reachable point, the maintenance task execution process and composite travel time matrix after communication constraint correction are obtained by calculating the composite walking path segment and determining the overlap of the spatial buffer.

[0008] Furthermore, based on the loop adjacency list and standardized task list, the collaborative maintenance reward calculation function is obtained through loop health weighted scoring, thermal coupling degradation coefficient calculation, and collaborative maintenance reward formula, including: Based on the load data and fault history records of smart meters, the health score of each circuit is obtained by extracting the operation indicators of each circuit and weighting and summing them. Based on the physical installation distance and common busbar marking in the loop adjacency table, the coupling degradation coefficient of each faulty loop to adjacent loops in the same enclosure is calculated using the thermal coupling degradation coefficient formula. Based on the circuit health score, coupling degradation coefficient, and preset circuit load importance weight, the collaborative maintenance reward calculation function is obtained through the collaborative maintenance reward formula.

[0009] Furthermore, based on the enhanced task list, composite travel time matrix, and collaborative maintenance reward calculation function, a complete state transition model is obtained through multi-objective normalization and state space definition. This model is then solved using nested solutions of temperature communication-sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming. A set of non-dominated path solutions is obtained through Pareto front screening, including: Based on the enhanced task list, composite travel time matrix, collaborative maintenance reward calculation function, and work group list, a complete state transition model containing cost function and constraint rules is obtained by constructing a normalized reference value through a single-objective greedy algorithm, defining a seven-tuple state space, and building a set of state transition feasibility check rules. Based on the enhanced task list, cluster analysis was performed by introducing a weighted anisotropy measure with a material thermal inertia difference penalty term to obtain multiple task clusters and task cluster division results; Based on the task cluster partitioning results, composite travel time matrix and complete state transition model, cluster access order is allocated through cluster-level dynamic programming, and then dynamic programming path search is performed on the tasks within each cluster, taking into account the equivalent service time of temperature rise perception and collaborative maintenance rewards, to obtain the optimal access order and estimated arrival time of tasks within the cluster. Based on the optimal access order and estimated arrival time of the tasks within the cluster, a set of non-dominated path schemes containing task execution order, estimated arrival time, and collaborative maintenance suggestions is obtained through a non-dominated sorting method.

[0010] Furthermore, based on the enhanced task list, composite travel time matrix, collaborative maintenance reward calculation function, and work group list, a complete state transition model containing cost functions and constraint rules is obtained by constructing a normalized reference value through a single-objective greedy algorithm, defining a seven-tuple state space, and building a set of state transition feasibility checks. This includes: A greedy construction algorithm was run with the single objective of minimizing the total effective travel time, minimizing the weighted repair waiting time, minimizing the imbalance of the workload of the shift team, and maximizing the cumulative collaborative maintenance reward value, resulting in four normalized reference values. Based on the preset four-dimensional weight coefficients and the four normalized reference values, a multi-objective weighted total cost function is constructed through a linear weighted sum formula to obtain the single-step cost calculation rule for state transition; the four-dimensional weight coefficients include efficiency weight, reliability weight, equilibrium weight, and collaborative reward weight. Based on the enhanced task list, the state space is defined in the form of a seven-tuple, and four types of sequential check rules are constructed: feasibility of temperature and time window, feasibility of communication closed loop, feasibility of personnel role, and feasibility of collaborative maintenance time consumption. This results in a set of state transition feasibility check rules, and thus a complete state transition model.

[0011] Furthermore, the updated complete maintenance path plan, obtained by combining the non-dominated path scheme set with real-time status snapshots and a list of newly added tasks, through reverse path re-inspection value assessment and remaining task replanning, includes: Based on the set of non-dominated path schemes, a real-time state snapshot and a set of re-inspection point markers are obtained by state variable aggregation and breadth-first search tracing. Based on the real-time status snapshot, the list of newly added tasks, and the set of re-inspection point markers, a set of replanning tasks is obtained by merging the sets, and an updated complete maintenance path scheme is obtained based on the set of replanning tasks.

[0012] Furthermore, based on the updated complete maintenance path scheme, the updated model parameter configuration file and system performance evaluation report are obtained through statistical fitting and parameter rolling updates, including: Based on the updated complete maintenance path scheme, an execution record dataset is obtained through data cleaning and structured storage; Based on the execution record dataset, an updated model parameter configuration file is obtained through median statistics, linear regression fitting, and comparative analysis. Based on the execution record dataset, a system performance evaluation report is obtained by calculating the time window satisfaction rate, travel time prediction accuracy, and actual benefits of collaborative maintenance.

[0013] Secondly, the present invention provides a dynamic planning system for junction box maintenance paths based on multi-objective optimization, comprising: Data Enhancement Module: Used to acquire basic data sets, and obtain an enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function through temperature sensing equivalent service duration modeling, communication blind zone task closed-loop construction and multi-loop collaborative maintenance reward; Multi-objective planning module: Based on the enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function, it obtains a complete state transition model through multi-objective normalization and state space definition, solves it through nested solutions of temperature communication sensing weighted clustering, cluster-level dynamic programming and intra-cluster dynamic programming, and obtains a set of non-dominated path schemes through Pareto front screening; Online replanning module: Based on the set of non-dominated path schemes, combined with real-time status snapshots and a list of newly added tasks, it obtains an updated complete maintenance path scheme through reverse path re-inspection value assessment and remaining task replanning; Closed-loop optimization module: Based on the updated complete maintenance path scheme, it is used to obtain the updated model parameter configuration file and system performance evaluation report through statistical fitting and rolling parameter updates.

[0014] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: This invention utilizes temperature-sensing equivalent service duration modeling, communication blind spot task closed-loop construction, and multi-loop collaborative maintenance rewards to form an enhanced task list and composite travel time matrix. Based on multi-objective normalization and temperature-communication-sensing weighted clustering, and cluster-level dynamic programming nested solution, it obtains non-dominated path solutions. It unifies temperature, communication, and collaborative maintenance constraints into a multi-objective optimization framework, solving problems such as inaccurate maintenance time prediction, failure to account for communication backtracking, and difficulty in quantifying collaborative benefits, thus improving the time window satisfaction rate and travel prediction accuracy of path planning. Furthermore, through refined methods such as material thermal inertia difference modeling, communication reachability search, and loop health coupling assessment, it enhances the model's adaptability to high-temperature operations, complex communication environments, and multi-loop degradation effects, reducing the frequency of repeated dispatches and maintenance costs. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the steps of the dynamic planning method for junction box maintenance paths based on multi-objective optimization in this invention. Figure 2 This is a flowchart illustrating the steps involved in solving the problem using nested dynamic programming and temperature-sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming in this embodiment. Figure 3 This is a flowchart illustrating the steps involved in obtaining the updated complete maintenance path scheme in this embodiment. Figure 4 This is a schematic diagram of the structure of the junction box maintenance path dynamic planning system based on multi-objective optimization of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof.

[0017] The term "and / or" in the following text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0018] Example 1: like Figure 1 As shown, this invention provides a dynamic planning method for junction box maintenance paths based on multi-objective optimization, including: The basic data set is obtained, and the enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function are obtained through temperature sensing equivalent service duration modeling, communication blind zone task closed loop construction and multi-loop collaborative maintenance reward. Based on the enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function, a complete state transition model is obtained through multi-objective normalization and state space definition. The model is solved by nested solution of temperature communication sensing weighted clustering, cluster-level dynamic programming and intra-cluster dynamic programming. The set of non-dominated path schemes is obtained by Pareto front screening. Based on the set of non-dominated path schemes, combined with real-time status snapshots and a list of new tasks, an updated complete maintenance path scheme is obtained through reverse path re-inspection value assessment and remaining task replanning. Based on the updated complete maintenance path scheme, the updated model parameter configuration file and system performance evaluation report are obtained through statistical fitting and rolling parameter updates.

[0019] The temperature sensing equivalent service duration modeling involves calculating the required cooling time for the junction box at any arrival time based on the material thermal inertia coefficient of the junction box's enclosure material, equipment orientation factor, and real-time environmental meteorological data. This calculation is achieved using a linear superposition formula for surface temperature and a simplified formula based on Newton's law of cooling. This cooling time is then superimposed on a baseline maintenance time to obtain a dynamically changing temperature rise sensing equivalent service duration function. This function is used in path planning to reflect the impact of differences in maintenance time between metal and sheet molding compound enclosures due to high-temperature exposure on the overall task timing. The communication blind zone task closed-loop construction involves determining the location of the junction box based on the long-distance radio received signal strength indicator. If the junction box is in a communication signal dead zone, a geographic information system is used to search for the nearest reachable communication point around the junction box that meets the signal threshold conditions. Based on this reachable point, a closed-loop atomic subtask sequence is constructed, including a physical maintenance subtask, a walking-to-communication-point subtask, a data synchronization subtask, and a walking-back subtask. Simultaneously, the sequence is modified for exemption based on the coverage range of the portable repeater carried by the work team, resulting in a communication constraint-corrected maintenance task execution flow and a composite travel time matrix for personnel and vehicle separation. This matrix explicitly incorporates the additional walking and data synchronization time caused by the communication dead zone into the cost structure of the path planning. The multi-circuit collaborative maintenance reward is based on the number of outgoing lines within the same junction box. Historical temperature rise, number of faults, load fluctuations, and service life of each circuit are weighted to obtain a health score for each circuit. The coupling degradation coefficient of a faulty circuit to adjacent circuits within the same enclosure is calculated based on the physical installation distance between circuits and their shared busbar relationship. When repairing a faulty circuit at a junction box, the reward value for optional collaborative repair actions is calculated based on the health scores of adjacent circuits, the coupling degradation coefficient, and the importance weight of the circuit load. This reward is used in dynamic programming state transitions to reduce the overall cost of simultaneously repairing multiple adjacent low-healthy circuits, thereby achieving a long-term economic benefit assessment of covering multiple potential repair needs with a single dispatch. Temperature communication sensing weighted clustering is based on the composite travel time between tasks. A weighted dissimilarity metric, using the product of the difference in thermal inertia coefficients of equipment materials and the average daily solar radiation temperature rise during the planned maintenance period as a penalty term, is introduced to perform cluster analysis on the maintenance tasks in the enhanced task list. Tasks in communication blind spots are prioritized to be clustered separately, resulting in several task clusters and task cluster partitioning results. This is used to group tasks with similar temperature response characteristics and communication constraints into the same solution unit to reduce the state space size of subsequent dynamic programming. The cluster-level dynamic programming is performed recursively using the assigned cluster bitmap and the current shift number as the state and the composite travel time between cluster centers as the transition cost to obtain the access order of the task clusters assigned to each shift. This is used to coordinate the allocation and access order of task clusters among multiple shifts.Intra-cluster dynamic programming involves using a seven-tuple state space within the task clusters assigned to each work group. This space includes the visited equipment location map, current equipment number, current time, current work group number, work group personnel role occupancy state matrix, repeater working state Boolean values, and the set of executed collaborative maintenance loops. During state transitions, the equivalent service duration function for temperature rise sensing, the maintenance task execution flow after communication constraint correction, and the collaborative maintenance reward calculation function are called sequentially. A memoized search is performed using a multi-objective weighted total cost function as the basis for state transition costs to obtain the optimal access order, estimated arrival time, and collaborative maintenance within the cluster. The proposed method is to achieve precise path optimization at the cluster task level, considering the triple constraints of temperature, communication, and collaborative maintenance. The reverse path re-inspection value assessment is conducted when dynamic replanning is triggered. For newly faulty equipment, the method traces upstream through the directed electrical topology graph to maintain the serviced equipment, calculating the electrical topology distance between the maintained upstream equipment and the newly faulty equipment, as well as the additional time cost of returning for re-inspection. The reverse path re-inspection value is obtained based on the ratio of the faulty equipment's importance weight to the aforementioned electrical topology distance and additional time cost. This value is used to include high-value re-inspection points in the replanning task set to realize opportunities for forward re-inspection of maintained upstream equipment. The remaining task replanning uses the current position in the current real-time state snapshot as a virtual starting point. It merges the set of unexecuted tasks, newly added tasks, and high-value re-inspection point markers obtained through reverse path re-inspection value assessment into a replanning task set. It then uses temperature communication sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming to resolve the problem. The newly generated path is then concatenated with the already traveled path to obtain an updated complete maintenance path scheme. This is used to quickly generate adjustment schemes adapted to the current working conditions when unexpected tasks are inserted or execution deviates. Statistical fitting is based on the actual travel trajectory collected after the scheme is executed. The system tracks the arrival time of each task, the equivalent service duration, the communication synchronization time, and the collaborative maintenance execution records. It updates the basic walking distance parameters, floor conversion factor, material thermal inertia coefficient, and coupling degradation coefficient using median statistics and linear regression fitting methods, respectively. This ensures that the model parameters gradually approximate real physical laws based on actual operating data. The rolling parameter update involves feeding back the updated parameter values ​​obtained through statistical fitting to the temperature sensing equivalent service duration modeling, the communication blind spot task closed-loop construction, and the multi-loop collaborative maintenance reward calculation via configuration file overwriting. This forms a closed-loop iterative mechanism for parameter adaptive optimization.

