A method for intelligent scheduling of emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning

Abstract

This invention discloses an intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning. The method includes: acquiring relevant repair data in real time and generating comprehensive disaster scheduling data after preprocessing; identifying high-risk power grid equipment within the typhoon's impact area based on the comprehensive disaster scheduling data and generating a list of tasks to be repaired; inputting the characteristic information corresponding to the tasks to be repaired into a fault priority assessment model to obtain repair priorities and generate a dynamic repair task sequence; generating an initial scheduling plan based on the dynamic repair task sequence, available repair team information, repair material information, and traffic condition data; generating an initial repair path and a backup repair path according to the initial scheduling plan, and dynamically adjusting the repair path during the journey based on updated traffic condition data. This method can improve the real-time performance, accuracy, and path adaptability of emergency repair resource scheduling during typhoon disasters, and shorten fault response time.

Description

Technical Field

[0001] This invention belongs to the field of power grid system dispatching technology, specifically relating to an intelligent dispatching method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning. Background Technology

[0002] During the passage of extreme weather disasters such as typhoons, the mid-disaster phase, power grid facilities often experience widespread power outages due to strong winds, torrential rains, landslides, fallen trees, and flooded roads. Rapid and efficient disaster relief is crucial for minimizing power outages and reducing economic losses and social impact. However, the disaster environment is extremely harsh and unpredictable. Not only are meteorological factors such as typhoon path, intensity, wind speed, and precipitation constantly changing, but road conditions, infrastructure condition, and accessibility to repair sites in affected areas can also change abruptly due to traffic congestion, flooding, damaged bridges, or other obstacles, posing significant challenges to resource allocation and route planning.

[0003] Existing power grid fault repair and dispatching methods largely rely on manual experience, static emergency plans, existing geographic information systems (GIS), and conventional operations research optimization algorithms. After a power outage, the dispatch center typically determines the repair sequence based on system alarm order, fault impact range, user importance, or basic topology, and plans routes for repair vehicles based on static road networks or conventional shortest path algorithms. Some existing technologies, such as CN121639171A, further introduce a power grid-transportation network joint simulation mechanism, integrating power grid monitoring data, GIS information, historical fault records, and traffic condition data. This, combined with the location of repair teams, available resources, traffic flow, and load transfer capacity, optimizes the allocation of repair tasks, vehicle routes, and the power restoration process. Such methods can improve the rationality of repair dispatching to a certain extent in typical distribution network fault repair scenarios.

[0004] However, existing technologies still have many shortcomings in typhoon disaster emergency repair scenarios. Current methods primarily rely on abnormal power grid monitoring or fault alarms after a fault occurs as the trigger, resulting in a highly reactive repair response that struggles to adapt to the constantly changing disaster environment of typhoon paths, intensities, and the extent of their influence. For power grid equipment that has not yet generated a clear fault alarm but is already in areas of strong winds, heavy rain, or high exposure risk, existing methods lack dynamic identification and risk differentiation capabilities, easily leading to untimely detection of high-risk equipment and delayed repair preparation.

[0005] Secondly, while existing methods may incorporate traffic condition data or road flow data, they typically rely primarily on road congestion, travel time, or static road accessibility as the main basis for route optimization. They fail to adequately consider sudden road blockages caused by typhoons, such as flooding, fallen trees, bridge damage, landslides, and traffic control. Once repair vehicles have departed, if a sudden blockage or road infrastructure damage occurs along the planned route, existing route planning methods struggle to determine in a timely manner whether the original route remains passable, and are unable to promptly develop stable and reliable alternative routes. This can easily lead to repair teams being obstructed en route, delaying repair efforts.

[0006] Furthermore, typhoon disaster relief and dispatch involve heterogeneous data from multiple sources, including meteorology, power grids, geography, transportation, and on-site disaster conditions. Existing methods typically use power grid monitoring data, GIS information, historical fault data, and transportation data as primary inputs, but they lack sufficient utilization of real-time weather forecast data, equipment spatial exposure status, micro-topographic conditions, unstructured disaster information, and road infrastructure damage information. Due to differences in time scale, spatial scale, data format, and update frequency among different data sources, the lack of effective data cleaning, spatiotemporal alignment, and correlation fusion mechanisms can easily lead to inaccurate fault risk assessment, unreasonable repair prioritization, and delayed route planning.

[0007] Furthermore, existing emergency repair task allocation methods primarily focus on factors such as the severity of the fault, the scope of impact, the importance of the user, the location of the repair team, and resource consumption. They lack comprehensive coordination regarding the safety risks to repair teams, the availability of materials, task waiting times, multiple concurrent faults, and resource capacity constraints during typhoon disasters. When multiple pieces of equipment fail simultaneously or multiple areas are affected by the disaster at the same time, problems such as uneven allocation of repair teams, mismatch between materials and task requirements, and difficulty in balancing low-risk routes with high-priority tasks can easily arise, limiting the response speed and utilization efficiency of repair resources.

[0008] Therefore, existing power grid fault repair and dispatch technologies still have technical problems when facing the highly dynamic, highly uncertain, and multi-point concurrent repair scenarios during typhoon disasters. These problems include lagging identification of high-risk equipment, inaccurate assessment of repair priorities, insufficient collaborative allocation of repair resources, poor adaptability of route planning to sudden road blockages, and insufficient capacity for dispatching backup routes. These issues make it difficult to meet the actual needs of rapid, safe, and continuous dispatch of repair resources during typhoon disasters. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a method for intelligent scheduling of emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning.

[0010] The objective of this invention can be achieved through the following technical solutions: This invention provides an intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning, comprising the following steps: S1. Real-time acquisition of typhoon disaster emergency repair data, and preprocessing of the typhoon disaster emergency repair data to generate comprehensive disaster dispatch data; wherein, the typhoon disaster emergency repair data includes typhoon meteorological data, power grid equipment data, geographical environment data and traffic condition data; S2. Based on the disaster-related comprehensive dispatch data, identify high-risk power grid equipment within the typhoon's impact range, and generate a list of tasks to be repaired based on the operating status of the high-risk power grid equipment. S3. Input the meteorological impact information, equipment operation information, spatial location information and load impact information corresponding to the list of tasks to be repaired into the fault priority assessment model to obtain the repair priority of each task to be repaired, and generate a dynamic repair task sequence based on the repair priority. S4. Based on the dynamic emergency repair task sequence, available emergency repair team information, emergency repair material information and traffic condition data, generate an initial scheduling scheme between the emergency repair team and the emergency repair task, so as to match the emergency repair resources with the emergency repair task. S5. Based on the initial scheduling plan and the traffic data, generate an initial repair route and a backup repair route for the repair team to the location of the repair task. S6. During the march of the emergency repair team, the traffic data is updated in real time. When traffic congestion or damage to road facilities is detected on the initial emergency repair route, the emergency repair route is dynamically adjusted based on the backup emergency repair route and the updated traffic data, and the adjusted emergency repair route is sent to the corresponding emergency repair team.

[0011] Furthermore, the typhoon meteorological data includes typhoon path data, typhoon intensity data, wind speed data, wind direction data, precipitation data, and typhoon impact range data; The power grid equipment data includes equipment ledger data, equipment spatial location data, equipment operating status data, fault alarm data, equipment importance data, and load impact data; The geographic environment data includes topographic data, elevation data, digital elevation data, and micro-topographic data; The traffic data includes road network topology data, road congestion data, road waterlogging data, road blockage data, and road facility damage data.

[0012] Furthermore, the preprocessing specifically includes: Anomaly screening was performed on the emergency repair data during the typhoon disaster, and abnormal data that exceeded the preset reasonable range was removed; Missing data in the emergency repair data related to the typhoon disaster were supplemented, and the data formats from different sources were standardized. The typhoon meteorological data, power grid equipment data, geographical environment data, and traffic condition data are time-aligned according to a unified time base. Spatial registration was performed on the typhoon meteorological data, power grid equipment data, geographical environment data, and traffic condition data according to unified spatial coordinates; Based on the data after time alignment and spatial registration, the disaster-related integrated dispatch data is generated by associating and fusing the data according to the spatial location of the equipment, the typhoon impact range, and the road network location.

[0013] Furthermore, the determination of high-risk power grid equipment within the typhoon's impact area based on the comprehensive disaster dispatch data specifically includes: Extract the current typhoon center location, typhoon impact range, and spatial location of each power grid device from the disaster comprehensive dispatch data; The typhoon center location and the equipment spatial location are uniformly converted into latitude and longitude coordinates, and then converted into radians when performing trigonometric function calculations to obtain the first... Equipment space location of individual power grid devices and the current moment Typhoon center location ; Based on the spatial location of the equipment and the location of the typhoon center, calculate the first... The spatial distance of each power grid device relative to the center of the typhoon ,in: in, Indicates the first At the current moment, each power grid device Relative to the spatial distance from the typhoon center, Represents the Earth's average radius. and They represent the first The longitude and latitude of each power grid device, and They represent the current time. The longitude and latitude of the typhoon center; Extract the current time from the typhoon impact range data. typhoon impact radius When the spatial distance Less than or equal to the typhoon's radius of influence At that time, the corresponding power grid equipment will be identified as candidate power grid equipment within the typhoon's impact range; Extract typhoon meteorological data, geographical environment data and power grid equipment data corresponding to the candidate power grid equipment from the disaster-in-disaster integrated dispatch data, and determine the meteorological exposure status, geographical exposure status and equipment operation status of the candidate power grid equipment; When the candidate power grid equipment has a fault alarm, abnormal operating status, or when the meteorological exposure status and geographical exposure status of the candidate power grid equipment reach the preset risk conditions, the corresponding candidate power grid equipment will be identified as the high-risk power grid equipment.

[0014] Furthermore, the process of generating a list of tasks to be repaired based on the operating status of the high-risk power grid equipment specifically includes: The operating status of the high-risk power grid equipment is obtained, including fault alarm status, abnormal equipment operating parameter status, power outage status, and load loss status. Based on the operating status, the high-risk power grid equipment is identified in the task status judgment, and the high-risk power grid equipment that is in fault alarm status, abnormal equipment operating parameter status, or power outage status is identified as the objects to be repaired. Based on the equipment type, spatial location, fault alarm status, abnormal status of equipment operating parameters, power outage status, and load loss status of the object to be repaired, a corresponding repair task is generated. Based on the equipment type and fault alarm status of the object to be repaired, determine the emergency repair materials required for the emergency repair task and the estimated emergency repair workload; The meteorological impact information, equipment operation information, spatial location information, load impact information, emergency repair material information, and estimated emergency repair workload corresponding to the task to be repaired are written into the task data item. The task data items generated repeatedly for the same object to be repaired within adjacent time windows are merged to form the list of tasks to be repaired.

[0015] Furthermore, the fault priority assessment model is constructed based on an ensemble learning model and a multi-attribute decision model, and the fault priority assessment model includes a feature extraction module, a fault risk prediction module, a risk fusion module, and a priority quantification module. The input to the fault priority assessment model is the meteorological impact information, equipment operation information, spatial location information and load impact information corresponding to the list of tasks to be repaired, and the output is the repair priority of each task to be repaired. The feature extraction module is used to extract a task feature vector for each task to be repaired. The task feature vector includes meteorological impact features formed by the meteorological impact information, equipment operation features formed by the equipment operation information, spatial exposure features formed by the spatial location information, and load impact features formed by the load impact information. The fault risk prediction module is used to input the task feature vector into the gradient boosting tree model and the random forest model respectively to obtain the first fault risk value and the second fault risk value of the corresponding task to be repaired. The risk fusion module is used to perform weighted fusion of the first fault risk value and the second fault risk value to obtain a comprehensive fault risk value for the corresponding emergency repair task, wherein: in, Indicates the first The overall fault risk value of each task awaiting emergency repair. This represents the first output of the gradient boosting tree model. The first fault risk value for each task awaiting emergency repair. The output of the random forest model represents the first... The second fault risk value for a task awaiting emergency repair. Indicates the fusion weight, and The value of is between 0 and 1; The priority quantification module is used to obtain the repair priority of the corresponding repair task based on the comprehensive fault risk value, equipment importance, load impact information and waiting time for emergency repair.

