A shortest path planning method and system based on water management pipe network inspection

Abstract

The present application relates to the technical field of urban water management, and particularly relates to a shortest path planning method and system based on water network inspection, which comprises the following steps: acquiring road network data, network facility data, work order task data, traffic and construction data, SCADA monitoring data, historical inspection trajectory data and inspection personnel state data; for each to-be-inspected facility point, a virtual access edge between the to-be-inspected facility point and an access node is established, and a fusion graph model containing facility nodes, road nodes and virtual access edges is formed; based on the fusion graph model, the comprehensive cost of each edge is calculated in combination with network risk attributes, task business attributes, traffic states and monitoring alarm states; a plurality of to-be-inspected tasks are clustered and sorted to determine the task access sequence; and a heuristic search algorithm containing space distance estimation and business attribute estimation in the heuristic function is used to search and generate an inspection path section by section. The present application improves the inspection efficiency and path planning accuracy.

Description

Technical Field

[0001] This invention relates to the field of urban water management technology, and in particular to a shortest path planning method and system based on water pipeline network inspection. Background Technology

[0002] Routine inspections of urban water supply, drainage, and integrated water facilities require inspectors to sequentially reach water facilities such as valve wells, fire hydrants, and pressure monitoring points located in different areas within limited working hours to complete on-site inspections and data recording. Currently, the planning of inspection routes typically relies on Geographic Information System (GIS) platforms, work order management systems, and ordinary map navigation, employing shortest path algorithms with road distance or travel time as optimization objectives to calculate travel routes on road network maps.

[0003] However, water facilities are typically located in greenbelts, within residential areas, along roadsides, or within pumping station areas. Ordinary road navigation cannot directly reach these facilities, creating a disconnect between road accessibility and facility inspection availability. Furthermore, existing route planning does not incorporate water service attributes such as pipe aging, leakage alarms, task priority, time window constraints, and facility access difficulty into the path cost, failing to balance prioritizing high-risk facilities with task urgency. Additionally, current multi-point inspection sorting often relies on manual experience or simple nearest-neighbor sorting, failing to comprehensively consider network pressure zoning and task priority differences, leading to irrational cross-zoning travel and task sorting. Moreover, existing shortest path algorithms are mostly based on static road networks; when dynamic events such as road closures or new emergency tasks occur, the entire route needs to be recalculated, resulting in low response efficiency.

[0004] Therefore, there is an urgent need for a shortest path planning method and system based on water pipeline network inspection. Summary of the Invention

[0005] (a) Technical problems to be solved In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a shortest path planning method and system based on water pipeline network inspection, which solves the technical problems of incomplete connection between road accessibility and facility inspectionability, unreasonable task sequencing, and low route response efficiency under dynamic events in the existing inspection route planning.

[0006] (II) Technical Solution To achieve the above objectives, the main technical solutions adopted by the present invention include: In a first aspect, the present invention provides a shortest path planning method based on water pipeline network inspection, comprising: Acquire road network data, pipeline facility data, work order task data, traffic and construction data, SCADA monitoring data, historical inspection trajectory data, and inspection personnel status data. The road network data includes accessible roadside areas. For each facility point to be inspected, an access line segment is determined and an access node is generated along the accessible road. A virtual access edge is established between the facility point to be inspected and the access node, forming a fusion graph model that includes facility nodes, road nodes and the virtual access edge. Based on the fusion graph model, and combining the pipeline risk attributes of the pipeline facility data, the task business attributes of the work order data, the traffic status of the traffic and construction data, and the monitoring alarm status of the SCADA monitoring data, the comprehensive cost of each edge in the fusion graph model is calculated. Based on the work order task data and the comprehensive cost, multiple tasks to be inspected are clustered and sorted to determine the task access order. Based on the fusion graph model, the comprehensive cost, and the task access order, a heuristic search algorithm that includes spatial distance estimation and business attribute estimation in the heuristic function is used to generate inspection paths segment by segment.

[0007] Optionally, the method further includes: During the inspection process, dynamic events are detected based on the traffic and construction data, SCADA monitoring data, and inspection personnel status data. When the dynamic event meets the preset triggering conditions, the affected edges and affected nodes in the fusion graph model are determined, the completed tasks are locked, a local subgraph is constructed based on the affected edges and affected nodes, and the remaining paths are replanned.

[0008] Optionally, the preset triggering condition is that the maximum value among the following—the edge weight change rate, the ratio of the patrol personnel's yaw distance to the yaw threshold, the urgency of the new task, and the road closure level—exceeds a preset threshold.

[0009] Optionally, for each facility point to be inspected, determining the access segment and generating the access node along the accessible roadside includes: For each facility point to be inspected, a set of candidate edges is selected from the accessible road edges within a preset search radius; Each candidate edge in the candidate edge set is filtered step by step according to a preset filtering rule to obtain a filtered candidate edge set. For each candidate edge in the filtered candidate edge set, the access cost is calculated based on the access distance, reachability penalty, historical trajectory deviation penalty, and travel mode matching penalty; Select the candidate edge with the minimum access cost as the optimal access line segment, and generate a projection access point on the optimal access line segment as the access node.