[0020] This embodiment transforms the thermal inertia difference of the junction box material into an equivalent service duration dependent on arrival time. By determining the reachable points of junction boxes in communication blind spots and constructing a closed-loop task sequence that includes physical maintenance and data synchronization, by establishing the thermal coupling degradation relationship between circuits and calculating the collaborative maintenance reward value, and by embedding the above three constraints into the dynamic programming state transition cost function and using clustering and nested two-level dynamic programming to solve the problem, it realizes the joint dynamic programming of temperature perception, communication perception and collaborative maintenance perception for the maintenance path of distribution network junction boxes under a multi-objective optimization framework. This improves the time window satisfaction rate of the maintenance path scheme under actual high-temperature conditions, reduces the invalid travel mileage caused by secondary backtracking in communication blind spot tasks, and reduces the frequency of repeated dispatching caused by ignoring the thermal coupling degradation between circuits, thereby improving the overall operational efficiency and power supply reliability of junction box maintenance scheduling.

[0021] Furthermore, this embodiment provides a step-by-step approach to obtain a basic data set, and to generate an enhanced task list, a composite travel time matrix, and a collaborative maintenance reward calculation function through temperature-sensing equivalent service duration modeling, communication blind spot task closed-loop construction, and multi-loop collaborative maintenance rewards. The steps include: Based on a multi-source heterogeneous system, a basic data set is obtained through data extraction, field mapping and format unification rules, including a standardized task list, equipment ledger dictionary, electrical topology directed graph, circuit adjacency list, work group list, road network configuration parameter set, meteorological time series data and regional communication signal strength heat map; Based on the equipment ledger dictionary and meteorological time series data, the equivalent service duration function for temperature rise sensing of each junction box maintenance task is obtained through material thermal inertia coefficient mapping and surface temperature linear estimation model. Specifically, this includes: obtaining a list of equipment labeled with material thermal inertia coefficients based on the box material field in the equipment ledger dictionary through a material thermophysical property lookup table; obtaining the orientation factor calculation rules for each device based on the equipment latitude and longitude coordinates and meteorological time series data through geographic information system shadow analysis and solar azimuth angle calculation; and obtaining the equivalent service duration function for temperature rise sensing through the surface temperature linear superposition formula and the simplified formula of Newton's law of cooling based on the material thermal inertia coefficient labeled equipment list, orientation factor calculation rules, meteorological time series data, and preset safety operation thresholds. Based on a standardized task list, a regional communication signal strength heatmap, and a work team list, the maintenance task execution process and composite travel time matrix after communication constraint correction are obtained through signal threshold determination, nearest communication reachable point search, and walking segment accumulation methods. Specifically, this includes: obtaining communication scenario classification labels based on the latitude and longitude coordinates of equipment in the standardized task list and the regional communication signal strength heatmap through signal threshold determination; obtaining the coordinates of the nearest communication reachable point for each communication blind zone equipment based on the coordinates of equipment in the communication blind zone and the regional communication signal strength heatmap through signal satisfaction point search and open area priority rules; and obtaining the maintenance task execution process and composite travel time matrix after communication constraint correction based on the coordinates of the nearest communication reachable point through composite walking path segment calculation and spatial buffer overlap determination. Based on the loop adjacency list and standardized task list, a collaborative maintenance reward calculation function is obtained through loop health weighted scoring, thermal coupling degradation coefficient calculation, and collaborative maintenance reward formula. Specifically, this includes: obtaining the health score of each loop by extracting operating indicators from each loop and weighted summing them based on smart meter load data and fault history records; calculating the coupling degradation coefficient of each faulty loop to adjacent loops in the same enclosure based on the physical installation distance and common busbar marking in the loop adjacency list using the thermal coupling degradation coefficient formula; and obtaining the collaborative maintenance reward calculation function based on the loop health score, coupling degradation coefficient, and preset loop load importance weights using the collaborative maintenance reward formula. An enhanced task list is obtained based on the maintenance task execution process modified by the temperature rise sensing equivalent service duration function and communication constraints.

[0022] The standardized task list is a structured data set containing task number, associated equipment number, fault circuit number, fault occurrence time, and scheduled time window, generated by matching work order data from the fault reporting system and inspection plan system through equipment number association. The equipment ledger dictionary is a set of equipment attribute data extracted from the production management system, including equipment number, equipment type, enclosure material, latitude and longitude coordinates, installation location text, number of outgoing circuits, and load level of each circuit. The electrical topology directed graph is a directed graph structure representing the power supply hierarchy between equipment, constructed based on key-value pairs of the list of superior and subordinate equipment numbers recorded in the power distribution geographic information system. The circuit adjacency table is an adjacency table structure that records the physical installation distance between any two circuits within the same equipment and whether they share a busbar, with the physical installation distance in centimeters. The regional communication signal strength heat map is a heat map generated by spatial overlay analysis of operator narrowband IoT coverage grid data and long-distance radio gateway configuration library, based on latitude and longitude grids. The grid represents the signal strength distribution of the unit; the material thermophysical property reference table is a pre-defined mapping table between the enclosure material and the material thermal inertia coefficient. The material thermal inertia coefficient is obtained by proportionally normalizing the measured data of the thermal conductivity of the enclosure material. Typically, the thermal inertia coefficient corresponding to stainless steel is set to 1.2, that of sheet molding compound fiberglass is set to 0.7, and that of polycarbonate is set to 0.4. The material thermal inertia coefficient characterizes the difference in surface temperature rise rate of enclosures of different materials under the same solar radiation conditions; the surface temperature linear superposition formula is a surface temperature estimation formula that linearly superimposes the ambient temperature, the solar radiation temperature rise reference value, the material thermal inertia coefficient, and the orientation factor under the assumption of steady-state thermal equilibrium. The solar radiation temperature rise reference value is obtained from the measured value of the additional heat generated by solar radiation on the equipment surface at noon on a sunny day under typical meteorological conditions. It is usually taken as 15℃, which is the typical additional heat generated by solar radiation on the equipment surface at noon on a sunny day.

[0023] The safe operating threshold is set according to the high-temperature operation classification of the power industry, and is usually set at 45℃. When the surface temperature of the equipment exceeds this threshold, maintenance personnel must wait for the enclosure to cool down before they can operate it. The simplified formula of Newton's law of cooling is a formula for calculating the waiting time to approximate the natural convection heat dissipation process as a linear cooling process. The natural convection heat dissipation coefficient is usually set at 0.15℃ / min, the thermal conductivity of stainless steel is usually set at 16.2, and the thermal conductivity of sheet molding compound fiberglass is usually set at 0.4.

[0024] The communication scenario classification marker is a marker value determined by comparing the long-distance radio received signal strength indication value (RSSI) of the device's location with a preset long-distance radio signal blind zone threshold, where the long-distance radio signal blind zone threshold is -120dBm. The communication reachable point is a coordinate point within a preset radius around the device in the communication blind zone that satisfies an RSSI greater than or equal to the preset long-distance radio signal blind zone threshold. If no coordinate point satisfies the condition within the preset radius, the nearest node in the geographic information system with the land use type of open area is taken, where the preset radius is usually 100m.

[0025] The circuit health score is a numerical value ranging from 0 to 100, with lower values ​​indicating more severe circuit degradation. It is obtained by extracting the temperature rise amplitude, number of failures, load peak-to-valley fluctuation amplitude, and years of operation for each circuit, and then weighting and summing them accordingly. The thermal coupling degradation coefficient formula is based on the exponential decay law of heat conduction with distance and the enhanced electromagnetic coupling effect of the common busbar, and its form is: ,in, This is the coupling degradation coefficient. Based on the coupling coefficient, For distance attenuation factor, Physical installation distance, Add a coupling coefficient to the common busbar. This is a common busbar indicator function, taking the value 1 when two circuits share a busbar, and 0 otherwise. The circuit load importance weight is a preset weight value based on the load level of the circuit. The weight is obtained from a preset load level and weight mapping table based on the highest load level of the circuit: Level 1 load corresponds to a weight of 0.9; Level 2 load corresponds to a weight of 0.6; and Level 3 load corresponds to a weight of 0.3. If the same circuit carries multiple load levels simultaneously, the weight is determined by the highest load level, and these weights are not accumulated repeatedly.