[0016] Furthermore, S3 specifically includes: Read the tasks to be repaired one by one from the list of tasks to be repaired, and extract the meteorological impact information, equipment operation information, spatial location information and load impact information of the corresponding tasks to be repaired from the disaster comprehensive scheduling data; Based on the meteorological impact information, equipment operation information, spatial location information, and load impact information, a task feature vector corresponding to the task to be repaired is constructed, and the task feature vector is input into the fault priority assessment model to obtain the comprehensive fault risk value of the corresponding task to be repaired. Based on the comprehensive fault risk value, equipment importance, load impact value, and waiting time for emergency repair, the data is input into the priority quantification module to calculate the priority index of the corresponding emergency repair task, where: in, Indicates the first The priority index of each task awaiting emergency repair. Indicates the first The overall fault risk value of each task awaiting emergency repair. Indicates the first The load impact value corresponding to each emergency repair task. Indicates the first The importance of the equipment corresponding to each emergency repair task. Indicates the first The correction value for the waiting time for each pending repair task. , and These represent the weighting coefficients corresponding to the comprehensive failure risk value, equipment importance, and repair wait time correction value, respectively. The priority index is used as the repair priority of the corresponding emergency repair task, and the emergency repair tasks in the emergency repair task list are sorted from high to low according to the repair priority to generate the dynamic emergency repair task sequence. When the data related to emergency repairs during the typhoon disaster is updated, or when tasks to be repaired are added, deleted, or merged in the list of tasks to be repaired, the updated meteorological impact information, equipment operation information, spatial location information, and load impact information are re-extracted, and the repair priority of each task to be repaired and the dynamic repair task sequence are updated.

[0017] Furthermore, S4 specifically includes: Read the tasks to be repaired from the dynamic emergency repair task sequence, and obtain the repair priority, location, repair materials information and estimated repair workload of each task to be repaired; Obtain the information on available emergency repair teams, which includes the current location of the emergency repair teams, their operational capabilities, their availability status, and information on the materials they are carrying. Based on the traffic data, the predicted travel time for each repair team to reach the location of each repair task is determined, and based on the estimated repair workload and the operational capacity of the repair teams, the estimated repair time for each repair team to handle each repair task is determined. Using the repair teams as the allocation objects and the tasks to be repaired as the task objects, allocation decision variables between the repair teams and the tasks to be repaired are established. Initial scheduling evaluation values ​​are generated based on the predicted travel time, the estimated repair time, the repair priority, and the disaster safety risks, wherein: in, This represents the initial scheduling evaluation value. This indicates the number of available repair teams. This indicates the number of tasks requiring emergency repair. Indicates the first The emergency repair team and the first The allocation decision variables among the pending repair tasks, when the... The emergency repair team was assigned to the first The value is 1 when there is an emergency repair task pending; otherwise, the value is 0. Indicates the first The emergency repair team handled the first The comprehensive response time required for each pending repair task is determined by the predicted traffic time and the estimated repair time. Indicates the first The emergency repair team carried out the first The corresponding disaster safety risks when an emergency repair task is pending. Indicates the first The priority of each pending repair task. , and These represent the weighting coefficients corresponding to the overall response time, disaster safety risk, and repair priority, respectively. With the goal of minimizing the initial scheduling evaluation value, and under the conditions of satisfying the task allocation constraints, the operational capacity constraints of the emergency repair team, and the matching constraints of emergency repair materials, the allocation relationship between the emergency repair team and the emergency repair task is optimized to obtain the candidate scheduling scheme between the emergency repair team and the emergency repair task. The task allocation constraints include that the same emergency repair task can be assigned to at most one available emergency repair team; the emergency repair team's operational capacity constraints include that the expected emergency repair workload corresponding to the emergency repair task assigned to the same emergency repair team does not exceed the operational capacity of that emergency repair team; and the emergency repair material matching constraints include that the information on the materials already carried by the emergency repair team matches the information on the emergency repair materials required for the emergency repair task, or that insufficient materials are replenished through the material sources determined by the emergency repair material information. The initial scheduling scheme is selected from the candidate scheduling schemes to meet the current disaster scheduling objectives. The initial scheduling scheme includes the correspondence between the tasks to be repaired and the repair teams, the execution order of the repair teams, the source of the required repair materials, and the estimated arrival time.

[0018] Furthermore, S5 specifically includes: According to the initial scheduling scheme, the location of the repair task corresponding to each repair team is determined, and the current location of the repair team is taken as the starting point of the path, and the location of the corresponding repair task is taken as the ending point of the path. From the traffic condition data, road network topology data, road congestion data, road water accumulation data, road blockage data, and road facility damage data are extracted to construct a dynamic road network corresponding to the current time. Based on the traffic status of each road segment in the dynamic road network, the traffic cost of each road segment is determined, wherein: in, Indicates the current time Lower road network nodes With road network nodes The value of passage between the two sections of road Represents road network nodes With road network nodes Basic travel time between the road sections Indicates the current time The degree of road congestion in this section of the road. Indicates the current time The impact value of road water accumulation on this section of the road. Indicates the current time The status value of whether there is road blockage or damage to road facilities on this section of the road is as follows. This indicates the adjustment coefficient corresponding to the degree of road congestion. This represents the adjustment coefficient corresponding to the impact value of road water accumulation. The penalty coefficient represents the amount of time required to prevent road blockage or damage to road infrastructure; the basic travel time... , Represents road network nodes With road network nodes The length of the road segment between them Represents road network nodes With road network nodes The basic traffic speed of the road section under normal traffic conditions; Based on the toll value of each road segment, a set of candidate emergency repair routes from the starting point of the route to the ending point of the route is searched in the dynamic road network, and the sum of the toll values ​​of each road segment in each candidate emergency repair route is taken as the total toll value of the corresponding candidate emergency repair route. The path with the lowest total value of passage and which meets the conditions for safe passage of the repair team is selected from the set of candidate repair paths and is taken as the initial repair path. From the candidate emergency repair route set, select the route whose overlap with the initial emergency repair route is lower than a preset ratio and whose total passage value meets the preset backup conditions, and use it as the backup emergency repair route. The initial repair path and the backup repair path are associated with the corresponding repair teams and the corresponding repair tasks to be repaired, forming repair path data, and the repair path data is written into the initial scheduling scheme.

[0019] Furthermore, S6 specifically includes: During the march of the repair team, the current location of the repair team, the road condition information reported by the vehicle terminal, and the updated traffic condition data are obtained in real time, and the updated traffic condition data are matched with the road segments in the initial repair route and the backup repair route. Based on the updated traffic data, the traffic status and cost of each segment in the initial repair route are updated, and the total cost of the initial repair route at the current time is calculated, where: in, Representing a path At the present moment The total value of the path agency Representing a path Central network nodes With road network nodes The section of road between, Indicates the current time The toll value of the following road sections; If there are abnormal road sections in the initial emergency repair route that are congested, blocked, flooded, or damaged, or if the total route value of the initial emergency repair route at the current time increases by more than a preset adjustment threshold compared to the total route value when the initial emergency repair route was issued, it is determined that the initial emergency repair route does not meet the current traffic conditions. After determining that the initial repair route does not meet the current traffic conditions, the current location of the repair team is used as the new route starting point, and the location of the corresponding repair task is used as the route ending point. The adjusted repair route is determined based on the backup repair route and the updated traffic data. When the backup repair route meets the current traffic conditions, the access node of the backup repair route is determined at the current location of the repair team, and the path connecting segment from the current location of the repair team to the access node and the remaining segment of the backup route from the access node to the location of the task to be repaired are spliced ​​together to obtain the adjusted repair route. When the backup repair route does not meet the current traffic conditions, the local road network affected by the abnormal road section is re-searched based on the updated traffic data to generate an alternative repair route from the current location of the repair team to the location of the repair task, and the alternative repair route is determined as the adjusted repair route. The adjusted repair path is associated with the corresponding repair team and the corresponding repair task, the repair path data in the initial scheduling scheme is updated, and the adjusted repair path is sent to the corresponding repair team.

[0020] Compared with the prior art, the present invention has the following advantages: (1) In the context of typhoon disaster emergency repairs, existing technologies typically rely on fault alarms or SCADA anomalies as the primary trigger for emergency repair scheduling, which can easily lead to delayed emergency repair responses. In particular, for power grid equipment that has not yet issued a clear fault alarm but is already within the typhoon's strong wind circle, rainstorm area, or terrain risk zone, existing methods struggle to identify potential fault risks in a timely manner, resulting in emergency repair resources being passively deployed only after the fault has escalated. This invention acquires typhoon meteorological data, power grid equipment data, geographical environment data, and traffic condition data in real time, and generates comprehensive disaster-related dispatch data. Based on the typhoon's impact range and the spatial location of power grid equipment, it identifies high-risk power grid equipment, enabling the dispatch targets to extend beyond equipment that has already failed to cover equipment in a high-risk state. This improves the proactiveness and predictability of disaster-related emergency repair scheduling, and reduces cascading power outages and missed repair windows caused by untimely identification of risky equipment.

[0021] (2) Although existing technologies can determine repair priorities based on factors such as fault severity, impact range, and user importance, in typhoon disaster scenarios, fault risk is often determined by the typhoon path, wind speed changes, precipitation intensity, equipment spatial exposure, equipment operating status, and load impact. Simply ranking based on fault information or static task urgency is insufficient to accurately reflect the process of fault evolution from potential risk to actual fault. This invention inputs meteorological impact information, equipment operating information, spatial location information, and load impact information corresponding to the task to be repaired into the fault priority assessment model, and calculates the repair priority by combining the comprehensive fault risk value, equipment importance, load impact value, and waiting time for repair. This ensures that the repair sequence reflects both the probability of fault occurrence and the degree of load loss, while also taking into account the processing needs of important equipment and tasks with long waiting times. This avoids high-risk tasks being occupied by low-value tasks and also prevents the long-term backlog of some medium- and low-priority tasks from leading to the expansion of secondary risks.

[0022] (3) In the prior art, task allocation usually focuses on the location of the repair team, available resources, and the urgency of the task. When facing multiple concurrent failures in a typhoon disaster, problems such as the closest location but mismatched materials, the highest priority but excessive risk of passage, and repair teams with insufficient capacity being assigned tasks are easily encountered. This leads to the need for secondary adjustments during actual execution, reducing repair efficiency. This invention generates an initial scheduling plan based on dynamic repair task sequences, available repair team information, repair material information, and traffic condition data. In the scheduling evaluation, it comprehensively considers the predicted passage time, expected repair time, repair priority, and safety risks in the disaster, so that a unified matching relationship is formed between repair teams, repair tasks, repair materials, and road conditions. This reduces the probability of the scheduling plan failing during the execution phase due to capacity mismatch or insufficient materials, and improves resource utilization efficiency and task completion stability during multi-point concurrent repairs.

[0023] (4) Existing technologies, even when considering traffic flow or road congestion, typically prioritize vehicle arrival time or the shortest path as the main optimization objective. They fail to adequately consider sudden road disasters such as road flooding, road blockages, bridge damage, and fallen trees during typhoons. This results in planned routes that appear feasible at the start but may become completely ineffective during travel due to physical blockages. This invention constructs a dynamic road network based on road network topology data, road congestion data, road flooding data, road blockage data, and road facility damage data. It determines the road segment passability value based on basic travel time, congestion level, flooding impact, and blockage status. This allows road conditions to be expressed not only by congestion time but also by incorporating physical damage and impassable conditions of roads during disasters into route planning. This improves the adaptability of emergency repair routes to sudden road changes under extreme weather disasters and reduces the risk of emergency repair vehicles being blocked or losing control while detouring.