[0026] Specifically, data extraction tools are used to extract equipment ledger data, electrical topology relationship data, work order data, team personnel data, environmental meteorological data, road network driving route data, and communication signal coverage data from the production management system, power distribution geographic information system, fault reporting system, inspection plan system, human resource management system, meteorological service application interface, high-precision map service interface, and telecommunications operator data platform. The extracted raw data undergoes field mapping, unifying fields with inconsistent names but identical meanings across different source systems to preset standard field names. Missing values ​​are handled according to preset filling rules, with missing installation location text filled with an empty string by default, and missing floor information filled with 1 by default. All data is uniformly converted to the WGS84 latitude and longitude coordinate system. After the above processing, a standardized task list is formed, where each task record includes a task number, associated equipment number, fault circuit number, fault occurrence time, and scheduled time window. An equipment ledger dictionary is also formed, where each equipment record includes equipment number, equipment type, enclosure material, etc. The system generates the following data: latitude and longitude coordinates, installation location text, number of outgoing circuits, and load level of each circuit; it also generates a directed electrical topology graph, where nodes represent equipment numbers and directed edges represent power supply relationships between higher and lower level equipment; it generates a circuit adjacency list, using equipment numbers as indexes to record the physical installation distance and common busbar markings between any two circuits within a given equipment; it generates a work group list, using work group numbers as indexes to record the members of each work group and the number of portable repeaters carried; it generates a road network configuration parameter set, including the map service interface's call address, key, and request frequency limit parameters; and it generates meteorological time-series data, including hourly timestamps and ambient temperature values. And solar radiation intensity values; forming a regional communication signal intensity heat map, including the mean RSSI or mean RSRP values ​​in 50m×50m grid units.

[0027] Then, each equipment record in the equipment ledger dictionary is traversed, the value of the enclosure material field is extracted, and a preset material thermophysical property lookup table is consulted. This table includes the following: if the enclosure material is stainless steel, the thermal inertia coefficient is assigned a value of 1.2; if it is sheet molding compound fiberglass, a value of 0.7 is assigned; and if it is polycarbonate, a value of 0.4 is assigned. The assigned values ​​are stored in the equipment list labeled with material thermal inertia coefficients. For each piece of equipment, based on its latitude and longitude coordinates and the timestamp in the meteorological time series data, the shading situation of the equipment relative to the sun at that moment is calculated using the shadow analysis function in the geographic information system. Combined with the solar azimuth angle, the current state of the equipment—whether it is in western sunlight, direct sunlight, or shade—is determined. If it is in western sunlight, the orientation factor is assigned a value of 1.2; if it is in direct sunlight, a value of 1.0 is assigned; and if it is in shade, a value of 0.5 is assigned. Based on the above-obtained equipment list labeled with material thermal inertia coefficients, the orientation factor calculation rules, and the meteorological time series data, a linear superposition formula for surface temperature is used. The estimated surface temperature of the calculation device at any arrival time, where This is an estimated value for the surface temperature of the equipment. This is the ambient temperature value. The baseline value for solar radiation temperature rise. The thermal inertia coefficient of the material. Orientation factor. If the calculated surface temperature estimate exceeds the safe operating threshold. Then, the cooling time can be calculated using the simplified formula of Newton's law of cooling. ,in, The coefficient of heat dissipation by natural convection. The thermal conductivity of the material is the coefficient of thermal conductivity of stainless steel. Take 16.2, the corresponding material of sheet molding compound fiberglass. Take 0.4; if the estimated surface temperature is less than or equal to the safe operating threshold, then wait for the cooling time. Set the equipment's baseline maintenance time. and and fixed safe operation preparation time Add them together to obtain the equivalent service duration of temperature rise sensing for the device at that arrival time. Since the waiting time for cooling depends on the estimated surface temperature of the device at the arrival time, and the estimated surface temperature of the device changes with the arrival time, the final result is a temperature rise sensing equivalent service duration function that dynamically changes with the arrival time.

[0028] Next, for each task in the standardized task list, the latitude and longitude coordinates of its associated device are extracted, and the communication signal strength value of the grid where the coordinates are located is queried in the regional communication signal strength heatmap. If the device's communication method is long-range radio and the long-range radio received signal strength indication (RSSI) is less than the preset long-range radio signal blind zone threshold, the task is marked as a communication blind zone task; otherwise, it is marked as a communication reachable task. For each communication blind zone task, with the device coordinates as the center and a preset radius as the search radius, the grids that meet the signal threshold conditions are searched in the regional communication signal strength heatmap, and the center coordinates of the grid closest to the device coordinates that meets the signal threshold conditions are taken as the coordinates of the nearest communication reachable point of the device; if there are no grids that meet the signal threshold conditions within the search radius, the land use type data of the geographic information system is called to find the coordinates of the open area node with the land use type of square, green space, or road intersection closest to the device coordinates as the coordinates of the nearest communication reachable point. The installation location text determines the installation scenario category of the device. If the installation location contains keywords such as "utility pole" or "roadside," or the device type is a JP cabinet and the coordinates are within the road boundary, it is classified as a Category A roadside parking scenario. If the installation location contains keywords such as "stairwell," "unit," or "floor," it is classified as a Category B residential community walking scenario. If there are no accessible roads within the search radius centered on the device coordinates, it is classified as a Category C farmland without road network scenario. For Category A scenarios, the parking space coordinates are taken from the device's own coordinates; for Category B scenarios, the parking space coordinates are taken from the main entrance / exit coordinates of the residential community; for Category C scenarios, the parking space coordinates are taken from the coordinates of the nearest accessible road node. Based on the parking space coordinates, device coordinates, and the coordinates of the nearest communicably accessible point, the walking time from the parking space to the device is calculated. Walking time from the device to the communication reachable point and walking time from communication access point to parking space Walking speed Take a speed of 5 km / h. For any two task points... and Call the high-precision map service interface to obtain data from the task points. Parking space coordinates to task point Driving time at parking space coordinates The combined travel time between the two task points is... ,in For the task point Walking return time, For the task point Walking entry time, For the task point Walking time from the device to a point where communication is accessible. For the task point Walking time from a communication-reachable point back to the parking space, the latter two only if the task point... For tasks in communication blind spots and not covered by repeaters, the combined travel time between all pairs of task points constitutes a composite travel time matrix for personnel and vehicles. Simultaneously, a closed-loop atomic sub-task sequence is constructed for each task: for communication-reachable tasks, the sequence includes walking in, physical maintenance, in-situ data synchronization, and walking back; for communication blind spot tasks where the team does not carry a portable repeater or the equipment is not within the repeater's coverage radius, the sequence includes walking in, physical maintenance, walking to the communication point, data synchronization, and walking back, with data synchronization time taken as 2 minutes and in-situ data synchronization time taken as 1 minute; if the team carries a portable repeater and the equipment is within a 500m repeater coverage radius, the communication blind spot task is executed according to the sequence of communication-reachable tasks.

[0029] Next, for each circuit of each device recorded in the circuit adjacency table, the load curve of the past 30 days is extracted from the smart meter load data, and the ratio of peak-to-valley difference to average value is calculated as the load fluctuation amplitude. The number of faults that occurred in the circuit in the past 12 months is counted from the fault reporting system. The temperature rise amplitude of the most recent infrared thermometer record and the commissioning date of the circuit are extracted from the production management system to calculate the service life. The above four indicators are weighted and summed according to the following weights: temperature rise amplitude (40%), number of faults (25%), load fluctuation (20%), and service life (15%), respectively, to obtain a circuit health score ranging from 0 to 100. ,in For equipment number, Number the loops. For each pair of adjacent loops recorded in the loop adjacency table, extract their physical installation distance. and common busbar indicator function According to the thermal coupling degradation coefficient formula Calculate the coupling degradation coefficient of the faulty circuit to adjacent circuits. When the maintenance team arrives at a certain piece of equipment to perform faulty circuit maintenance, they traverse all other circuits within that equipment and select those with health scores. Less than the preset loop health score threshold The circuits constitute a set of low-health circuits. For each loop in the set of low-health loops, extract its loop load importance weight. Circuit health score and coupling degradation coefficient Through the collaborative maintenance reward formula The reward value is obtained for the collaborative maintenance actions that the device can choose to perform at the current arrival time. Additional time spent on collaborative maintenance actions. This is the sum of the maintenance baseline time for each additional circuit, where the maintenance baseline time for each additional circuit is determined based on the equipment type, and is typically taken as a value. .

[0030] Finally, each task in the standardized task list is associated with the temperature rise sensing equivalent service duration function, the maintenance task execution process after communication constraint correction, and the collaborative maintenance reward calculation function obtained in the above steps. For each task, based on the equipment type and installation scenario category of its associated equipment, the baseline maintenance duration, walking time parameters, and communication scenario classification of the task are determined; based on the enclosure material of its associated equipment, the material thermal inertia coefficient corresponding to the task is determined; based on its associated equipment number, the physical installation distance and common busbar relationship of each loop within the equipment are extracted from the loop adjacency table to determine the set of adjacent loops that need to be considered when making collaborative maintenance decisions for this task. The above information is organized into an enhanced task list indexed by the task number. Each task record in the enhanced task list contains, in addition to the fields in the original standardized task list, a reference to the temperature rise sensing equivalent service duration function, a sequence of task-closed-loop atomic subtasks, and a reference to the collaborative maintenance reward calculation function.

[0031] The natural convection heat dissipation coefficient is obtained by fitting engineering experience data and is usually taken as 0.15℃ / min; the long-distance radio signal blind zone threshold is obtained by the extremely weak signal threshold defined in the long-distance radio wide area network specification and is usually taken as -120dBm; the communication reachability search radius is obtained by statistical analysis of the average distance from the gate of a typical community to the farthest building and is usually taken as 100m; the portable repeater coverage radius is obtained by the measured effective coverage distance of long-distance portable radio repeaters in urban obstructed environments and is usually taken as 500m; the weights of each indicator in the circuit health score are allocated according to the defect level classification principle in the application specification for infrared diagnosis of live equipment and combined with the experience of operation and maintenance experts; the circuit health score threshold is calculated by converting the temperature rise standard of the boundary between general defects and severe defects in the application specification for infrared diagnosis of live equipment and is usually taken as 75; the basic coupling coefficient is obtained by statistical analysis of the time interval between successive failures of multiple circuits in the same enclosure in historical fault data. The values ​​are fitted and are typically set to 0.6; the distance attenuation factor is calibrated based on the physical law of heat conduction attenuation with distance and is typically set to 0.05; the common busbar additional coupling coefficient is obtained based on the comparative analysis of the fault correlation strength between common busbar circuits and non-common busbar circuits and is typically set to 0.3; the circuit load importance weight is quantified based on the power supply reliability requirements of each load level in the power industry load classification standard; the collaborative maintenance additional circuit maintenance benchmark time is obtained based on the statistical data of single circuit maintenance time for different types of equipment; the walking speed is obtained based on the measured value of the normal walking speed of an adult carrying lightweight tools and is typically set to 5 km / h; the data synchronization time is calculated based on the long-distance narrowband radio communication rate and typical data transmission volume, with on-site data synchronization typically set to 1 minute and communication point data synchronization typically set to 2 minutes; the fixed safe operation preparation time is obtained based on the statistical data of the average time spent on pre-operation safety confirmation and tool preparation as specified in the work instruction and is typically set to 3 minutes. In this embodiment, the values ​​of safe operation threshold, circuit health score threshold, long-distance radio signal blind zone threshold, and material thermal inertia coefficient are only examples, and those skilled in the art can set them according to actual conditions. This embodiment does not impose any restrictions on these values.

[0032] This embodiment transforms the material differences of the junction box into a temperature rise sensing equivalent service duration function that dynamically changes with the arrival time. By determining the communication reachable points for junction boxes in communication blind spots and constructing a closed-loop atomic subtask sequence containing physical maintenance and data synchronization subtasks, and by establishing the thermal coupling degradation relationship between loops and calculating the collaborative maintenance reward value, the above three enhancement constraints are uniformly associated with a standardized task list to form an enhanced task list. This achieves accurate modeling of maintenance tasks in three dimensions: temperature, communication, and multi-loop collaboration, improves the adaptability of subsequent path planning to actual working conditions, and reduces task timing deviations and repeated dispatch frequency caused by ignoring high-temperature waiting, communication backtracking, and inter-loop coupling degradation.

[0033] Furthermore, this embodiment provides a method based on an enhanced task list, a composite travel time matrix, and a collaborative maintenance reward calculation function. It obtains a complete state transition model through multi-objective normalization and state space definition, solves it using nested solutions of temperature communication-sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming, and obtains a set of non-dominated path solutions through Pareto front screening. The method includes: Based on an enhanced task list, a composite travel time matrix, a collaborative maintenance reward calculation function, and a team list, a complete state transition model containing cost functions and constraint rules is obtained by using a single-objective greedy algorithm to obtain normalized reference values, defining a seven-tuple state space, and constructing a set of state transition feasibility checks. Specifically, this includes: running a greedy algorithm with the single objectives of minimizing total effective travel time, minimizing weighted repair waiting time, minimizing team workload imbalance, and maximizing cumulative collaborative maintenance reward value to obtain four normalized reference values; based on preset four-dimensional weight coefficients and the four obtained normalized reference values, a multi-objective weighted total cost function is constructed using a linear weighted sum formula to obtain the single-step cost calculation rules for state transition; the four-dimensional weight coefficients include efficiency weight, reliability weight, balance weight, and collaborative reward weight; based on the enhanced task list, the state space is defined in the form of seven-tuples, and four types of sequential check rules are constructed for temperature and time window feasibility, communication closed-loop feasibility, personnel role feasibility, and collaborative maintenance time feasibility, resulting in a set of state transition feasibility checks, thus obtaining a complete state transition model. Based on the enhanced task list, cluster analysis was performed by introducing a weighted anisotropy measure with a penalty term for material thermal inertia difference to obtain several task clusters and task cluster division results; Based on the task cluster partitioning results, composite travel time matrix and complete state transition model, cluster access order is allocated through cluster-level dynamic programming. Then, precise dynamic programming path search is performed on the tasks within each cluster, taking into account the equivalent service time of temperature rise perception and collaborative maintenance rewards, to obtain the optimal access order and estimated arrival time of tasks within the cluster. Based on the optimal access order and estimated arrival time of tasks within the cluster, a set of non-dominated path schemes containing task execution order, estimated arrival time, and collaborative maintenance suggestions is obtained through screening using a non-dominated sorting method.

[0034] The complete state transition model is a mathematical model that can be used for dynamic programming recursive solutions by encapsulating the multi-objective weighted total cost function and the state transition feasibility check rule set. Its core is to calculate the transition cost between adjacent states using the single-step cost calculation rule and to determine the legality of the state transition using the state transition feasibility check rule set. The multi-objective weighted total cost function is a comprehensive cost evaluation function formed by unifying the dimensions and weights of four optimization objectives—total effective travel time, weighted repair waiting time, team workload imbalance, and cumulative collaborative maintenance reward value—using a linear weighted sum formula. Its form is as follows: ,in Total effective travel time Weighted repair waiting time, To address the uneven workload among work teams. This is the cumulative collaborative maintenance reward value; These are the normalized reference values ​​obtained through a single-objective greedy construction algorithm; The four-dimensional weighting coefficients set for the dispatcher, and satisfying the following conditions: The aforementioned uneven workload among work teams Defined as the standard deviation of the total working hours of each shift, the calculation formula is: ;in The total number of work groups For the work group The total man-hours include travel time, equivalent service time for temperature rise sensing, and additional time spent on collaborative maintenance. This is the arithmetic mean of the total working hours for all work groups. When the difference in the number of personnel between work groups exceeds... hour, Replace with the average total working hours per person in the work group. The single-step cost calculation rule for state transition is based on the multi-objective weighted total cost function, which calculates the cost from the current state at each state transition. Transition to the next state The rules governing the costs incurred are in the form of ,in The additional effective travel time for this state transition The added weighted repair wait time for this state transition This represents the change in workload imbalance caused by this state transition. This is the reward value obtained if collaborative maintenance is selected during this state transition.

[0035] The seven-tuple state space is a discrete set of states defined by seven dimensions: the visited device location map, the current device number, the current time, the current shift number, the shift personnel role occupancy state matrix, the repeater working state Boolean value, and the set of executed collaborative maintenance loops. It can be represented as ,in This is a map showing the locations of the visited devices. The device number is the current device number. For the current moment, Assign the current work group number. The state matrix for the roles of team members. This is a Boolean value representing the repeater's operating status. This refers to the set of collaborative maintenance loops that have been executed; the state matrix of the roles occupied by the team members. It is Boolean matrix, For the work group The total number of personnel. The three columns of the matrix correspond to the following three predefined roles: R1 (Major), R2 (Safety Supervisor / Assistant), and R3 (Data Recorder and Synchronizer). A value of 0 indicates that the role is currently available, and 1 indicates that the role is currently occupied. Each subtask in the task closed-loop atomic subtask sequence is marked with a list of required roles; when performing the personnel role feasibility check, it is necessary to check... The system checks whether there are any idle personnel (i.e., element = 0) in the corresponding column of the corresponding work group. If there are idle personnel for all required roles, the system is considered feasible. The state transition feasibility check rule set is a set of sequential judgment rules that includes feasibility checks for temperature and time windows, communication closed-loop feasibility checks, personnel role feasibility checks, and collaborative maintenance time feasibility checks. These rules are used to filter legal state transition paths during the dynamic programming recursion process.

[0036] Temperature communication sensing weighted clustering is a method based on a composite travel time matrix. It introduces a weighted dissimilarity measure into the traditional K-means clustering algorithm, using the product of the absolute value of the difference in the thermal inertia coefficients of two equipment materials and the average daily solar radiation temperature rise during the planned maintenance period as a penalty term. This method then performs cluster analysis on maintenance tasks in the enhanced task list. The weighted dissimilarity measure is a corrected distance formula used to calculate the degree of difference between any two maintenance task points, and its form is: ,in For the task point With mission points The weighted dissimilarity measure between them, where Task points in the composite travel time matrix With mission points The combined travel time between them and Task points and mission points The thermal inertia coefficient of the material of the associated equipment, The average solar radiation temperature rise during the planned maintenance period. The difference penalty coefficient is used; cluster-level dynamic programming is a task cluster allocation algorithm that uses the assigned cluster bitmap and the current shift number as the state and the composite travel time between cluster centers as the transition cost for recursive solution; intra-cluster dynamic programming is an exact path planning algorithm that uses the visited equipment point map, the current equipment number, the current time, the current shift number, the shift personnel role occupation state matrix, the repeater working state Boolean value, and the set of executed collaborative maintenance loops as the seven-tuple state space within the task clusters assigned to each shift. In the state transition, the temperature rise sensing equivalent service duration function, the maintenance task execution process after communication constraint correction, and the collaborative maintenance reward calculation function are called in sequence, and the memoized search is performed using the multi-objective weighted total cost function as the basis for state transition cost; Pareto front screening is an operation that removes dominated solutions and retains non-dominated solutions from the solution results corresponding to multiple sets of weight coefficient combinations through non-dominated sorting method; the set of non-dominated path schemes is a set of several non-dominated maintenance path schemes obtained after Pareto front screening, where each scheme includes the task execution order, the estimated arrival time, and the collaborative maintenance suggestion.

[0037] Furthermore, the setting methods for each preset coefficient and threshold are explained below: Difference Penalty Coefficient The value is determined by adjusting the clustering results and is usually set to [value]. Its function is to balance the relative importance of spatial distance and material thermal inertia differences in clustering; the number of clusters Based on the total number of tasks Number of available work groups according to The calculations ensure that the number of tasks within each task cluster does not exceed 15 to facilitate accurate dynamic programming solutions later; four-dimensional weight coefficients. Based on the daily maintenance conditions, in the efficiency-first mode... Take the larger value, power supply priority mode Take the larger value; in balanced mode, each weight is allocated according to a preset ratio; normalized reference value. The algorithm is obtained by running a greedy construction algorithm for each corresponding single objective. The greedy construction algorithm adopts the strategy of selecting the next task point that minimizes the cost increment of the current single objective until all task points have been visited. The non-dominated sorting method adopts a fast non-dominated sorting algorithm, which determines the Pareto dominance relationship based on the values ​​of each scheme on the four optimization objectives and retains all schemes that are not dominated by any other schemes.