[0024] (5) Existing technologies typically use the determined emergency repair route as the basis for vehicle execution, resulting in insufficient backup plans. If the original route is suddenly blocked, the dispatch system needs to recalculate the route or rely on manual command, which can easily cause emergency repair vehicles to stagnate and wait. This delay is even more pronounced when communication is unstable or road information changes rapidly. This invention generates a backup emergency repair route at the same time as the initial emergency repair route, and requires that the backup emergency repair route maintain a certain difference from the initial emergency repair route. This ensures that the backup route is not a simple local variation of the original route, but still has the possibility of independent passage when key road sections are blocked. Therefore, when the initial emergency repair route is congested or road facilities are damaged, it can be quickly switched or adjusted in combination with updated traffic data, shortening the route replanning time and improving the continuous travel capability of emergency repair vehicles.

[0025] (6) In existing technologies, route optimization is mostly completed before the emergency repair task is issued. There is insufficient continuous verification of route passability during the march of the emergency repair team, resulting in the inability to promptly reflect changes in the risk of the original route to the scheduling plan. This invention updates traffic condition data in real time during the march of the emergency repair team and matches the updated traffic condition data with road segments in the initial emergency repair route and backup emergency repair route. When traffic congestion, road blockage, excessive road water accumulation, or damage to road facilities are detected, the emergency repair route is dynamically adjusted based on the backup emergency repair route and the updated traffic condition data and then sent to the corresponding emergency repair team. This transforms route planning from a one-time static planning to a closed-loop adjustment during the march, thereby reducing the impact of sudden changes in road conditions during a disaster on the timeliness of emergency repairs and improving the real-time performance and safety of emergency repair scheduling.

[0026] (7) Existing technologies tend to prioritize the shortest arrival time in emergency repairs during extreme disasters. However, during typhoons, repair teams are themselves exposed to strong winds, torrential rain, road flooding, and facility damage. Simply pursuing speed may lead to repair vehicles entering high-risk sections or highly exposed areas, resulting in repair interruptions or even personnel risks. This invention introduces disaster safety risks into the initial dispatching scheme and determines the cost of passage by combining road flooding, road blockages, and facility damage status in route planning. This allows the dispatching results to comprehensively balance repair efficiency and personnel safety, thereby avoiding repair resources from entering obviously impassable or high-risk areas. This ensures timely repairs while improving the safety and sustainability of disaster-related repair operations.

[0027] (8) In the prior art, when multiple faults occur simultaneously, the task list is prone to expansion due to repeated alarms, repeated tasks within adjacent time windows, or multiple types of abnormal information of the same equipment, resulting in repeated dispatching of orders by the dispatching system or repeated occupation of repair resources by the same fault point. The present invention generates repair tasks based on the operating status of high-risk power grid equipment and merges the task data items repeatedly generated for the same repair object within adjacent time windows, so that the repair task list can maintain a consistent correspondence with the actual repair object, thereby reducing repeated scheduling, repeated dispatching and task conflicts, improving the accuracy of dynamic repair task sequences, and making subsequent priority assessment and resource allocation more stable. Attached Figure Description

[0028] Figure 1 This is a general flowchart of an embodiment of the present invention; Figure 2 This is a flowchart illustrating the identification and emergency repair priority assessment of high-risk power grid equipment according to an embodiment of the present invention. Figure 3 This is a flowchart of emergency repair scheduling and dynamic path adjustment according to an embodiment of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0030] like Figure 1 , Figure 2 , Figure 3 As shown, this embodiment specifically provides a method for intelligent scheduling of emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning, including the following steps: S1. Real-time acquisition of emergency repair data during typhoon disasters, and preprocessing of the emergency repair data during typhoon disasters to generate comprehensive dispatch data during disasters; In one specific implementation, step S1 can be implemented as follows: During the typhoon's passage, the dispatch platform accesses typhoon disaster emergency repair data according to a preset collection cycle. This data includes typhoon meteorological data, power grid equipment data, geographical environment data, and traffic condition data. After all types of data are accessed, they are uniformly written into the disaster data cache, and data indexes are established according to collection time, spatial location, and data source to facilitate subsequent identification of high-risk power grid equipment, generation of emergency repair tasks, assessment of repair priorities, and route planning.

[0031] Typhoon meteorological data can be provided jointly by meteorological monitoring stations, radar echo data, typhoon track prediction data, and gridded meteorological forecast data. Specifically, it includes typhoon track data, typhoon intensity data, wind speed data, wind direction data, precipitation data, and typhoon impact area data. Typhoon track data characterizes the movement trajectory of the typhoon center over time; typhoon intensity data characterizes the typhoon's central pressure, maximum wind speed, or wind force level; wind speed and direction data are used to determine the direction and intensity of wind loads on power grid lines, towers, substations, and other facilities; precipitation data is used to assess the risk of road flooding, landslides, and equipment dampness; and typhoon impact area data delineates spatial areas that may be affected by strong winds or heavy precipitation at different times. By simultaneously collecting measured and predicted meteorological data, the problem of delayed dispatch response caused by relying solely on a single real-time observation can be avoided, enabling comprehensive disaster dispatch data to reflect the evolving trend of the typhoon's impact area.

[0032] Power grid equipment data can be provided by distribution automation systems, online monitoring devices, fault alarm systems, and equipment ledger systems. Specifically, it includes equipment ledger data, equipment spatial location data, equipment operating status data, fault alarm data, equipment importance data, and load impact data. Equipment ledger data includes equipment type, years of operation, associated line, associated substation, and maintenance records. Equipment spatial location data includes coordinates of poles, switching stations, distribution rooms, line segmentation points, and substations. Equipment operating status data includes voltage, current, switch status, temperature, humidity, and communication status. Fault alarm data includes trip alarms, grounding alarms, overcurrent alarms, communication interruption alarms, and protection action information. Equipment importance data can be determined based on the power supply range, user type, and whether it involves critical loads such as hospitals, transportation hubs, and emergency command centers. Load impact data characterizes the potential power outage impact range and load loss scale caused by equipment anomalies or faults. By simultaneously incorporating static equipment ledger data and real-time operating status into the data scope, reactive repairs based solely on alarm information can be avoided, improving the completeness of repair task identification.

[0033] Geographic environmental data can be provided by geographic information systems, digital elevation models, topographic databases, and historical disaster distribution data, specifically including topographic data, elevation data, digital elevation data, and micro-topographic data. Topographic data is used to distinguish spatial environments such as plains, mountains, coastal low-lying areas, and densely populated urban areas; elevation data and digital elevation data are used to assess water accumulation, slope runoff, and landslide risk; micro-topographic data is used to identify localized high-risk locations such as windward slopes, leeward slopes, low-lying road sections near rivers, and bridge and culvert areas. Combining geographic environmental data with meteorological data can reflect the actual exposure differences of different power grid facilities and road sections under the same wind and rain conditions, making subsequent high-risk area assessments more closely reflect the on-site situation during disasters.

[0034] Traffic data can be jointly provided by urban traffic management platforms, navigation service platforms, road video recognition results, emergency repair vehicle terminals, and on-site disaster reporting information. Specifically, it includes road network topology data, road congestion data, road flooding data, road blockage data, and road infrastructure damage data. Road network topology data describes road nodes, road connections, traffic directions, and road grades; road congestion data reflects current vehicle speed, traffic index, and queue length; road flooding data reflects road flooding depth and passability risk; road blockage data characterizes road closures, traffic control, and tree falls blocking access; and road infrastructure damage data characterizes the damage status of bridges, culverts, tunnels, slopes, and critical road facilities under typhoon influence. Traffic data is not only used to calculate the travel time of emergency repair vehicles but also to determine whether a route is actually passable, avoiding the misclassification of impassable roads as viable routes during the disaster phase.

[0035] During data preprocessing, the scheduling platform first standardizes the format and fields of the incoming data, converting data from different sources into a unified data structure. For example, the time field is standardized to the same time base, the spatial field to the same coordinate system, and a correspondence is established between equipment number, line number, road number, and area number. For data of the same object from multiple systems, matching is performed according to object code, spatial location, and timestamp to form a complete data record.

[0036] To address data quality issues caused by communication disruptions, sensor malfunctions, and delays in manual reporting during disasters, the dispatch platform conducts anomaly screening and data completion for typhoon-related emergency repair data. Anomaly screening can be performed by considering physical boundaries and historical fluctuation ranges. For example, data such as negative wind speeds, abrupt changes in precipitation, equipment voltage exceeding identifiable ranges, and missing road congestion indices with abnormally zero vehicle speeds can be marked as anomalies and entered into the review process. For short-term missing continuous monitoring data, time interpolation can be used for completion; for meteorological gaps in the spatial grid, spatial completion can be performed by combining surrounding grids or nearby monitoring points.

[0037] In one alternative implementation, spatially missing values ​​in gridded meteorological data can be imputed using an inverse distance-weighted method: in, Indicates the position to be completed At any moment The meteorological data completion value, Indicates the first neighboring sampling locations At any moment Meteorological observations, This indicates the number of neighboring sampling locations that participated in the completion. Indicates the position to be completed and neighboring sampling locations Spatial distance between them Indicates the first The weights corresponding to the nearest sampling locations This represents the distance attenuation coefficient. Weights are assigned based on the principle that closer locations have a greater impact, ensuring that meteorological conditions in neighboring areas contribute more significantly to the missing location. This approach better adapts to the characteristics of typhoon precipitation and wind speed, which vary spatially but exhibit significant local differences. This completion method reduces the impact of meteorological monitoring blind spots on the identification of high-risk equipment and the assessment of path risks.

[0038] After anomaly screening and missing data completion, the dispatch platform aligns typhoon meteorological data, power grid equipment data, geographic environment data, and traffic condition data according to a unified time benchmark. For data with different update frequencies, more real-time data can be mapped to a unified dispatch time window. For example, typhoon path prediction data is mapped to the current dispatch time according to the prediction release time and prediction lead time; equipment operation status data is entered into the current dispatch time according to the most recently valid sampled value; and traffic condition data is entered into the current dispatch time according to the latest road conditions. Time alignment avoids using data from different time slices in the same dispatch decision.

[0039] The dispatch platform also performs spatial registration of various data according to a unified spatial coordinate system. The spatial locations of power grid equipment, typhoon center locations, meteorological grids, road nodes, and road sections are all transformed into a unified coordinate system, and then associations are established based on the spatial relationships between equipment locations, road sections, and the typhoon's impact area. For example, power grid equipment is mapped to the corresponding meteorological grid, the location of the equipment is associated with the nearest road node or reachable road section, and information on road flooding or road damage is associated with the corresponding road network edge. Through spatial registration, equipment risk, road risk, and meteorological risk can be calculated within the same spatial range.

[0040] After data cleaning, time alignment, and spatial registration, the dispatch platform integrates and correlates data based on equipment spatial location, typhoon impact range, and road network location relationships to generate comprehensive disaster dispatch data. This comprehensive disaster dispatch data can be stored using a combination of structured data tables, spatial index tables, and dynamic status tables. The structured data tables record equipment status, meteorological status, traffic status, and task status; the spatial index tables record the spatial relationships between equipment, roads, meteorological grids, and disaster-stricken areas; and the dynamic status tables record equipment risk, road accessibility, and data update time at different dispatch times. This data organization method allows subsequent steps to directly access meteorological impact, operational status, spatial exposure, load impact, and road accessibility information corresponding to the same equipment, reducing computational delays caused by repeated matching of multi-source data.