[0038] Specifically, such as Figure 2 As shown, the construction steps of the complete state transition model are first performed, and the number of tasks in the enhanced task list is denoted as... The number of available work groups in the work group list is denoted as... Based on the total effective travel time. Minimum, weighted repair waiting time Minimum, uneven workload among work teams Minimum and cumulative collaborative maintenance reward value The maximum is a single optimization objective, and a greedy construction algorithm is used to run each task once. The execution process of the greedy construction algorithm is as follows: starting from the team's location, maintain the set of visited tasks and the current time. In each run, select the task that minimizes the current single-objective cost increment from the unvisited tasks as the next visit point. If multiple tasks simultaneously meet the condition, select the task with the smaller task number, until all tasks have been visited. Record the objective function value obtained in each run, and use it as the first normalization reference value. Second normalized reference value Third normalized reference value and the fourth normalized reference value Receive preset four-dimensional weighting coefficients. The four-dimensional weighting coefficients satisfy Furthermore, all coefficients are not less than 0. Based on the linear weighted sum formula, a multi-objective weighted total cost function is constructed. In the state transition process of dynamic programming, from state... Transition to state Calculate the cost per step. ,in The additional travel time added for this transfer The weighted waiting time added for this transfer This represents the change in workload imbalance within the work teams caused by this transfer. Let the reward value obtained by choosing collaborative maintenance in this transition be the basis for the single-step cost calculation rule for state transition. Define a 7-tuple state space, where states... ,in This is a map of visited device points, with a length of... The binary bit string representation of the first bit is... A bit value of 1 indicates the first One task has been accessed; The current device number, with a value of to , Indicates the work team's location; The current time is expressed in minutes. This is the current work group number, with a value of [value]. to ; The state matrix for the roles of team members is structured as follows: A boolean flag indicating whether each role in the personnel role attribute matrix is ​​currently occupied; This is a Boolean value representing the repeater's operating status. This indicates that the repeater is turned on and is in working condition. This indicates that the repeater is not turned on or is not being carried. For the set of completed collaborative maintenance loops, record the loop numbers that have been maintained in collaborative maintenance actions to avoid repeatedly awarding collaborative maintenance rewards to the same loop. Construct a set of state transition feasibility check rules, including: temperature and time window feasibility checks, i.e., calculating the arrival point of the task. The moment The equivalent service duration function for temperature rise sensing is called to obtain... ,like Greater than Then it is determined to be infeasible, where For the task point The latest arrival time within the scheduled time window; a communication loop feasibility check, i.e., if the task point... For communication blind spots and Then check Does this cause subsequent tasks to time out? For the task point The walking time from the communication reachable point back to the parking space is taken as the value of the task point. The walking distance between the corresponding reachable point coordinates and the parking space coordinates divided by the walking speed The resulting quotient; personnel role feasibility check, i.e., traversing task points. The role requirements of each subtask in the corresponding closed-loop atomic subtask sequence are checked. Are there any available personnel? A feasibility check for collaborative maintenance time consumption should be performed, i.e., if collaborative maintenance is chosen, the additional time consumption should be checked. Does it cause subsequent tasks to time out? The multi-objective weighted total cost function, the single-step cost calculation rules for state transitions, and the set of rules for checking the feasibility of state transitions are encapsulated to obtain a complete state transition model.

[0039] Then, the temperature communication-sensing weighted clustering step is performed. The material thermal inertia coefficient of each task-associated device is extracted from the enhanced task list. The ambient temperature and solar radiation intensity values ​​for each hour within the planned maintenance period are extracted from meteorological time-series data. The average value of the baseline value of solar radiation temperature rise during this period is calculated and denoted as the average solar radiation temperature rise. For any two task points in the enhanced task list. and Obtain the composite travel time from the composite travel time matrix. Calculate the absolute value of the difference in thermal inertia coefficients of the materials. Substitute into the weighted dissimilarity measurement formula The difference penalty coefficient Pick The aforementioned weighted dissimilarity metric replaces the Euclidean distance metric in the traditional K-means clustering algorithm, performing cluster analysis on all task points in the enhanced task list (excluding communication blind zone tasks that have already been prioritized for separate clustering). The K-means algorithm uses the K-means++ method for initial cluster center selection, and the iteration termination condition is that the maximum geographical offset of all cluster centers between two iterations is less than 1 meter. Number of clusters. according to Confirmed, among which The total number of tasks. Number of available work groups This indicates rounding up. During clustering, tasks categorized as communication blind spots are preferentially clustered separately. Specifically, during initialization, each communication blind spot task is first selected as an independent candidate for initial cluster center, and then K-means initialization is performed on the remaining reachable tasks. After clustering iterations converge, the following is obtained: Task clusters The corresponding task cluster partitioning results, with each task cluster containing no more than [number of tasks]. indivual.

[0040] The next step involves nested solutions of cluster-level and intra-cluster dynamic programming. First, cluster-level dynamic programming is performed: the state is defined as... ,in For the allocated cluster bitmap, with a length of The binary bit string representation of the first bit is... Position Indicates the first One task cluster has been assigned; This is the current work group number. The initial state is... , This represents an all-zero bitmap. During state transition, an unassigned task cluster is selected. (Right now The Position ), and assign it to the current work group Or switch to another available shift. The transfer cost is calculated from the current shift's last location (or shift base if the shift has not yet been assigned any task clusters) to the task cluster. Composite travel time between cluster centers The cluster center is defined as the task cluster. The geometric center of the latitude and longitude coordinates of all task points within the area is used to obtain the access order of the task clusters assigned to each work group by recursively solving the optimal cluster allocation scheme. For each work group... The assigned task cluster sequence is denoted as ,in To be assigned to work groups The There are several task clusters, and then precise dynamic programming is performed within each cluster: for each work group A specific cluster of tasks assigned Let the number of task points it contains be... The initial state is the last state of the work group before executing the task cluster. The state space is defined using a seven-tuple state space. In each state transition That is, from the task point Transfer to mission point In the process, the following steps are performed in order: calculate arrival time. ,in The round-trip walking time to the reachable point is already included; the task point is calculated by calling the equivalent service duration function based on temperature rise sensing. exist Surface temperature estimate and waiting time for the temperature to drop and receive equivalent service time. ,in Based on the standard maintenance time, Fixed safety operation preparation time; classification and labeling based on communication scenario and repeater operating status Boolean values. Determine the task closed-loop sequence; if it is a task in a communication blind zone and The total time for the task loop to close is then... Otherwise ,in The time taken for in-situ data synchronization, i.e., when the task point... For tasks that are reachable through communication or tasks that are in communication dead zones but are covered by portable repeaters carried by the work team, the time required for maintenance personnel to complete data synchronization on-site after physical maintenance is calculated based on the long-distance narrowband radio communication rate and typical data transmission volume, and is usually taken as [value missing]. Correspondingly, This indicates the time taken for data synchronization at communication points, typically taken as a value. If the task point The reward value returned by the corresponding collaborative maintenance reward calculation function If the value is greater than 0, then the evaluation will determine whether to select to execute the collaborative maintenance action. If selected, additional time will be consumed. The cost is calculated based on the sum of the maintenance reference time for the additional circuits, and a negative reward is obtained in the state transition cost; the comprehensive cost of this transition is calculated according to the single-step cost calculation rules for state transition. ; Update the time for the next state Update the map of visited device locations. Update the set of collaborative maintenance loops that have been executed. ,in The total time for task closure is calculated, and if collaborative maintenance is performed, a corresponding loop number is added. A memoized search method is used to traverse all legal state transition paths, recording the minimum cumulative cost to reach each state and its corresponding predecessor state. Finally, the optimal access order and estimated arrival time of tasks within the cluster are obtained through backtracking. To improve search efficiency, this embodiment employs a pruning strategy during the memoized search process. This pruning strategy includes at least cost pruning or time window pruning. Cost pruning involves terminating the search from that path if the cumulative cost to reach a certain state is greater than or equal to the recorded minimum cumulative cost to reach the same state. Time window pruning involves terminating the search from that state if the estimated arrival time from the current state to any remaining unvisited task point exceeds the latest arrival time of its scheduled time window.

[0041] Finally, the Pareto front screening step is performed. Multiple sets of four-dimensional weight coefficient combinations are preset. For each set of weight coefficient combinations, a complete solution process involving temperature communication sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming is executed once. The four-dimensional angular function values ​​corresponding to the optimal solution obtained under that set of weights are recorded. The solution results are used to form a candidate solution set, and the Pareto dominance relationship is determined using the fast non-dominated sorting algorithm: for any two candidate solutions... and ,like Non-inferior in all four objectives And strictly superior to at least one objective Then it is called Dominate The Pareto front is formed by removing all solutions dominated by other candidate solutions from the candidate solution set and retaining all solutions that are not mutually dominant. Each solution in the Pareto front corresponds to a complete maintenance path planning scheme, which includes the task execution order assigned to each shift, the estimated arrival time of each task, whether to recommend performing coordinated maintenance actions, and the specific loop number for coordinated maintenance. The dispatcher can select a scheme from the Pareto front and issue it for execution based on actual operation and maintenance needs.

[0042] For example, this embodiment assumes that the enhanced task list for a certain day includes There are 1 maintenance task pending, and the number of available work teams is [number missing]. First, calculate the number of clusters. That is, weighted clustering of temperature communication sensing is performed on 28 task points using a weighted dissimilarity metric, resulting in 3 task clusters, among which cluster 1 is a single cluster. Contains 10 task points, cluster Contains 9 task points, cluster It contains 9 task clusters. The cluster center coordinates of each task cluster are as follows: Assuming a work group ,team ,team The coordinates of the station are respectively Perform cluster-level dynamic programming, state Recursion: Starting from the initial state, the clusters Assigned to work groups The cost is from to The composite journey time; cluster Assigned to work groups The cost is from to The composite journey time; cluster Assigned to work groups The cost is from to The composite travel time is used to obtain the cluster allocation scheme for the work group. responsible cluster ,team responsible cluster ,team responsible cluster Perform precise dynamic programming within each cluster separately, based on work groups. responsible cluster For example, it contains 10 task points, numbered as follows: to From the work group Starting from the base, the optimal access order is obtained through state space search. In the mission During the maintenance process, the reward value returned by the collaborative maintenance reward calculation function is used. It is recommended to perform collaborative maintenance, and additionally inspect the health score of the same component within the enclosure if it is lower than the recommended level. Adjacent loops increase time consumption. .go through The group weight coefficients were obtained by solving the problem. The non-dominated solutions constitute the Pareto front, among which the schemes The corresponding weight is Total effective travel time Weighted repair waiting time Uneven workload among work teams Cumulative collaborative maintenance reward value ;plan The corresponding weight is Total effective travel time Weighted repair waiting time Uneven workload among work teams Cumulative collaborative maintenance reward value ;plan The corresponding weight is Total effective travel time Weighted repair waiting time Uneven workload among work teams Cumulative collaborative maintenance reward value If the dispatcher's primary objective for the day is to ensure power supply, then they can choose the appropriate option. If the primary goal is to improve long-term economic benefits, then the following option can be selected. In this embodiment, the values ​​of the difference penalty coefficient, the number of clusters, and the number of weight coefficient combinations are merely examples. Those skilled in the art can set them according to actual conditions, and this embodiment does not impose any restrictions on them.

[0043] This embodiment embeds the difference in material thermal inertia into the cluster dissimilarity metric as a penalty term. It significantly compresses the state space size by using a double-layer nested solution of cluster-level dynamic programming and intra-cluster exact dynamic programming. It achieves joint optimization of triple constraints by sequentially calling the temperature rise sensing equivalent service duration function, the maintenance task execution process after communication constraint correction, and the collaborative maintenance reward calculation function during the state transition of intra-cluster dynamic programming. By traversing multiple sets of weight coefficients to generate Pareto fronts for dispatchers to choose flexibly, it realizes efficient and accurate solution of distribution network junction box maintenance paths under a multi-objective optimization framework. This improves the computational efficiency of path planning in large-scale task scenarios, ensures the time window satisfaction rate of maintenance plans under actual high-temperature conditions and communication blind zone constraints, and provides dispatchers with a quantitative decision-making basis for balancing operational efficiency, power supply reliability, and long-term economic benefits.

[0044] Furthermore, this embodiment provides a step-by-step approach to obtain an updated complete maintenance path plan based on a set of non-dominated path solutions combined with real-time status snapshots and a list of newly added tasks, through reverse path re-inspection value assessment and remaining task replanning, including: Based on the set of non-dominated path solutions, a real-time state snapshot and a set of high-value re-inspection point markers are obtained by state variable aggregation and breadth-first search tracing. Based on real-time status snapshots, newly added task lists, and high-value re-inspection point marker sets, a replanning task set is obtained by merging the sets, and an updated complete maintenance path plan is obtained based on the replanning task set.

[0045] The real-time status snapshot is a real-time status record obtained by aggregating the GPS coordinates, task completion confirmation receipts, and current time data transmitted via the team's mobile terminals during the execution of the plan. This record includes the current location, current time, set of completed tasks, the current occupancy status of team members' roles, and the current working status of the repeater. It provides initial state input for dynamic replanning. The high-value re-inspection point mark set is a set of indicators for each faulty device in the new task list. It involves tracing upstream repaired devices through an electrical topology directed graph, calculating the electrical topology distance between the repaired upstream devices and the newly faulty device, and the additional time cost of returning for re-inspection. The reverse path re-inspection value is obtained based on the ratio of the faulty device's importance weight to the electrical topology distance and additional time cost. The set of indicators for upstream repaired devices whose reverse path re-inspection value is greater than a preset threshold is used to capture opportunities for forward re-inspection of the repaired upstream devices. The reverse path re-inspection value is a quantitative indicator measuring the marginal benefit of returning from the current location to the repaired upstream devices for re-inspection. Its calculation formula is... ,in For the upstream equipment that has been repaired Compared to newly added faulty equipment The value of reverse path re-examination For newly added faulty equipment Importance weight, To access the upstream equipment that has already been repaired To newly added faulty equipment Electrical topology distance, The additional time cost of returning for re-inspection The time value conversion factor is preset; the electrical topology distance is the distance from the repaired upstream equipment in the directed electrical topology graph. The node departs to reach the newly faulty device The minimum number of directed edges traversed by a node is calculated using a breadth-first search algorithm and is used to quantify the tightness of electrical coupling between two devices; the additional time cost of returning for re-inspection is the time spent returning to the upstream device that has already been repaired from the current position. The parking space coordinates are then used to locate the newly added faulty equipment. The combined travel time required to find the parking space coordinates and to travel directly from the current location to the newly added faulty equipment. The difference in the composite travel time required to locate the parking space coordinates is calculated using the following formula: ,in From the current location to the upstream equipment that has been repaired The combined travel time for parking spaces To access the upstream equipment that has already been repaired From parking space to newly faulty equipment The combined travel time for parking spaces From the current location to the newly added faulty equipment The composite travel time of parking spaces; the replanning task set is the set of tasks obtained by combining the unexecuted tasks in the real-time status snapshot, all new tasks in the new task list, and the upstream equipment marked in the high-value re-inspection point mark set through set union operation, and is used as the input task set for dynamic replanning solution.

[0046] The setting methods for the above preset coefficients and thresholds are explained below: Time Value Conversion Coefficient It is determined based on the economic equivalence relationship between unit travel time cost and unit re-inspection revenue, and is usually taken as a value. Its function is to convert the extra time cost of returning for re-inspection into a value with the same dimension as the importance weight; and to preset the value threshold for reverse path re-inspection. This value is calculated based on the minimum re-inspection value of the route-based re-inspection schemes actually adopted in historical scheduling decisions, and is typically taken as [value missing]. When the value of reverse path re-examination Value greater than the preset reverse path re-examination value threshold The upstream equipment that has been repaired will be ready at that time. Included in the set of high-value re-inspection point markers.

[0047] Specifically, such as Figure 3 As shown, the process begins with obtaining real-time status snapshots and a set of high-value re-inspection point markers. The dispatcher selects an execution plan from the set of non-dominated path plans and pushes the task execution order, estimated arrival time, and collaborative maintenance suggestions in the plan to the mobile terminals of the corresponding work teams, thus obtaining the execution plan. During the execution of the plan, the work team's mobile terminals upload GPS coordinates at preset time intervals and upload a task completion confirmation receipt upon completion of each task. The preset time interval is dynamically adjusted according to the work team's current movement speed. This interval parameter can be modified in the system configuration file. For example, the time interval is set to 30 seconds when the work team's current movement speed is greater than 30 km / h, and 2 minutes when the work team's current movement speed is less than or equal to 30 km / h. After receiving the above-mentioned data, the system extracts the current location using a state variable aggregation method. Current moment Completed task collection (Includes all completed task numbers), current occupancy status of team member roles. (Indicates the current occupancy status of each member and role in each shift) and the current working status of the repeater. (Values) or ), assembled into a real-time state snapshot. When a new task list is added. Upon arrival, for each newly added faulty device in the new task list Let its importance weight be . Importance weight of faulty equipment The weighting is determined by the highest load level of the faulty device, consistent with the rules for assigning importance weights to circuit loads; specifically, level one, level two, and level three loads correspond to weights of 0.9, 0.6, and 0.3, respectively. This is determined by adding the faulty device to the electrical topology directed graph. As the endpoint, with all upstream equipment that has been repaired. Using candidate starting points, the electrical topology distance is calculated using a breadth-first search algorithm. If from Unable to reach the destination after departure but For each upstream device that has been repaired and has an existing path. Calculate the additional time cost of returning for re-inspection. ,in To start from the current position Upstream equipment that has been repaired The combined travel time for parking spaces To access the upstream equipment that has already been repaired From parking space to newly faulty equipment The combined travel time for parking spaces To start from the current position To newly added faulty equipment The combined travel time for parking spaces is obtained from the combined travel time matrix. Substitute into the formula for calculating the value of reverse path re-examination The time value of money discount factor Pick If the calculated value of the reverse path re-examination Value greater than the preset reverse path re-examination value threshold Then the upstream equipment that has been repaired will be... Mark it as a high-value re-inspection point and add it to the high-value re-inspection point marker set. When multiple high-value re-inspection points exist, the re-inspection value is performed according to the described reverse path. Sort by high to low, prioritizing the points with the highest re-inspection value; otherwise, do not mark them.

[0048] Then, the remaining tasks are replanned and path concatenation is performed. The original tasks that have not yet been executed are extracted from the real-time state snapshot to form a set of unexecuted tasks. Set of unexecuted tasks New task list and a set of high-value re-inspection point markers The three are combined using a set union operation to obtain the replanning task set. The current position in a real-time snapshot. As a virtual starting point, the current moment As of the starting moment, the current occupied state of the team members' roles. As the initial occupancy state, the current operating state of the repeater. As the initial repeater state, to replan the set of tasks. As the complete set of tasks to be visited, the solution is recalculated using the complete solution process of temperature communication sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming to obtain the updated remaining path scheme. This updated remaining path scheme is then compared with the set of completed tasks. The corresponding traveled route records are spliced ​​together, meaning the traveled route records are preserved and the process starts from the current position. The remaining route options are appended to obtain the updated complete maintenance route option. If, during the replanning process, it is found that the estimated arrival time of a task exceeds the latest arrival time of its scheduled time window, a timeout alarm message is sent to the scheduler, indicating the timeout task number and the estimated timeout duration.

[0049] This embodiment aggregates data transmitted from the team's mobile terminals in real time during the execution of the solution to form a real-time status snapshot. It uses breadth-first search to trace the upstream repaired equipment of the newly added faulty equipment and calculates the re-inspection value of the reverse path to screen high-value re-inspection points. By merging unexecuted tasks, newly added tasks, and high-value re-inspection points, it calls the hybrid solution process for rapid replanning. By splicing the replanning results with the beginning and end of the already traveled path, it obtains the updated complete maintenance path solution. This realizes online dynamic adjustment for sudden task insertion or execution deviation from the working conditions, improves the response speed of the maintenance scheduling system to real-time changes in working conditions, and reduces the extra mileage and dispatch costs caused by dispatching vehicles separately for newly added tasks or ignoring along-the-way re-inspection opportunities. This further improves the overall efficiency and economy of junction box maintenance.

[0050] Furthermore, this embodiment provides a step-by-step approach based on the updated complete maintenance path scheme, which involves obtaining an updated model parameter configuration file and a system performance evaluation report through statistical fitting and rolling parameter updates. The steps include: Based on the updated complete maintenance path scheme, an execution record dataset is obtained through data cleaning and structured storage; Based on the execution record dataset, the updated model parameter configuration file was obtained through median statistics, linear regression fitting, and comparative analysis. Based on the execution record dataset, a system performance evaluation report is obtained by calculating the time window satisfaction rate, the accuracy of travel time prediction, and the actual benefits of collaborative maintenance.

[0051] The execution record dataset is a structured data set formed after the updated complete maintenance path plan has been executed. This dataset consists of the entire trajectory data of the team's mobile terminals, the actual arrival time and maintenance time records for each task, communication synchronization time records, collaborative maintenance execution records, and timeout alarm information. The data is cleaned and structured for storage, providing factual basis for model parameter updates and system performance evaluation. The updated model parameter configuration file is a set of parameters updated to their current optimal values ​​and stored in configuration file format. This is obtained by statistically analyzing the execution record dataset using median statistics, linear regression fitting, and comparative analysis. At least one of the following parameters—basic walking distance parameters, floor conversion coefficient, material thermal inertia coefficient, and coupling degradation coefficient—is updated to its current optimal value. This configuration file is used for subsequent modeling of temperature sensing equivalent service duration, construction of communication blind zone task loops, and calculation of multi-loop collaborative maintenance rewards. The basic walking distance parameter is a statistical parameter characterizing the average walking distance from the parking space to the building where the equipment is located in a Class B community. Its initial value is obtained based on statistical analysis of the average distance from the gate of a typical community to the farthest building, and is typically set to a value of [value missing]. The floor conversion factor is a parameter that represents the equivalent increase in walking distance for maintenance personnel carrying tool kits up and down stairs for each additional floor. Its initial value is determined based on the ratio of walking speed to stair-climbing speed, and is typically set to a value of [value missing]. The thermal inertia coefficient of materials is a physical parameter characterizing the difference in surface temperature rise rate of enclosures made of different materials under the same solar radiation conditions. It includes the thermal inertia coefficients for stainless steel, sheet molding compound (FRP), and polycarbonate. Its initial value is obtained by proportionally normalizing the measured thermal conductivity data of each material. The coupling degradation coefficient is a parameter characterizing the degree of thermal coupling degradation of adjacent loops within the same enclosure. It includes the basic coupling coefficient, distance attenuation factor, and common busbar additional coupling coefficient. Its initial value is obtained by statistical fitting of historical fault data. The system performance evaluation report is a summary report generated after calculating three evaluation indicators: time window fulfillment rate, travel time prediction accuracy, and actual benefits of collaborative maintenance. The time window fulfillment rate is defined as the actual arrival time of all executed tasks not being later than the latest arrival time of the scheduled time window. The ratio of the number of tasks to the total number of tasks is used to evaluate the degree to which the path planning scheme adheres to the time window constraints. The mean absolute error of travel time prediction is the average of the absolute values ​​of the differences between the actual arrival time of each task and the estimated arrival time in the execution plan. The smaller the value, the smaller the overall prediction deviation. The travel time prediction accuracy is the ratio of the number of tasks whose travel time prediction absolute error is less than or equal to the preset allowable deviation threshold to the total number of tasks. The larger the value, the higher the prediction accuracy. It is used to evaluate the accuracy of the composite travel time matrix and dynamic programming model in predicting travel time. The preset allowable deviation threshold is based on a default value of 15 minutes and is set differently according to task priority and power supply level. It is adaptively and dynamically calibrated by the closed-loop optimization module based on the accuracy distribution of historical execution records, the proportion of overdue tasks, and the satisfaction feedback from maintenance personnel. The actual benefit of collaborative maintenance is the sum of the subsequent independent dispatch costs avoided by the collaborative maintenance actions actually performed in the execution plan. It is used to evaluate the actual economic effect of the multi-loop collaborative maintenance reward mechanism.