[0041] S2. Based on the comprehensive dispatch data during the disaster, identify high-risk power grid equipment within the typhoon's impact range, and generate a list of tasks to be repaired based on the operating status of the high-risk power grid equipment. In one specific implementation, step S2 can be implemented as follows: Within each scheduling time window, the scheduling platform reads the current typhoon center location, typhoon impact range data, and spatial location of each power grid device from the disaster comprehensive scheduling data. Power grid devices may include transmission towers, distribution line segmentation points, ring main units, switching stations, distribution rooms, and substations, etc. The typhoon center location is obtained from real-time meteorological monitoring or typhoon path prediction results, and the spatial location of power grid devices is obtained from equipment ledgers or geographic information systems. To ensure that the typhoon location, power grid device location, and subsequent road network location can be calculated under the same spatial reference, the scheduling platform first converts coordinates from different sources into latitude and longitude coordinates, and converts latitude and longitude angles into radians before performing trigonometric function calculations.

[0042] The typhoon center location and equipment spatial location were uniformly converted into latitude and longitude coordinates, and then converted into radians during trigonometric function calculations to obtain the [missing information]. Equipment space location of individual power grid devices and the current moment Typhoon center location ; Based on the spatial location of the equipment and the location of the typhoon center, calculate the... The spatial distance of each power grid device relative to the center of the typhoon ,in: in, Indicates the first At the current moment, each power grid device Relative to the spatial distance from the typhoon center, Represents the Earth's average radius. and They represent the first The longitude and latitude of each power grid device, and They represent the current time. The longitude and latitude of the typhoon center are used; a spherical distance is used instead of a straight-line distance because the dispatching range during a typhoon disaster usually covers multiple areas including cities, suburbs, and coastal areas, with a large span between equipment locations. Using a spherical distance can reduce the spatial error caused by directly subtracting latitude and longitude coordinates, making the judgment of the typhoon's impact range more stable.

[0043] The dispatch platform extracts the current time from the typhoon impact range data. typhoon impact radius When meteorological departments provide different wind circle radii, the radius of a level 7, level 10, or level 12 wind circle can be selected based on the actual dispatch level. When the meteorological department provides the radii of the wind circles in different quadrants, the quadrant in which the power grid equipment is located can be determined first based on the azimuth of the equipment relative to the typhoon center, and then the wind circle radius of the corresponding quadrant can be selected as the radii. .when When, explain the first A power grid device is located within the current typhoon's impact area, and the dispatch platform marks this power grid device as a candidate power grid device; when At that time, the power grid equipment will not be included in the high-risk equipment identification queue at the current moment, but it can still participate in the identification again in the next scheduling time window.

[0044] After identifying candidate power grid equipment, the dispatch platform continues to extract typhoon meteorological data, geographical environment data, and power grid equipment data corresponding to the candidate equipment from the comprehensive disaster dispatch data. This data is used to determine the meteorological exposure status, geographical exposure status, and equipment operating status of the candidate power grid equipment. The meteorological exposure status can be determined based on data such as wind speed, precipitation intensity, and the angle between wind direction and line alignment within the meteorological grid where the equipment is located. The geographical exposure status can be determined based on data such as the altitude, terrain slope, micro-topography windwardness, and low-lying area of ​​the equipment's location. The equipment operating status can be determined based on fault alarms, abnormal operating parameters, power outage status, communication status, and load loss status.

[0045] To make the determination of whether meteorological and geographical exposure conditions meet preset risk conditions more precise, the dispatch platform can calculate exposure risk values ​​for candidate power grid equipment. : in, Indicates the first The candidate power grid devices at the current moment Exposure risk value, Indicates the first Real-time wind speed at the location of each candidate power grid device. This represents the normalized baseline value for wind speed. Indicates the first The intensity or cumulative precipitation at the location of each candidate power grid device. This represents the normalized baseline value for precipitation. Indicates the first Geographic exposure factors corresponding to each candidate power grid device. This indicates the influencing factor between the current wind direction and the equipment routing. , , and These represent the weighting coefficients corresponding to wind speed, precipitation, geographical exposure, and wind direction angle, respectively.

[0046] By incorporating wind speed, precipitation, geographical environment, and wind direction angle into the same risk quantification index, it is understood that damage to power grid equipment during typhoons is not determined by a single meteorological factor. For example, strong winds can easily cause conductors to gallop, towers to tilt, or foreign objects to slack on them; heavy precipitation can easily cause equipment to become damp, roads to flood, and foundations to loosen; low-lying or windward terrain can amplify the severity of localized damage; and the angle between the wind direction and the power line alignment can affect the direction of wind loads on the power lines. By normalizing the data before weighting the calculations, it is possible to avoid judgment bias caused by directly adding data of different dimensions, making the risk assessment of candidate power grid equipment more consistent with the actual situation on the ground during the disaster.

[0047] In one alternative implementation, geographical exposure factor It can be determined by the following formula: in, Indicates the first Geographic exposure factors corresponding to each candidate power grid device. Indicates the first The normalized slope value of the location of each candidate power grid device Indicates the first Normalized values ​​of low-lying water accumulation sensitivity at the locations of candidate power grid equipment. Indicates the first Normalized values ​​of windward exposure of the micro-topography at the location of each candidate power grid device. Indicates the first Normalized values ​​of historical disaster sensitivity in the region where each candidate power grid device is located. These represent the weighting coefficients corresponding to slope, sensitivity to water accumulation in low-lying areas, windward exposure of micro-topography, and sensitivity to historical disaster damage, respectively.

[0048] Including topographic slope, low-lying areas with water accumulation, windward exposure, and historical damage in the geographical exposure factor is because the risk of equipment damage during typhoons depends not only on wind speed and precipitation intensity but also on the local topography where the equipment is located. Equipment located on windward slopes, in low-lying areas, or in areas with a high incidence of historical damage is more prone to pole collapse, water ingress, foundation loosening, or external force damage, even under the same meteorological conditions. By normalizing and weighting different geographical factors, the geographical exposure factor can more accurately reflect the disaster amplification effect of the equipment's location.

[0049] Influencing factors between wind direction and equipment routing It can be determined by the following formula: in, Indicates the current time Wind direction and the Influence factors between the corresponding line routes of each candidate power grid device. Indicates the current time The angle between the wind direction and the route direction, Indicates the current time The wind angle, Indicates the first The line direction angle of each candidate power grid device.

[0050] By adopting The influence of wind direction is characterized because the closer the wind direction is to being perpendicular to the line's orientation, the greater the lateral wind load on the line, and the higher the risk of conductor galloping, tower stress shifting, and foreign object snagging. Conversely, the closer the wind direction is to being parallel to the line's orientation, the smaller the lateral effect. This influencing factor allows the differences in disaster impact caused by different line orientations under the same wind speed conditions to be incorporated into the high-risk equipment identification process, making the assessment of high-risk equipment more precise.

[0051] In one specific determination method, if a candidate power grid device has a fault alarm, abnormal operating status, power outage, or communication interruption, the dispatching platform can directly identify the candidate power grid device as a high-risk power grid device; if the candidate power grid device has not yet shown a clear fault alarm, but has exposed a risk value... If a device reaches a preset risk threshold and its geographical exposure status indicates that it is located in a low-lying area, windward slope, coastal area with strong winds, or area with a history of high-risk disasters, it can also be identified as a high-risk power grid device. This approach allows for the simultaneous inclusion of both devices that have already failed and those that are highly likely to fail in the near future into the emergency repair scheduling, preventing the scheduling system from only being able to respond passively after a failure occurs.

[0052] After identifying high-risk power grid equipment, the dispatch platform obtains its operational status. Operational status can include fault alarm status, abnormal equipment operating parameters, power outage status, and load loss status. Fault alarm status can include trip alarms, grounding alarms, overcurrent alarms, protection operation alarms, and communication interruption alarms; abnormal equipment operating parameters can include voltage exceeding limits, sudden current changes, abnormal temperature, abnormal humidity, and abnormal switch status; power outage status can be determined based on the affected users, the affected transformer area, or the affected line segment; load loss status can be determined based on the current power outage load, the impact on important users, and the difficulty of power restoration.

[0053] The dispatch platform determines the task status of high-risk power grid equipment based on its operational status. If a high-risk power grid equipment is in a fault alarm state, has abnormal operating parameters, or is in a power outage state, the dispatch platform identifies it as a device requiring emergency repair. For high-risk power grid equipment that is only at high exposure risk but has not yet experienced operational abnormalities, it can be included in the monitoring targets or reserve repair targets according to the dispatch strategy; when the equipment experiences operational abnormalities or its exposure risk continues to increase in a subsequent time window, it is then converted into a device requiring emergency repair. This tiered approach avoids the occupation of repair resources by a large number of non-faulty devices while maintaining continuous monitoring of potentially high-risk equipment.

[0054] For each device requiring emergency repair, the dispatching platform generates a corresponding repair task based on the equipment type, equipment location, fault alarm status, abnormal equipment operating parameters, power outage status, and load loss status. The repair task can include a task number, corresponding equipment number, equipment type, task location, fault type, fault level, load impact range, impact on key users, generation time, and task status. The equipment type is used to determine the repair operation method, the task location is used for subsequent path planning, the fault type and fault level are used to determine the repair difficulty, and the load impact range and impact on key users are used for subsequent repair priority assessment.

[0055] The dispatch platform also determines the required repair materials and estimated workload based on the equipment type and fault alarm status of the object to be repaired. Repair materials may include spare parts, wires, cables, fuses, switchgear, insulating materials, drainage tools, climbing tools, and emergency power generation equipment. The estimated workload can be determined comprehensively based on equipment type, fault type, historical repair time, site accessibility, and the difficulty of operations during a disaster. For example, the repair materials and procedures for water ingress in a power distribution room and line breakage differ, as do the personnel skill requirements for tower tilting and switch protection activation. Therefore, it is necessary to determine the material requirements and workload simultaneously when generating a repair task to facilitate the subsequent matching of repair teams with the repair task.

[0056] The dispatching platform writes the meteorological impact information, equipment operation information, spatial location information, load impact information, repair material information, and estimated repair workload corresponding to the tasks to be repaired into the task data items. To avoid repeated alarms from equipment during a disaster causing multiple repair tasks to be generated for the same equipment, the dispatching platform merges the task data items generated repeatedly within adjacent time windows for the same object to be repaired. During merging, the earliest generation time, latest operating status, highest fault level, maximum load impact value, and latest material requirements are retained, and the status record of the task data item is updated. By merging duplicate tasks, duplicate dispatching and resource consumption can be reduced, ensuring that the list of tasks to be repaired is consistent with the actual objects to be repaired.

[0057] S3. Input the meteorological impact information, equipment operation information, spatial location information and load impact information corresponding to the list of tasks to be repaired into the fault priority assessment model to obtain the repair priority of each task to be repaired, and generate a dynamic repair task sequence based on the repair priority. In one embodiment, step S3 can be implemented as follows: After generating a list of tasks to be repaired, the dispatching platform reads each task one by one and extracts the meteorological impact information, equipment operation information, spatial location information, and load impact information corresponding to each task from the comprehensive disaster dispatching data. Meteorological impact information may include real-time wind speed, wind speed change rate, wind direction, precipitation intensity, cumulative precipitation, and the relative distance between the task location and the typhoon center; equipment operation information may include equipment alarm type, voltage and current anomalies, switch status, communication status, historical fault frequency, and equipment service life; spatial location information may include equipment latitude and longitude, terrain slope, altitude, micro-topography exposure status, low-lying water accumulation sensitivity, and road accessibility; load impact information may include current power outage load, number of affected users, number of important users, and the scope of power restoration impact. By using meteorology, power grid operation, spatial environment, and load consequences as inputs, it avoids dispatching tasks solely based on alarm order or a single fault level, enabling repair priorities to reflect the comprehensive changes in fault risk and power outage impact during a disaster.