[0052] The calculation methods for the above evaluation indicators are explained below: Time window satisfaction rate The calculation formula is: ,in The number of tasks whose actual arrival time is no later than the latest arrival time within the scheduled time window. The total number of all tasks performed; actual benefits of collaborative maintenance. The calculation formula is: ,in This represents the total number of collaborative maintenance actions actually performed. For the first The importance weight of the circuit loads of adjacent circuits involved in the second coordinated maintenance. For the first The health score of adjacent circuits involved in the second collaborative maintenance before the maintenance. The preset loop health score threshold is typically set to 75.

[0053] Specifically, after the updated and complete maintenance route plan is executed, the system first exports the entire GPS trajectory data from the team's mobile terminals, and then extracts the actual arrival time of each task from the task completion confirmation receipt. The actual start and end times of maintenance are used to calculate the actual equivalent service duration. , For the first The actual start time of the maintenance for each task. For the first The actual end time of each maintenance task; extract the actual communication synchronization time of each communication blind spot task from the communication records. This includes the time spent on in-situ data synchronization or the time spent on communication point data synchronization; extracting a list of actually executed collaborative maintenance actions from the collaborative maintenance confirmation record, including the circuit number being collaboratively maintained and the additional time spent on collaborative maintenance. and the circuit health score of the circuit before collaborative maintenance. The original data was cleaned to remove abnormal records caused by terminal signal interruption or human error. The cleaned data was then organized into a structured data table indexed by task number. Each record contained task number, associated device number, actual arrival time, actual equivalent service duration, actual communication synchronization time, whether collaborative maintenance was performed, and related fields, forming an execution record dataset.

[0054] Then, all task records belonging to the walking scenario within the Class B community were filtered out from the execution record dataset, and the actual walking distance from the parking space coordinates to the device coordinates, calculated from the GPS trajectory, was extracted from each task record. Calculate all actual walking distances The median was used as the updated baseline walking distance parameter. From the execution record dataset, filter out all Class B scenario task records containing floor information, and extract the actual walking time for each task record. With floor number By floor number The independent variable is the actual walking time. As the dependent variable, a univariate linear regression is performed using the least squares method, and the fitting formula is: The regression coefficients Divide by walking speed Get the updated floor conversion factor ;in For the first The actual walking time for the walking scenario task within the Class B community. The intercept term in the linear regression represents the baseline walking time when the building is on the first floor, i.e., the walking time from the parking space to the entrance of the building. The regression coefficient, i.e., the slope term of the linear regression, represents the additional walking time for each additional floor. For the first The floor number of the device's location corresponding to each task is a positive integer. All metal enclosure task records and sheet molding compound enclosure task records are filtered from the execution record dataset, and the arrival time and actual equivalent service duration of each task record are extracted. Standard maintenance time Fixed safe operation preparation time And the corresponding ambient temperature and solar radiation intensity values, to calculate the actual waiting time for cooling. Using the actual cooling time as the objective variable and the material thermal inertia coefficient in the temperature rise sensing equivalent service duration function as the parameter to be calibrated, a nonlinear least squares fitting method is used to update the material thermal inertia coefficients for stainless steel and sheet molding compound fiberglass, respectively. The objective function is to minimize the sum of squared residuals of all training samples. ;in To calculate the theoretical waiting time for cooling based on the temperature rise sensing equivalent service duration function, the iteration termination condition for fitting is that the change in the objective function value is less than [a certain value]. ; Filter out all task records that performed collaborative maintenance actions from the execution record dataset, and extract the circuit health score of each circuit before maintenance. And a marker indicating whether the circuit will fail within a subsequent preset observation period (e.g., 3 months). The relationship between the probability of failure and the loop health score and coupling degradation coefficient is fitted using a logistic regression model, and the basic coupling coefficient is updated using the maximum likelihood estimation method. Distance attenuation factor and the additional coupling coefficient of the common busbar The input feature of the logistic regression model is the loop health score. The coupling degradation coefficient is used to output a label indicating whether a fault has occurred in the loop within a preset observation period. Write the updated parameter values ​​to the model parameter configuration file, overwriting the original parameter values ​​to obtain the updated model parameter configuration file.

[0055] The final steps for generating the system performance evaluation report include: extracting the total number of all executed tasks from the execution record dataset. The number of tasks whose actual arrival time is no later than the latest arrival time within the scheduled time window is counted. According to the formula Calculate the time window satisfaction rate; extract the actual arrival time of each task from the execution record dataset. And extract the corresponding estimated arrival time from the issued execution plan. According to the formula Calculate the first The absolute error of the travel time prediction for each task, and simultaneously according to the formula Calculate the mean absolute error of travel time prediction This metric is used to quantify the overall average deviation of the predicted arrival times for all tasks, where For the first The actual arrival time of each task For the first The estimated arrival time of each task in the execution plan; the number of tasks whose absolute error in travel time prediction is less than or equal to the preset allowable deviation threshold for that task. According to the formula The accuracy of the travel time prediction was calculated. This metric is used to quantify the percentage of tasks where the prediction error is within the acceptable range for business operations; it also extracts the total number of collaborative maintenance actions actually performed from the execution record dataset. and the importance weight of the circuit loads of adjacent circuits involved in each collaborative maintenance. Compared with the circuit health score before maintenance According to the formula Calculate the actual benefits of collaborative maintenance; summarize the above three evaluation indicators along with the statistical summary information of the execution record dataset to generate a system performance evaluation report.

[0056] This embodiment systematically collects and cleans actual operational data after the scheme is completed to form an execution record dataset. It updates the basic walking distance parameters through median statistics, updates the floor conversion coefficient through linear regression fitting, updates the material thermal inertia coefficient through nonlinear least squares fitting, and updates the coupling degradation coefficient through logistic regression and maximum likelihood estimation. It generates a system performance evaluation report by calculating the time window satisfaction rate, travel time prediction accuracy, and actual benefits of collaborative maintenance. This achieves closed-loop adaptive optimization of model parameters as they continuously approach the real physical laws with actual operational data, improving the temperature sensing accuracy, communication constraint modeling accuracy, and economic rationality of collaborative maintenance decisions in subsequent maintenance path planning. This ensures the continuous effectiveness of this method and the ability to control operation and maintenance costs during long-term operation.

[0057] Example 2: like Figure 4 As shown in the figure, this application provides a dynamic planning system for junction box maintenance paths based on multi-objective optimization. The system includes a data augmentation module, a multi-objective planning module, an online replanning module, and a closed-loop optimization module. Data Enhancement Module: Used to acquire basic data sets, and obtain an enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function through temperature sensing equivalent service duration modeling, communication blind zone task closed-loop construction and multi-loop collaborative maintenance reward; Multi-objective programming module: Based on the enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function, it obtains a complete state transition model through multi-objective normalization and state space definition, and solves it by nesting temperature communication sensing weighted clustering, cluster-level dynamic programming and intra-cluster dynamic programming, and obtains a set of non-dominated path solutions through Pareto front screening; Online replanning module: Based on the set of non-dominated path solutions, combined with real-time status snapshots and a list of newly added tasks, it obtains an updated complete maintenance path solution through reverse path re-inspection value assessment and replanning of remaining tasks. Closed-loop optimization module: Based on the updated complete maintenance path plan, it obtains the updated model parameter configuration file and system performance evaluation report through statistical fitting and rolling parameter updates.

[0058] The data enhancement module, multi-objective programming module, online replanning module, and closed-loop optimization module are all located on the server. The server receives data transmitted from the acquisition devices and performs further analysis. The acquisition devices include: smart meters and temperature sensors installed in the junction boxes to collect load data, temperature rise data, and box surface temperature data for each outgoing circuit; a GPS positioning terminal installed on the maintenance vehicle to transmit the vehicle's location coordinates in real time; a mobile handheld terminal carried by maintenance personnel to upload task completion confirmation receipts and actual arrival times; and long-distance radio communication gateways and narrowband IoT base stations deployed in the distribution network to provide regional communication signal coverage data. The data enhancement module receives a set of basic data from multiple heterogeneous systems, including data acquisition equipment and production management systems, power distribution geographic information systems, fault reporting systems, and meteorological service interfaces. After modeling the equivalent service duration of temperature sensing, constructing closed-loop tasks for communication blind spots, and calculating rewards for multi-loop collaborative maintenance, it generates an enhanced task list, a composite travel time matrix, and a collaborative maintenance reward calculation function. This data is then output to the multi-objective programming module. The multi-objective programming module performs multi-objective normalization and complete state transition model construction based on the received data. It then generates a Pareto front solution set through nested solutions of temperature communication sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming. The dispatcher selects a solution and outputs the chosen solution to the online replanning module. During the solution execution process, the online replanning module continuously receives real-time status data and new task lists from the acquisition equipment, triggers reverse path re-inspection value assessment and remaining task replanning, and outputs an updated complete maintenance path solution. The closed-loop optimization module collects actual execution record data after the solution is completed, and performs rolling updates on basic walking distance parameters, floor conversion coefficients, material thermal inertia coefficients, and coupling degradation coefficients. The updated model parameter configuration file is then fed back to the data augmentation module, multi-objective planning module, and online replanning module, forming a closed-loop iterative mechanism for parameter adaptive optimization.