[0058] The fault priority assessment model can be built based on ensemble learning models and multi-attribute decision models. The model includes a feature extraction module, a fault risk prediction module, a risk fusion module, and a priority quantification module. The feature extraction module converts data from different sources and with different scales into task feature vectors that can participate in model calculations. The fault risk prediction module determines the fault risk of the equipment corresponding to the task to be repaired under the current typhoon impact conditions based on the task feature vectors. The risk fusion module merges the risk assessment results output by different learning models. The priority quantification module combines comprehensive fault risk, equipment importance, load impact, and waiting time for repair to form the final repair priority.

[0059] Regarding the first For each urgent repair task, the feature extraction module can construct a task feature vector. : in, Indicates the first The task feature vector of each task awaiting emergency repair. This indicates the meteorological impact characteristics formed by meteorological impact information. This represents the equipment operating characteristics formed by equipment operating information. This represents the spatial exposure characteristics formed by spatial location information. This represents the load impact characteristics formed by load impact information. The feature vector is grouped and concatenated to preserve the differences in the sources of various factors, facilitating the model's simultaneous learning of the nonlinear relationships between typhoon intensity, equipment status, spatial exposure, and load consequences. For example, under the same wind speed conditions, the probability of damage to overhead lines located on windward slopes differs from that of switching stations located in flat urban areas; similarly, for the same equipment anomaly, the urgency of repair differs between equipment carrying critical user loads and ordinary branch equipment.

[0060] Before inputting data into the model, the scheduling platform can normalize continuous features, encode discrete features such as alarm type, equipment type, and terrain category, and set data quality tags for missing but completed data. Normalization can reduce the numerical bias caused by different units of data such as wind speed, load loss, and distance on model training and inference; data quality tags can enable the model to maintain a certain fault tolerance when dealing with unstable communication data during disasters and reduce the impact of abnormal sampling on emergency repair scheduling.

[0061] The fault risk prediction module can include a gradient boosting tree model and a random forest model. The gradient boosting tree model is suitable for learning the progressive risk changes between meteorological intensity, equipment operating status, and load impact, and can form a relatively sensitive risk judgment for scenarios such as sudden increases in wind speed and continuous intensification of precipitation. The random forest model, through parallel voting of multiple decision trees, can reduce the risk of a single model being affected by abnormal samples or local noise. Using both types of models together can balance prediction sensitivity and stability, making it suitable for emergency repair scenarios during typhoon disasters where data changes rapidly, noise is abundant, and risk factors are intertwined.

[0062] During the training phase, training samples can be constructed using historical typhoon disaster repair records, equipment failure records, meteorological observation records, and load loss records. Each training sample uses the meteorological impact characteristics, equipment operation characteristics, spatial exposure characteristics, and load impact characteristics corresponding to a historical moment as input, and uses whether the equipment failed, the failure level, or the urgency of repair as labels. After the model training is completed, the current task feature vector is received and the risk value is output during the disaster scheduling phase. For the first... For each pending repair task, the gradient boosting tree model outputs the first fault risk value. The random forest model outputs a second failure risk value. .

[0063] The risk fusion module performs a weighted fusion of the first and second fault risk values ​​to obtain a comprehensive fault risk value: in, Indicates the first The overall fault risk value of each task awaiting emergency repair. This represents the first output of the gradient boosting tree model. The first fault risk value for each task awaiting emergency repair. The output of the random forest model represents the first... The second fault risk value for a task awaiting emergency repair. This represents the fusion weight, with values ​​between 0 and 1. The results from the two models are fused using a linear weighting method. This preserves the gradient boosting tree model's ability to capture complex, nonlinear risk changes while leveraging the random forest model to mitigate the volatility caused by anomalous data during a disaster. When the quality of real-time disaster data is high and risk changes are drastic, this weight can be appropriately increased. When data noise is high or there are many false alarms from sensors, the speed can be appropriately reduced. This makes the overall fault risk value more stable. This setting helps avoid misjudgment by a single model that could lead to the incorrect allocation of repair resources, thus improving the reliability of priority assessment results.

[0064] After obtaining the comprehensive fault risk value, the priority quantification module further combines the equipment importance, load impact value, and waiting time for emergency repair to calculate the first fault risk. Priority index of each pending repair task: in, Indicates the first The priority index of each task awaiting emergency repair. Indicates the first The load impact value corresponding to each emergency repair task. Indicates the first The importance of the equipment corresponding to each emergency repair task. Indicates the first The correction value for the waiting time for each pending repair task. These represent the weighting coefficients for the risk load impact item, equipment importance item, and waiting time for emergency repair item, respectively.

[0065] This is used to simultaneously express failure risk and load impact. If the failure risk is high and the load impact is large, this item will significantly improve task ranking; if the load impact value is extremely large, then... Smoothing can prevent a single overloaded task from suppressing other urgent tasks for an extended period, thereby maintaining the balance of the overall emergency repair task queue. This is used to reflect the importance of equipment in the power grid topology and power supply guarantee. For example, hub substations, important switching stations, and power supply nodes of hospitals or emergency command centers can be given higher importance. This is used to reflect the impact of waiting time on the urgency of the task, to prevent some medium-risk tasks from being delayed due to long waiting times, and to reduce the probability of secondary faults and the expansion of the impact of power outages on users.

[0066] In one alternative implementation, the waiting time correction value is... It can be dynamically determined based on the task waiting time: in, Indicates the first The correction value for the waiting time for each pending repair task. Indicates the first The waiting time from the generation of each pending repair task to the current scheduling time. This represents the waiting time growth factor. An exponential growth form is used because the longer the waiting time for disaster relief tasks, the higher the likelihood of fault spread, user complaints, secondary equipment damage, and increased recovery difficulty. Through exponential adjustment, tasks that have been unprocessed for a long time can be gradually moved up in the ranking, preventing low-initial-priority tasks from being ignored for extended periods.

[0067] The scheduling platform will prioritize indexes As the priority for corresponding emergency repair tasks, tasks in the emergency repair task list are sorted from highest to lowest priority to generate a dynamic emergency repair task sequence. The dynamic emergency repair task sequence can include information such as task number, equipment location, repair priority, comprehensive fault risk value, equipment importance, load impact value, waiting time, required materials, and estimated workload. The scheduling platform directly calls this sequence during the subsequent resource allocation phase, allowing high-risk, high-impact, high-importance tasks with long waiting times to be prioritized for emergency repair scheduling.

[0068] When data related to typhoon disaster relief and repair is updated, or when tasks are added, deleted, or merged in the list of tasks to be repaired, the dispatch platform re-extracts the updated meteorological impact information, equipment operation information, spatial location information, and load impact information, and recalculates the corresponding task feature vector, comprehensive fault risk value, and priority index. If the typhoon path deviates, local wind speeds significantly increase, road flooding worsens, or equipment operation status changes from abnormal to power outage, the repair priority of the relevant tasks can be increased accordingly. If a task has been repaired, a false alarm has been cleared, or the load has been restored through power transfer, that task can be removed from the dynamic repair task sequence or its ranking can be lowered.

[0069] S4. Based on the dynamic emergency repair task sequence, available emergency repair team information, emergency repair material information and traffic condition data, generate an initial scheduling scheme between emergency repair teams and tasks to be repaired, so as to match emergency repair resources with tasks to be repaired. In one specific implementation, step S4 can be implemented as follows: After obtaining the dynamic emergency repair task sequence, the scheduling platform reads the tasks to be repaired from high to low priority, and extracts the emergency repair priority, location, repair material information, and estimated repair workload for each task. The location of the task to be repaired is used to determine the accessibility between the starting point and the task point of the emergency repair team; the repair material information is used to determine whether the emergency repair team has the conditions to perform the task; and the estimated repair workload is used to measure the degree to which the emergency repair task occupies personnel, vehicles, and working time.

[0070] The dispatch platform simultaneously acquires information on available emergency repair teams. This information includes the team's current location, operational capabilities, availability status, carried supplies, current task occupancy, vehicle type, and personnel skill level. Team availability status can be determined based on whether the team is already on a task, on standby, located in a hazardous area, and vehicle availability. Team operational capabilities are determined based on team size, skill level, vehicle and equipment configuration, and historical average repair efficiency. By first determining team availability before task matching, duplicate assignments to teams already occupied or unsafe for deployment can be avoided.

[0071] Based on traffic data, the dispatch platform calculates the predicted travel time for each repair team to reach the location of each repair task. The predicted travel time can be determined based on the current road network topology, road congestion level, road flooding status, road blockage status, and road infrastructure damage status. For road sections with blockages, severe flooding, or damaged bridges, the travel time can be set to a maximum value or the section can be directly deemed impassable to avoid selecting routes that appear short but are actually inaccessible. The dispatch platform also estimates the estimated repair time required for each repair team to handle the repair tasks based on the expected workload and the operational capabilities of the repair teams.

[0072] In one alternative implementation, the first The emergency repair team handled the first The total response time required for each emergency repair task can be expressed as: in, Indicates the first The emergency repair team handled the first The total response time required for each emergency repair task. Indicates the first The emergency repair team will arrive at the [number]th location from its current position. Predicted passage time for each location awaiting emergency repair. Indicates the first The emergency repair team handled the first The estimated repair time for each pending repair task is calculated. Road travel time and on-site processing time are included in the response time because disaster relief efficiency depends not only on the speed of vehicle arrival but also on whether the team's capabilities match the type of fault. If only arrival time is considered, teams located nearby but lacking sufficient expertise may be prioritized, potentially extending the actual repair time; conversely, if only operational capabilities are considered, teams located further away may cause response delays. Evaluating the overall response time allows for task allocation to better align with the entire disaster relief process.

[0073] The dispatch platform uses repair teams as the allocation targets and pending repair tasks as the task targets, establishing allocation decision variables. When the first The emergency repair team was assigned to the first When there is an emergency repair task. The value is 1; when the first The emergency repair team was not assigned to the first When there is an emergency repair task. The value is set to 0. The dispatching platform generates an initial dispatching evaluation value based on the predicted travel time, estimated repair time, repair priority, and disaster safety risks: in, This represents the initial scheduling evaluation value. This indicates the number of available repair teams. This indicates the number of tasks requiring emergency repair. Indicates the first The emergency repair team and the first The allocation decision variables among the pending repair tasks, when the... The emergency repair team was assigned to the first The value is 1 when there is an emergency repair task pending; otherwise, the value is 0. Indicates the first The emergency repair team handled the first The comprehensive response time required for each emergency repair task is determined by the predicted traffic time and the estimated repair time. Indicates the first The emergency repair team carried out the first The corresponding disaster safety risks when an emergency repair task is pending. Indicates the first The priority of each pending repair task. , and These represent the weighting coefficients corresponding to the overall response time, disaster safety risk, and repair priority, respectively. Used to constrain emergency repair response efficiency; the longer the overall response time, the higher the dispatch evaluation value. To constrain the safety of operations during disasters, if the repair team needs to cross areas with strong winds, deep water accumulation, damaged roads, or traffic control zones, the safety risk during the disaster increases, and the dispatch evaluation value increases accordingly. This method reflects the principle that high-priority tasks should be assigned first, with higher priority repair tasks having a more significant impact on reducing the evaluation value. By incorporating time cost, risk cost, and task benefits into the calculation, the method avoids scheduling results that only pursue the nearest team or the shortest time, while ignoring personnel safety and critical load recovery needs in high-risk environments.

[0074] Safety risks during disasters The route can be determined based on a comprehensive assessment of factors such as wind speed, water depth, probability of road blockage, extent of road infrastructure damage, and traffic control measures in the areas traversed by the repair team. For example, when the route passes through areas heavily affected by strong winds, sections of road with excessive water accumulation, or sections of road with damaged bridges, The value increases; when the path mainly passes through unobstructed roads and is far from high-risk weather areas. The value is relatively small. This setting allows the dispatch platform to balance repair speed and team safety, reducing the likelihood of repair vehicles being forced to stop after entering high-risk roads.

[0075] During the optimization process, the scheduling platform aims to minimize the initial scheduling evaluation value. To determine the allocation relationship between repair teams and pending repair tasks, the solution must satisfy three constraints: task allocation, repair team operational capacity, and repair material matching. The task allocation constraint prevents multiple teams from accepting the same repair task; in resource-constrained scenarios, low-priority tasks can be temporarily deferred, while high-priority tasks are given priority in team allocation. The repair team operational capacity constraint prevents the same team from being assigned more tasks or workloads than its capacity. The repair material matching constraint ensures that the repair teams have sufficient materials to meet the task requirements, or that they can resupply from the nearest material storage point before executing the task.

[0076] In one alternative implementation, the constraint relationship can be expressed as: in, Indicates the first The emergency repair team and the first Allocation decision variables among the tasks awaiting emergency repair. This indicates the number of available repair teams. This indicates the number of tasks requiring emergency repair. Indicates the first The estimated workload for each pending repair task. Indicates the first The operational capabilities of the emergency repair teams within the current dispatch cycle; This indicates the number of available repair teams. This indicates the number of tasks requiring emergency repair. This indicates that a set of tasks must be processed; The first constraint states that the same emergency repair task can be assigned to at most one available repair team, reducing duplicate assignments and resource waste. The second constraint states that a set of tasks must be processed. Each task awaiting repair needs to be assigned to an available repair team to ensure that high-priority tasks, power supply tasks for critical users, or tasks with a high risk of fault propagation are not overlooked due to objective function optimization. The task set must be processed. The priority threshold can be determined based on the repair priority threshold, equipment importance threshold, load impact threshold, or the type of user experiencing a power outage. For example, tasks involving critical loads such as hospitals, emergency command centers, and transportation hubs, or tasks with a priority index exceeding a preset threshold, can be included in the set of tasks that must be processed. The third constraint states that the total workload of tasks assigned to the same repair team cannot exceed the team's operational capacity, preventing the scheduling plan from failing during the execution phase due to excessive team load.

[0077] Emergency repair supplies can be matched by comparing the task's supply requirements list with the team's existing supplies list. If the supplies carried by the repair team can cover the needs of the task to be repaired, the team can be directly assigned to the corresponding task. If there is a shortage, the dispatch platform further searches for reachable supply storage points to determine whether the repair team can complete the resupply within the expected time. For combinations that cannot directly meet the supply requirements and cannot be resolved through nearby resupply, the dispatch platform marks the combination as infeasible and does not participate in the generation of candidate dispatch solutions. Through supply matching constraints, it is possible to avoid teams being unable to carry out repairs after arriving on site due to insufficient spare parts, tools, or emergency equipment.

[0078] The scheduling platform can use integer programming, heuristic search, genetic algorithms, or non-dominated sorting genetic algorithms to solve the allocation relationships. For areas with a small number of tasks and a limited number of teams, integer programming can directly obtain a better allocation result. For scenarios with multiple concurrent failures and rapidly changing task numbers during typhoon disasters, heuristic search or genetic algorithms can be used to quickly generate candidate scheduling schemes. Candidate scheduling schemes can include information such as the task list corresponding to each repair team, the task execution order, the estimated departure time, the estimated arrival time, the estimated repair completion time, the source of materials, and the alternative teams.

[0079] After candidate dispatch schemes are generated, the dispatch platform selects an initial dispatch scheme based on the current dispatch objectives during the disaster. These objectives can be dynamically adjusted according to the typhoon stage and power supply requirements. For example, during the peak of the typhoon, the safety risk weight can be increased, prioritizing lower-risk repair combinations; when the typhoon's impact weakens and roads are largely restored, the repair priority weight can be increased to accelerate the restoration of critical loads; when power outages significantly affect important users, schemes that can restore critical loads such as hospitals, emergency command centers, and transportation hubs most quickly can be prioritized. Through dynamic adjustments to weights and dispatch objectives, the initial dispatch scheme can adapt to the repair needs at different stages of a typhoon disaster.

[0080] The final initial scheduling plan includes the correspondence between tasks to be repaired and repair teams, the execution order of repair teams, the source of required repair materials, and the estimated arrival time. This plan provides both the starting point and destination of the route planning for subsequent route planning and the supply basis for the allocation of repair materials. By simultaneously considering task priority, team capabilities, material matching, travel time, and disaster safety risks in the initial scheduling stage, secondary adjustments caused by task mismatch, insufficient materials, or excessively high road risks after the repair teams depart can be reduced, thereby improving the overall response efficiency of repair resources in multi-point concurrent fault scenarios.

[0081] S5. Based on the initial dispatch plan and traffic data, generate an initial repair route and a backup repair route for the repair team to the location of the task to be repaired. In one specific implementation, step S5 can be implemented as follows: After obtaining the initial scheduling plan, the scheduling platform reads the correspondence between repair teams and tasks to be repaired in the initial scheduling plan one by one, and determines the location of the task to be repaired corresponding to each repair team. For the first... The dispatch platform assigns a repair team, using the team's current location as the starting point and the location of the assigned repair task as the ending point. The team's current location can be determined by vehicle positioning terminals, dispatch platform check-in information, or team deployment information; the repair task location can be determined by the spatial location of the equipment to be repaired, fault location results, or coordinates reported on-site. By determining the starting and ending points beforehand, the route planning remains consistent with the previous resource scheduling results, preventing the route planning from deviating from the actual work assignment plan.

[0082] The dispatch platform extracts road network topology data, road congestion data, road flooding data, road blockage data, and road facility damage data from traffic condition data, and constructs a dynamic road network corresponding to the current moment. The dynamic road network can be represented as... ,in, Indicates the current time Dynamic road network Represents the set of road network nodes. Indicates the current time A set of road segments that can participate in route planning. Road network nodes can include road intersections, ramp entrances, bridge endpoints, tunnel endpoints, and turning points for emergency repair vehicles; road segments can include the road connections between adjacent road network nodes. Unlike ordinary static road networks, the status of road segments in dynamic road networks changes in real time with disaster information such as traffic congestion, road flooding, traffic control, fallen trees, and bridge damage, making them more suitable for route planning of emergency repair vehicles during typhoon disasters.

[0083] Before determining the cost of passage for each road segment, the dispatch platform first calculates the basic passage time based on the road segment length and normal traffic speed: in, Represents road network nodes With road network nodes Basic travel time between the road sections Represents road network nodes With road network nodes The length of the road segment between them This indicates the basic traffic speed of this road section under normal traffic conditions.

[0084] After obtaining the basic travel time, the dispatch platform calculates the travel cost of the road segment based on road congestion data, road waterlogging data, road blockage data, and road facility damage data: in, Indicates the current time Lower road network nodes With road network nodes The value of passage between the two sections of road Represents road network nodes With road network nodes Basic travel time between the road sections Indicates the current time The degree of road congestion in this section of the road. Indicates the current time The impact value of road water accumulation on this section of the road. Indicates the current time The status value of whether there is road blockage or damage to road facilities on this section of the road is as follows. This indicates the adjustment coefficient corresponding to the degree of road congestion. This represents the adjustment coefficient corresponding to the impact value of road water accumulation. This indicates the penalty coefficient corresponding to road blockage or damage to road facilities; This term is used to express the amplifying effect of congestion on basic travel time. The more congested the road, the more likely the actual travel time of vehicles will increase proportionally; therefore, the congestion term is used as a multiplicative correction to the basic travel time. This is used to describe the additional impact of water accumulation on the passage of emergency repair vehicles. Water accumulation not only reduces vehicle speed but may also increase the risk of vehicles wading through water and taking detours; therefore, water accumulation is treated as an independent penalty factor. Used to describe the impassable impact caused by road blockage or damage to road facilities. This applies when the road section is completely blocked by road closures, damaged bridges, landslides, or fallen trees. The value is 1; when there is no road blockage or damage to road facilities on this section of the road, The value is 0. Penalty coefficient. This cost function can be set to a value much higher than the cost of passing through ordinary road sections, allowing the path search algorithm to automatically avoid physically blocked road sections. Through this cost function, path planning considers not only speed but also passability and safety, reducing the likelihood of repair vehicles being directed to inaccessible roads during disasters.

[0085] Impact value of road water accumulation The determination can be made based on the depth and length of the floodwaters, as well as the wading capability of the repair vehicles. For ordinary repair vehicles, if the water depth is close to or exceeds the vehicle's safe wading threshold, [the vehicle can be...]. Set to a larger value; for vehicles with strong wading capabilities, the impact value at the same water depth can be appropriately reduced. This approach allows the same road section to have different passage significance for different repair vehicles, avoiding the blanket exclusion of all flooded roads and preventing high-risk flooded roads from being mistakenly selected as the preferred route.

[0086] After constructing a dynamic road network and calculating the toll value of each road segment, the scheduling platform searches for a set of candidate emergency repair routes from the starting point to the ending point of the route within the dynamic road network based on the toll value of each road segment. Candidate emergency repair routes can be obtained using Dijkstra's algorithm, A* algorithm, K-shortest path algorithm, or incremental path search algorithms suitable for dynamic road networks. For typhoon disaster scenarios, multiple candidate routes can be generated preferentially rather than just one, because a single route is prone to failure when road conditions change rapidly. Multiple route candidates provide a foundation for subsequent backup route selection and dynamic replanning.

[0087] For any path in the candidate emergency repair path set The scheduling platform can calculate the total cost of passage for this route: in, Representing a path At the present moment The total value of the path agency Representing a path Central network nodes With road network nodes The section of road between, Indicates the current time The cost of passage on the lower section of the road; The dispatching platform selects the initial repair route from the candidate repair route set, choosing the route with the lowest total cost of passage that also meets the safety conditions for the repair team's passage. These safety conditions include: no confirmed road blockages along the route; water depth in critical sections not exceeding the vehicle wading threshold; wind speeds in the route's traversal area not exceeding the safe driving threshold for repair vehicles; road infrastructure damage risk not exceeding a preset level; and the route meeting the required width and turning radius for repair vehicles. This configuration avoids selecting roads with serious safety hazards solely based on low cost of passage.

[0088] After determining the initial repair route, the dispatch platform continues to select a backup repair route from the set of candidate repair routes. The backup repair route should not highly overlap with the initial repair route; otherwise, if a critical section of the initial repair route is blocked, the backup route may also become unusable. Therefore, the dispatch platform can calculate the proportion of overlapping sections between the candidate routes and the initial repair route: in, Indicate candidate path Compared with the initial repair path The proportion of overlapping road sections between them Indicate candidate path The set of included road segments Indicates the initial repair path The set of included road segments This indicates the number of road segments shared by both paths. This indicates the number of road segments included in the initial repair route.

[0089] In one specific selection method, the dispatching platform filters paths from the candidate repair path set whose overlapping section ratio is lower than a preset ratio. Then, it selects paths with lower total toll collection value and that meet the backup conditions as backup repair paths. Preset backup conditions may include: the backup path's total toll collection value not exceeding a preset multiple of the initial repair path's total toll collection value; the backup path not having any confirmed blocked sections; the backup path passing through fewer than a preset number of high-risk flooded sections; and the backup path being able to connect to the repair team near its current location or at a node midway along the initial path. By simultaneously controlling the path overlap ratio and total toll collection value, it avoids backup paths that are independent but take excessively long detours, and also avoids backup paths that are shorter but highly overlapped with the initial path.

[0090] For urban areas with dense road networks, the dispatch platform can generate an initial repair route and at least one backup repair route for a repair team. For mountainous areas, coastal areas, or areas with numerous bridges and tunnels and sparse road networks, if the number of available roads is limited, the backup repair route can be expanded to a backup access point plus a partial detour. When the initial repair route is partially blocked, the repair vehicle can first travel to the nearest accessible node and then bypass the blocked section via the partial detour. This method can improve the flexibility of route dispatching in areas with poor road conditions.