[0059] In summary, this invention transforms the thermal inertia difference of the junction box material into an equivalent service duration function of temperature rise sensing that depends on the arrival time. It determines the communication reachable point for junction boxes in communication blind spots and constructs a closed-loop atomic subtask sequence containing physical maintenance subtasks and data synchronization subtasks. It establishes the thermal coupling degradation relationship between loops and calculates the reward value of optional collaborative maintenance actions. By embedding the above three constraints into a dynamic programming state transition cost function and using temperature and communication sensing weighted clustering and nested two-layer dynamic programming to solve the problem, this invention realizes the joint dynamic programming of temperature sensing, communication sensing, and collaborative maintenance sensing for the maintenance path of distribution network junction boxes under a multi-objective optimization framework. This solution improves the time window fulfillment rate of maintenance route plans under actual high-temperature conditions, reduces the invalid mileage caused by secondary backtracking for tasks in communication blind spots, reduces the frequency of repeated dispatches due to neglecting thermal coupling degradation between circuits, and quickly generates adjustment plans to adapt to new operating conditions through reverse path re-inspection value assessment and remaining task replanning when sudden tasks are inserted. At the same time, the parameter adaptive optimization mechanism ensures the continuous effectiveness of the model in long-term operation, thereby improving the overall operational efficiency, power supply reliability and maintenance economy of junction box maintenance scheduling.

[0060] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0061] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0062] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0063] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A dynamic programming method for junction box maintenance paths based on multi-objective optimization, characterized in that, include: The basic data set is obtained, and the enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function are obtained through temperature sensing equivalent service duration modeling, communication blind zone task closed loop construction and multi-loop collaborative maintenance reward. Based on the enhanced task list, composite travel time matrix, and collaborative maintenance reward calculation function, a complete state transition model is obtained through multi-objective normalization and state space definition. The model is solved by nested solutions of temperature communication sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming. A set of non-dominated path schemes is obtained by Pareto front screening. Based on the set of non-dominated path schemes, combined with real-time status snapshots and a list of new tasks, an updated complete maintenance path scheme is obtained through reverse path re-inspection value assessment and remaining task replanning. Based on the updated complete maintenance path scheme, the updated model parameter configuration file and system performance evaluation report are obtained through statistical fitting and rolling parameter updates.

2. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 1, characterized in that, The acquisition of the basic data set, through temperature sensing equivalent service duration modeling, communication blind spot task closed-loop construction, and multi-loop collaborative maintenance reward, yields an enhanced task list, a composite travel time matrix, and a collaborative maintenance reward calculation function, including: Based on a multi-source heterogeneous system, a basic data set is obtained through data extraction, field mapping and format unification rules, including a standardized task list, equipment ledger dictionary, electrical topology directed graph, circuit adjacency list, work group list, road network configuration parameter set, meteorological time series data and regional communication signal strength heat map; Based on the equipment ledger dictionary and meteorological time series data, the equivalent service duration function of temperature rise sensing for each junction box maintenance task is obtained through material thermal inertia coefficient mapping and surface temperature linear estimation model. Based on the standardized task list, regional communication signal strength heat map, and work group list, the maintenance task execution process and composite travel time matrix after communication constraint correction are obtained through signal threshold determination, nearest communication reachable point search, and walking segment accumulation method. Based on the aforementioned loop adjacency list and standardized task list, the collaborative maintenance reward calculation function is obtained through loop health weighted scoring, thermal coupling degradation coefficient calculation, and collaborative maintenance reward formula. Based on the temperature rise sensing equivalent service duration function and the maintenance task execution process modified by communication constraints, an enhanced task list is obtained.

3. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 2, characterized in that, Based on the equipment ledger dictionary and meteorological time-series data, the equivalent service duration function for temperature rise sensing of each junction box maintenance task is obtained through material thermal inertia coefficient mapping and surface temperature linear estimation model, including: Based on the box material field in the equipment ledger dictionary, the list of equipment labeled with material thermal inertia coefficient is obtained through the material thermophysical property lookup table; Based on the latitude and longitude coordinates of the equipment and meteorological time series data, the orientation factor calculation rules for each piece of equipment are obtained through geographic information system shadow analysis and solar azimuth angle calculation. Based on the material thermal inertia coefficient labeling equipment list, orientation factor calculation rules, meteorological time series data, and preset safety operation thresholds, the temperature rise sensing equivalent service duration function is obtained through the surface temperature linear superposition formula and the simplified formula of Newton's law of cooling.

4. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 2, characterized in that, Based on the standardized task list, regional communication signal strength heat map, and work group list, the maintenance task execution process and composite travel time matrix after communication constraint correction are obtained through signal threshold determination, nearest communication reachable point search, and walking segment accumulation method, including: Based on the latitude and longitude coordinates of the devices in the standardized task list and the heat map of regional communication signal strength, a communication scenario classification label is obtained by determining the signal threshold. Based on the coordinates of devices in communication blind spots and the heat map of regional communication signal strength, the coordinates of the nearest reachable communication point for each device in a communication blind spot are obtained through signal satisfaction point search and open area priority rules. Based on the coordinates of the nearest reachable point, the maintenance task execution process and composite travel time matrix after communication constraint correction are obtained by calculating the composite walking path segment and determining the overlap of the spatial buffer.

5. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 2, characterized in that, Based on the loop adjacency list and standardized task list, the collaborative maintenance reward calculation function is obtained through loop health weighted scoring, thermal coupling degradation coefficient calculation, and collaborative maintenance reward formula, including: Based on the load data and fault history records of smart meters, the health score of each circuit is obtained by extracting the operation indicators of each circuit and weighting and summing them. Based on the physical installation distance and common busbar marking in the loop adjacency table, the coupling degradation coefficient of each faulty loop to adjacent loops in the same enclosure is calculated using the thermal coupling degradation coefficient formula. Based on the circuit health score, coupling degradation coefficient, and preset circuit load importance weight, the collaborative maintenance reward calculation function is obtained through the collaborative maintenance reward formula.

6. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 1, characterized in that, Based on the enhanced task list, composite travel time matrix, and collaborative maintenance reward calculation function, a complete state transition model is obtained through multi-objective normalization and state space definition. This model is then solved using nested methods of temperature communication-sensing weighted clustering, cluster-level dynamic programming, and intra-cluster dynamic programming. A set of non-dominated path solutions is obtained through Pareto front screening, including: Based on the enhanced task list, composite travel time matrix, collaborative maintenance reward calculation function, and work group list, a complete state transition model containing cost function and constraint rules is obtained by constructing a normalized reference value through a single-objective greedy algorithm, defining a seven-tuple state space, and building a set of state transition feasibility check rules. Based on the enhanced task list, cluster analysis was performed by introducing a weighted anisotropy measure with a material thermal inertia difference penalty term to obtain multiple task clusters and task cluster division results; Based on the task cluster partitioning results, composite travel time matrix and complete state transition model, cluster access order is allocated through cluster-level dynamic programming, and then dynamic programming path search is performed on the tasks within each cluster, taking into account the equivalent service time of temperature rise perception and collaborative maintenance rewards, to obtain the optimal access order and estimated arrival time of tasks within the cluster. Based on the optimal access order and estimated arrival time of the tasks within the cluster, a set of non-dominated path schemes containing task execution order, estimated arrival time, and collaborative maintenance suggestions is obtained through a non-dominated sorting method.

7. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 6, characterized in that, Based on the enhanced task list, composite travel time matrix, collaborative maintenance reward calculation function, and work group list, a complete state transition model containing cost functions and constraint rules is obtained by constructing a normalized reference value through a single-objective greedy algorithm, defining a seven-tuple state space, and building a set of state transition feasibility checks. This includes: A greedy construction algorithm was run with the single objective of minimizing the total effective travel time, minimizing the weighted repair waiting time, minimizing the imbalance of the workload of the shift team, and maximizing the cumulative collaborative maintenance reward value, resulting in four normalized reference values. Based on the preset four-dimensional weight coefficients and the four normalized reference values, a multi-objective weighted total cost function is constructed through a linear weighted sum formula to obtain the single-step cost calculation rule for state transition; the four-dimensional weight coefficients include efficiency weight, reliability weight, equilibrium weight, and collaborative reward weight. Based on the enhanced task list, the state space is defined in the form of a seven-tuple, and four types of sequential check rules are constructed: feasibility of temperature and time window, feasibility of communication closed loop, feasibility of personnel role, and feasibility of collaborative maintenance time consumption. This results in a set of state transition feasibility check rules, and thus a complete state transition model.

8. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 1, characterized in that, The updated complete maintenance path plan, based on the set of non-dominated path solutions combined with real-time status snapshots and a list of newly added tasks, is obtained through reverse path re-inspection value assessment and remaining task replanning, including: Based on the set of non-dominated path schemes, a real-time state snapshot and a set of re-inspection point markers are obtained by state variable aggregation and breadth-first search tracing. Based on the real-time status snapshot, the list of newly added tasks, and the set of re-inspection point markers, a set of replanning tasks is obtained by merging the sets, and an updated complete maintenance path scheme is obtained based on the set of replanning tasks.

9. The dynamic planning method for junction box maintenance paths based on multi-objective optimization according to claim 1, characterized in that, The updated complete maintenance path scheme, through statistical fitting and rolling parameter updates, yields an updated model parameter configuration file and a system performance evaluation report, including: Based on the updated complete maintenance path scheme, an execution record dataset is obtained through data cleaning and structured storage; Based on the execution record dataset, an updated model parameter configuration file is obtained through median statistics, linear regression fitting, and comparative analysis. Based on the execution record dataset, a system performance evaluation report is obtained by calculating the time window satisfaction rate, travel time prediction accuracy, and actual benefits of collaborative maintenance.

10. A dynamic planning system for junction box maintenance paths based on multi-objective optimization, used to implement the dynamic planning method for junction box maintenance paths based on multi-objective optimization as described in any one of claims 1-9, characterized in that, The system includes: The data enhancement module is used to acquire a basic data set and obtain an enhanced task list, a composite travel time matrix, and a collaborative maintenance reward calculation function through temperature sensing equivalent service duration modeling, communication blind spot task closed-loop construction, and multi-loop collaborative maintenance reward. The multi-objective planning module is used to obtain a complete state transition model based on the enhanced task list, composite travel time matrix and collaborative maintenance reward calculation function through multi-objective normalization and state space definition. It then solves the model by nesting temperature communication sensing weighted clustering, cluster-level dynamic programming and intra-cluster dynamic programming, and obtains a set of non-dominated path schemes through Pareto front screening. The online replanning module is used to obtain an updated complete maintenance path plan based on the set of non-dominated path schemes, real-time status snapshots, and a list of newly added tasks, through reverse path re-inspection value assessment and remaining task replanning. The closed-loop optimization module is used to obtain the updated model parameter configuration file and system performance evaluation report based on the updated complete maintenance path scheme through statistical fitting and rolling parameter updates.

Images