[0091] The dispatching platform associates the initial and backup repair routes with the corresponding repair teams and tasks to be repaired, forming repair route data. This data includes route number, team number, task number, route start point, route end point, road segments, critical turning points, estimated travel time, total toll value, backup route access node, route update time, and abnormal road segment markers. After the repair route data is written into the initial dispatch plan, the dispatching platform can first distribute the initial repair route to the corresponding repair teams and retain the backup repair route for dynamic adjustments during the journey.

[0092] This implementation method eliminates the static shortest distance as the sole criterion for route planning. Instead, it transforms road congestion, flooding, blockages, and infrastructure damage during typhoons into calculable travel costs. An initial repair route ensures repair teams can quickly reach their assigned locations, while a backup route addresses unforeseen road conditions en route. Using these two types of routes in combination reduces the probability of repair vehicles being blocked en route, improving the continuity and safety of repair operations during typhoons.

[0093] S6. During the march of the emergency repair team, traffic data is updated in real time. When traffic congestion or damage to road facilities is detected on the initial repair route, the repair route is dynamically adjusted based on the backup repair route and the updated traffic data. The adjusted repair route is then sent to the corresponding emergency repair team.

[0094] In one specific implementation, step S6 can be implemented as follows: During the repair team's journey along the initial repair route, the dispatch platform continuously receives the repair team's current location, road condition information reported by vehicle terminals, and real-time updated traffic data. The repair team's current location can be reported by the vehicle-mounted positioning terminal at preset time intervals. The road condition information reported by the vehicle terminals can include on-site information such as road congestion ahead, road flooding, fallen trees, road closures, bridge abnormalities, and vehicles being unable to pass. The updated traffic data can come from traffic management platforms, navigation platforms, road monitoring and identification results, on-site inspection information, and disaster reporting information. The dispatch platform matches the updated traffic data with road segments in the initial repair route and backup repair routes to determine whether newly emerging abnormal road conditions affect the route currently being followed by the repair team.

[0095] During the road segment matching process, the dispatch platform can establish a correspondence based on road network node numbers, road names, spatial coordinate ranges, and road segment coverage areas. For point-like or linear disaster information such as road flooding, road blockages, or road facility damage, the disaster location can first be mapped to the nearest road network node or road segment, and then it can be determined whether the road segment belongs to the initial repair route or the backup repair route. For information reported from repair vehicle terminals, abnormal road segments can be identified by combining the vehicle's current location, driving direction, and the road topology ahead. This matching method can prevent disaster information from remaining only at the text or coordinate level and failing to be promptly applied to the ongoing repair routes.

[0096] Based on updated traffic data, the dispatch platform recalculates the traffic status and cost of passage for each segment of the initial repair route. If a segment changes from normal traffic to congestion, the congestion impact is increased; if water accumulation deepens on a segment, the water accumulation impact is increased; if a segment is completely blocked by road closures, bridge damage, slope collapses, or fallen trees, it is marked as an abnormal segment, and its traffic cost is set to maximum or impassable. Thus, the route status is updated synchronously with changes in the road environment during the disaster.

[0097] The dispatch platform can record the total value of the path when the initial emergency repair path is issued. And the total value of the path at the current moment. A comparison is made. When the initial repair route contains abnormal sections such as traffic congestion, road blockage, excessive road flooding, or damaged road facilities, the dispatch platform directly determines that the initial repair route does not meet the current traffic conditions. Even when there are no completely blocked sections, if the total cost of the route has significantly increased compared to when it was initially assigned, it can still be determined that the initial repair route does not meet the current traffic conditions. The judgment criteria can be expressed as follows: in, Indicates the initial repair path At the present moment The total value of the path agency Indicates the initial repair path At the time of issuance The total value of the path agency This indicates a preset adjustment threshold. When the increase in toll costs exceeds this threshold, even if the road is not completely blocked, it indicates that continuing along the original route may lead to significant delays or increased safety risks, thus requiring a route adjustment. This method avoids only rerouting when the road is completely impassable, allowing repair teams to avoid risky sections earlier.

[0098] After determining that the initial repair route does not meet the current traffic conditions, the dispatch platform uses the repair team's current location as the new route starting point and the location of the corresponding repair task as the route ending point. It then combines this with backup repair routes and updated traffic data to determine the adjusted repair route. The reason for re-determining the route starting point is that the repair team is already in motion, and the original station or departure location is no longer suitable as the route planning starting point. Using the real-time location as the new route starting point ensures that the adjusted repair route directly serves the current vehicle status, avoiding the generation of unexecutable routes.

[0099] When the backup repair route meets the current traffic conditions, the dispatch platform first determines the access node for the repair team's current location to connect to the backup repair route. The access node can be the backup route node closest to the repair team's current location and reachable safely via the current dynamic road network, or it can be a road intersection connecting to the backup repair route after bypassing the abnormal road section. The dispatch platform calculates the path connection segment from the repair team's current location to the access node, then extracts the remaining backup path segment from the access node to the location of the repair task, and concatenates the path connection segment with the remaining backup path segment to obtain the adjusted repair route. This processing method fully utilizes the backup repair route already generated in the previous step, reduces global rerouting time, and enables the repair team to quickly obtain an executable detour plan in case of sudden road conditions.

[0100] When alternative repair routes also fail to meet current traffic conditions, the dispatch platform re-searches the local road network affected by the abnormal road segment based on updated traffic data. The local road network can be defined by the current location of the repair team, the location of the task to be repaired, and road network nodes within a certain range around the abnormal road segment; alternatively, it can be defined by the current vehicle reachability and candidate roads in the task direction. During the re-search, the dispatch platform retains the travel cost of unaffected road segments and only updates the cost of the abnormal road segment and its adjacent road segments, thus avoiding repeated global calculations of the entire city road network. Through this local road network re-search, alternative repair routes from the current location of the repair team to the location of the task to be repaired can be obtained while maintaining computational efficiency.

[0101] In one alternative implementation, local re-search can be achieved using an incremental path planning algorithm. This type of algorithm updates the cost of affected road segments based on the initial path, recalculating only the node costs related to the abnormal road segments, without completely discarding already calculated path information. During typhoons, road conditions change frequently. If the global path is recalculated with each traffic data update, it will place excessive computational pressure on the dispatch platform and easily delay path distribution. Using a local update approach can improve the speed of re-planning, allowing repair vehicles to receive adjusted routes in time before approaching abnormal road segments.

[0102] After generating alternative repair routes, the dispatch platform will confirm these alternative routes as the adjusted repair routes. If multiple alternative repair routes simultaneously meet the passability requirements, the dispatch platform can prioritize routes with lower overall route value, fewer abnormal road sections, lower risk of road facility damage, and greater distance from the original abnormal road sections. This avoids alternative routes passing through the same disaster area again, improving the stability of the adjusted routes.

[0103] After determining the adjusted repair routes, the dispatch platform associates these routes with the corresponding repair teams and tasks, and updates the repair route data in the initial dispatch plan. The updates may include the new route origin, access nodes, route segments, estimated arrival time, total route cost, abnormal road segment markers, route update time, and reason for detour. For abnormal road segments deemed impassable, the dispatch platform can also synchronously write the abnormal status into traffic condition data for subsequent route planning by other repair teams, preventing different teams from repeatedly selecting the same blocked road segment.

[0104] After the adjusted repair route is generated, the dispatch platform sends the route to the corresponding repair team's vehicle-mounted or mobile terminal. The sent information may include detour instructions, key turning points, changes in estimated arrival time, locations of abnormal road sections, and precautions. If communication conditions are unstable, the dispatch platform can simultaneously send simplified route instructions and a list of key nodes to ensure the repair team can complete the detour according to the key nodes even in weak communication environments. After receiving the information, the repair team's vehicle terminal returns a confirmation message, and the dispatch platform continues to track the vehicle's current location and road conditions until the repair team reaches the location of the repair task.

[0105] Through the above processing, emergency repair route planning is transformed from a one-time pre-departure planning to a real-time closed-loop adjustment during the journey. This method can promptly identify the risk of initial emergency repair route failure when road conditions change rapidly during a typhoon, and generate executable alternative routes using backup repair routes or local re-search. Compared to relying solely on static shortest paths, this implementation method can reduce the occurrence of emergency repair vehicles being blocked en route, long waiting times for manual reassignment, or repeated detours, improving the continuity, safety, and timeliness of emergency repair teams arriving at the scene.

[0106] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

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

Claims

1. A method for intelligent scheduling of emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning, characterized in that, Includes the following steps: S1. Real-time acquisition of typhoon disaster emergency repair data, and preprocessing of the typhoon disaster emergency repair data to generate comprehensive disaster dispatch data; wherein, the typhoon disaster emergency repair data includes typhoon meteorological data, power grid equipment data, geographical environment data and traffic condition data; S2. Based on the disaster-related comprehensive dispatch data, identify high-risk power grid equipment within the typhoon's impact range, and generate a list of tasks to be repaired based on the operating status of the high-risk power grid equipment. S3. Input the meteorological impact information, equipment operation information, spatial location information and load impact information corresponding to the list of tasks to be repaired into the fault priority assessment model to obtain the repair priority of each task to be repaired, and generate a dynamic repair task sequence based on the repair priority. S4. Based on the dynamic emergency repair task sequence, available emergency repair team information, emergency repair material information and traffic condition data, generate an initial scheduling scheme between the emergency repair team and the emergency repair task, so as to match the emergency repair resources with the emergency repair task. S5. Based on the initial scheduling plan and the traffic data, generate an initial repair route and a backup repair route for the repair team to the location of the repair task. S6. During the march of the emergency repair team, the traffic data is updated in real time. When traffic congestion or damage to road facilities is detected on the initial emergency repair route, the emergency repair route is dynamically adjusted based on the backup emergency repair route and the updated traffic data, and the adjusted emergency repair route is sent to the corresponding emergency repair team.

2. The intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning as described in claim 1, characterized in that, The typhoon meteorological data includes typhoon path data, typhoon intensity data, wind speed data, wind direction data, precipitation data, and typhoon impact range data; The power grid equipment data includes equipment ledger data, equipment spatial location data, equipment operating status data, fault alarm data, equipment importance data, and load impact data; The geographic environment data includes topographic data, elevation data, digital elevation data, and micro-topographic data; The traffic data includes road network topology data, road congestion data, road waterlogging data, road blockage data, and road facility damage data.

3. The intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning as described in claim 1, characterized in that, The preprocessing specifically includes: Anomaly screening was performed on the emergency repair data during the typhoon disaster, and abnormal data that exceeded the preset reasonable range was removed; Missing data in the emergency repair data related to the typhoon disaster were supplemented, and the data formats from different sources were standardized. The typhoon meteorological data, power grid equipment data, geographical environment data, and traffic condition data are time-aligned according to a unified time base. Spatial registration was performed on the typhoon meteorological data, power grid equipment data, geographical environment data, and traffic condition data according to unified spatial coordinates; Based on the data after time alignment and spatial registration, the disaster-related integrated dispatch data is generated by associating and fusing the data according to the spatial location of the equipment, the typhoon impact range, and the road network location.

4. The intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning as described in claim 1, characterized in that, The determination of high-risk power grid equipment within the typhoon's impact area based on the comprehensive disaster dispatch data specifically includes: Extract the current typhoon center location, typhoon impact range, and spatial location of each power grid device from the disaster comprehensive dispatch data; The typhoon center location and the equipment spatial location are uniformly converted into latitude and longitude coordinates, and then converted into radians when performing trigonometric function calculations to obtain the first... Equipment space location of individual power grid devices and the current moment Typhoon center location ; Based on the spatial location of the equipment and the location of the typhoon center, calculate the first... The spatial distance of each power grid device relative to the center of the typhoon ,in: in, Indicates the first At the current moment, each power grid device Relative to the spatial distance from the typhoon center, Represents the Earth's average radius. and They represent the first The longitude and latitude of each power grid device, and They represent the current time. The longitude and latitude of the typhoon center; Extract the current time from the typhoon impact range data. typhoon impact radius When the spatial distance Less than or equal to the typhoon's radius of influence At that time, the corresponding power grid equipment will be identified as candidate power grid equipment within the typhoon's impact range; Extract typhoon meteorological data, geographical environment data and power grid equipment data corresponding to the candidate power grid equipment from the disaster-in-disaster integrated dispatch data, and determine the meteorological exposure status, geographical exposure status and equipment operation status of the candidate power grid equipment; When the candidate power grid equipment has a fault alarm, abnormal operating status, or when the meteorological exposure status and geographical exposure status of the candidate power grid equipment reach the preset risk conditions, the corresponding candidate power grid equipment will be identified as the high-risk power grid equipment.

5. The intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning as described in claim 1, characterized in that, The process of generating a list of tasks to be repaired based on the operating status of the high-risk power grid equipment specifically includes: The operating status of the high-risk power grid equipment is obtained, including fault alarm status, abnormal equipment operating parameter status, power outage status, and load loss status. Based on the operating status, the high-risk power grid equipment is identified in the task status judgment, and the high-risk power grid equipment that is in fault alarm status, abnormal equipment operating parameter status, or power outage status is identified as the objects to be repaired. Based on the equipment type, spatial location, fault alarm status, abnormal status of equipment operating parameters, power outage status, and load loss status of the object to be repaired, a corresponding repair task is generated. Based on the equipment type and fault alarm status of the object to be repaired, determine the emergency repair materials required for the emergency repair task and the estimated emergency repair workload; The meteorological impact information, equipment operation information, spatial location information, load impact information, emergency repair material information, and estimated emergency repair workload corresponding to the task to be repaired are written into the task data item. The task data items generated repeatedly for the same object to be repaired within adjacent time windows are merged to form the list of tasks to be repaired.

6. The intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning as described in claim 1, characterized in that, The fault priority assessment model is constructed based on an ensemble learning model and a multi-attribute decision model. The fault priority assessment model includes a feature extraction module, a fault risk prediction module, a risk fusion module, and a priority quantification module. The input to the fault priority assessment model is the meteorological impact information, equipment operation information, spatial location information and load impact information corresponding to the list of tasks to be repaired, and the output is the repair priority of each task to be repaired. The feature extraction module is used to extract a task feature vector for each task to be repaired. The task feature vector includes meteorological impact features formed by the meteorological impact information, equipment operation features formed by the equipment operation information, spatial exposure features formed by the spatial location information, and load impact features formed by the load impact information. The fault risk prediction module is used to input the task feature vector into the gradient boosting tree model and the random forest model respectively to obtain the first fault risk value and the second fault risk value of the corresponding task to be repaired. The risk fusion module is used to perform weighted fusion of the first fault risk value and the second fault risk value to obtain a comprehensive fault risk value for the corresponding emergency repair task, wherein: in, Indicates the first The overall fault risk value of each task awaiting emergency repair. This represents the first output of the gradient boosting tree model. The first fault risk value for each task awaiting emergency repair. The output of the random forest model represents the first... The second fault risk value for a task awaiting emergency repair. Indicates the fusion weight, and The value of is between 0 and 1; The priority quantification module is used to obtain the repair priority of the corresponding repair task based on the comprehensive fault risk value, equipment importance, load impact information and waiting time for emergency repair.

7. A method for intelligent scheduling of emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning, as described in claim 6, is characterized in that... S3 specifically includes: Read the tasks to be repaired one by one from the list of tasks to be repaired, and extract the meteorological impact information, equipment operation information, spatial location information and load impact information of the corresponding tasks to be repaired from the disaster comprehensive scheduling data; Based on the meteorological impact information, equipment operation information, spatial location information, and load impact information, a task feature vector corresponding to the task to be repaired is constructed, and the task feature vector is input into the fault priority assessment model to obtain the comprehensive fault risk value of the corresponding task to be repaired. Based on the comprehensive fault risk value, equipment importance, load impact value, and waiting time for emergency repair, the data is input into the priority quantification module to calculate the priority index of the corresponding emergency repair task, where: in, Indicates the first The priority index of each task awaiting emergency repair. Indicates the first The overall fault risk value of each task awaiting emergency repair. Indicates the first The load impact value corresponding to each emergency repair task. Indicates the first The importance of the equipment corresponding to each emergency repair task. Indicates the first The correction value for the waiting time for each pending repair task. , and These represent the weighting coefficients corresponding to the comprehensive failure risk value, equipment importance, and repair wait time correction value, respectively. The priority index is used as the repair priority of the corresponding emergency repair task, and the emergency repair tasks in the emergency repair task list are sorted from high to low according to the repair priority to generate the dynamic emergency repair task sequence. When the data related to emergency repairs during the typhoon disaster is updated, or when tasks to be repaired are added, deleted, or merged in the list of tasks to be repaired, the updated meteorological impact information, equipment operation information, spatial location information, and load impact information are re-extracted, and the repair priority of each task to be repaired and the dynamic repair task sequence are updated.

8. The intelligent scheduling method for emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning as described in claim 1, characterized in that, S4 specifically includes: Read the tasks to be repaired from the dynamic emergency repair task sequence, and obtain the repair priority, location, repair materials information and estimated repair workload of each task to be repaired; Obtain the information on available emergency repair teams, which includes the current location of the emergency repair teams, their operational capabilities, their availability status, and information on the materials they are carrying. Based on the traffic data, the predicted travel time for each repair team to reach the location of each repair task is determined, and based on the estimated repair workload and the operational capacity of the repair teams, the estimated repair time for each repair team to handle each repair task is determined. Using the repair teams as the allocation objects and the tasks to be repaired as the task objects, allocation decision variables between the repair teams and the tasks to be repaired are established. Initial scheduling evaluation values ​​are generated based on the predicted travel time, the estimated repair time, the repair priority, and the disaster safety risks, wherein: in, This represents the initial scheduling evaluation value. This indicates the number of available repair teams. This indicates the number of tasks requiring emergency repair. Indicates the first The emergency repair team and the first The allocation decision variables among the pending repair tasks, when the... The emergency repair team was assigned to the first The value is 1 when there is an emergency repair task pending; otherwise, the value is 0. Indicates the first The emergency repair team handled the first The comprehensive response time required for each pending repair task is determined by the predicted traffic time and the estimated repair time. Indicates the first The emergency repair team carried out the first The corresponding disaster safety risks when an emergency repair task is pending. Indicates the first The priority of each pending repair task. , and These represent the weighting coefficients corresponding to the overall response time, disaster safety risk, and repair priority, respectively. With the goal of minimizing the initial scheduling evaluation value, and under the conditions of satisfying the task allocation constraints, the operational capacity constraints of the emergency repair team, and the matching constraints of emergency repair materials, the allocation relationship between the emergency repair team and the emergency repair task is optimized to obtain the candidate scheduling scheme between the emergency repair team and the emergency repair task. The task allocation constraints include that the same emergency repair task can be assigned to at most one available emergency repair team; the emergency repair team's operational capacity constraints include that the expected emergency repair workload corresponding to the emergency repair task assigned to the same emergency repair team does not exceed the operational capacity of that emergency repair team; and the emergency repair material matching constraints include that the information on the materials already carried by the emergency repair team matches the information on the emergency repair materials required for the emergency repair task, or that insufficient materials are replenished through the material sources determined by the emergency repair material information. The initial scheduling scheme is selected from the candidate scheduling schemes to meet the current disaster scheduling objectives. The initial scheduling scheme includes the correspondence between the tasks to be repaired and the repair teams, the execution order of the repair teams, the source of the required repair materials, and the estimated arrival time.

9. A method for intelligent scheduling of emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning, as described in claim 1, is characterized in that... S5 specifically includes: According to the initial scheduling scheme, the location of the repair task corresponding to each repair team is determined, and the current location of the repair team is taken as the starting point of the path, and the location of the corresponding repair task is taken as the ending point of the path. From the traffic condition data, road network topology data, road congestion data, road water accumulation data, road blockage data, and road facility damage data are extracted to construct a dynamic road network corresponding to the current time. Based on the traffic status of each road segment in the dynamic road network, the traffic cost of each road segment is determined, wherein: in, Indicates the current time Lower road network nodes With road network nodes The value of passage between the two sections of road Represents road network nodes With road network nodes Basic travel time between the road sections Indicates the current time The degree of road congestion in this section of the road. Indicates the current time The impact value of road water accumulation on this section of the road. Indicates the current time The status value of whether there is road blockage or damage to road facilities on this section of the road is as follows. This indicates the adjustment coefficient corresponding to the degree of road congestion. This represents the adjustment coefficient corresponding to the impact value of road water accumulation. The penalty coefficient represents the amount of time required to prevent road blockage or damage to road infrastructure; the basic travel time... , Represents road network nodes With road network nodes The length of the road segment between them Represents road network nodes With road network nodes The basic traffic speed of the road section under normal traffic conditions; Based on the toll value of each road segment, a set of candidate emergency repair routes from the starting point of the route to the ending point of the route is searched in the dynamic road network, and the sum of the toll values ​​of each road segment in each candidate emergency repair route is taken as the total toll value of the corresponding candidate emergency repair route. The path with the lowest total value of passage and which meets the conditions for safe passage of the repair team is selected from the set of candidate repair paths and is taken as the initial repair path. From the candidate emergency repair route set, select the route whose overlap with the initial emergency repair route is lower than a preset ratio and whose total passage value meets the preset backup conditions, and use it as the backup emergency repair route. The initial repair path and the backup repair path are associated with the corresponding repair teams and the corresponding repair tasks to be repaired, forming repair path data, and the repair path data is written into the initial scheduling scheme.

10. A method for intelligent scheduling of emergency repair resources during typhoon disasters based on real-time prediction and dynamic path planning, as described in claim 1, is characterized in that... S6 specifically includes: During the march of the repair team, the current location of the repair team, the road condition information reported by the vehicle terminal, and the updated traffic condition data are obtained in real time, and the updated traffic condition data are matched with the road segments in the initial repair route and the backup repair route. Based on the updated traffic data, the traffic status and cost of each segment in the initial repair route are updated, and the total cost of the initial repair route at the current time is calculated, where: in, Representing a path At the present moment The total value of the path agency Representing a path Central network nodes With road network nodes The section of road between, Indicates the current time The toll value of the following road sections; If there are abnormal road sections in the initial emergency repair route that are congested, blocked, flooded, or damaged, or if the total route value of the initial emergency repair route at the current time increases by more than a preset adjustment threshold compared to the total route value when the initial emergency repair route was issued, it is determined that the initial emergency repair route does not meet the current traffic conditions. After determining that the initial repair route does not meet the current traffic conditions, the current location of the repair team is used as the new route starting point, and the location of the corresponding repair task is used as the route ending point. The adjusted repair route is determined based on the backup repair route and the updated traffic data. When the backup repair route meets the current traffic conditions, the access node of the backup repair route is determined at the current location of the repair team, and the path connecting segment from the current location of the repair team to the access node and the remaining segment of the backup route from the access node to the location of the task to be repaired are spliced ​​together to obtain the adjusted repair route. When the backup repair route does not meet the current traffic conditions, the local road network affected by the abnormal road section is re-searched based on the updated traffic data to generate an alternative repair route from the current location of the repair team to the location of the repair task, and the alternative repair route is determined as the adjusted repair route. The adjusted repair path is associated with the corresponding repair team and the corresponding repair task, the repair path data in the initial scheduling scheme is updated, and the adjusted repair path is sent to the corresponding repair team.

Images