[0010] Optionally, the access cost is obtained by weighting and summing the access distance, the reachability penalty, the historical trajectory deviation penalty, and the travel mode matching penalty according to their respective preset weights; Wherein, the access distance is the vertical projection distance from the facility point to be inspected to the candidate side; The reachability penalty is determined as follows: if a candidate edge is marked as restricted after being filtered step by step, a corresponding penalty value is assigned according to the degree of restriction; if the candidate edge is normally accessible, the penalty value is zero. The historical trajectory deviation penalty is determined based on the frequency with which the historical inspection trajectory passes through the candidate edge; The method for determining the penalty for matching the passage mode is as follows: if the passage mode allowed by the facility point to be inspected matches the passage mode allowed by the candidate edge, the penalty value is zero; otherwise, a corresponding penalty value is assigned according to the degree of mismatch.

[0011] Optionally, calculating the comprehensive cost of each edge in the fusion graph model specifically includes: The distance cost, travel time cost, pipeline risk cost, business preference cost, turning penalty cost, and uncertainty cost of each edge in the fusion graph model are obtained respectively. The distance cost, travel time cost, pipeline risk cost, business preference cost, and uncertainty cost are weighted and summed according to their respective preset weights to obtain the basic comprehensive cost of each side. When the heuristic search algorithm expands each edge, the turning penalty cost is calculated based on the actual entry direction, normalized, and then superimposed on the basic comprehensive cost to obtain the final comprehensive cost of each edge.

[0012] Optionally, the step of clustering and sorting multiple tasks to be inspected based on the work order task data and the comprehensive cost to determine the task access order specifically includes: Based on the work order task data, obtain the task characteristic information of each inspection task. The task characteristic information includes task location, task priority, task time window, task service duration, task pressure zone, and task inspection area. Based on the task characteristic information of each inspection task, the clustering distance between any two inspection tasks is calculated; the clustering distance is obtained by weighted summation of the spatial distance between the two tasks, the path cost calculated based on the comprehensive cost, the affiliation difference of the inspection area, the affiliation difference of the pressure zone, the task priority difference, and the task time window cutoff difference. All tasks to be inspected are divided into several task clusters based on the clustering distance; Within each task cluster, tasks that meet preset urgency conditions are prioritized, and the access order within the cluster is determined in a way that minimizes the incremental cost of the path between adjacent tasks. The task clusters are sorted according to their cluster center location, highest priority within the cluster, and earliest deadline within the cluster. The access order within each task cluster is then concatenated to obtain the global task access order.

[0013] Optionally, a heuristic search algorithm that incorporates spatial distance estimation and business attribute estimation in the heuristic function is used to generate the inspection path segment by segment, specifically including: According to the task access order, heuristic search is performed on the fusion graph model, starting from the current inspection position or the previous task node and ending at the target task node. During the heuristic search process, candidate edges and nodes are pruned based on the passability status and access permissions of the traffic and construction data, as well as the task time window of the work order task data, and search branches that do not meet the constraints are discarded. When a feasible path has been found, the cumulative cost of the current feasible path is used as the upper bound, and search branches whose cumulative cost exceeds the upper bound are pruned; the cumulative cost is the sum of the comprehensive costs of each edge on the traversed path. When expanding to the target task node, the path returns to the starting point level by level from the target task node along the recorded backtracking path to generate the current segment path; By concatenating the various path segments according to the order in which the tasks are accessed, a complete inspection path is obtained.

[0014] Optionally, the heuristic function of the heuristic search includes a spatial distance estimation term and a business attribute estimation term; The spatial distance estimation term is determined based on the geometric distance between the current node and the target task node; the business attribute estimation term includes at least one of the following: risk area crossing estimation term, time window urgency estimation term, and facility access reachability estimation term; The risk area crossing estimation term is determined based on the high-risk pipe section area, construction sensitive area, or pressure zone boundary traversed by the passageway from the current node to the target task node; the time window urgency estimation term is determined based on the relationship between the expected arrival time from the current node to the target task node and the target task time window; and the facility access reachability estimation term is determined based on the access difficulty of the facility point corresponding to the target task node.

[0015] Secondly, this application provides a shortest path planning system based on water pipeline network inspection, including: The data acquisition module is used to acquire road network data, pipeline facility data, work order task data, traffic and construction data, SCADA monitoring data, historical inspection trajectory data, and inspection personnel status data. The road network data includes the sides of passable roads. The fusion graph construction module is used to determine the access line segment and generate the access node on the accessible roadside for each facility point to be inspected, establish the virtual access edge between the facility point to be inspected and the access node, and form a fusion graph model containing facility nodes, road nodes and the virtual access edge. The comprehensive cost calculation module is used to calculate the comprehensive cost of each edge in the fusion graph model based on the fusion graph model, combined with the pipeline risk attributes of the pipeline facility data, the task business attributes of the work order data, the traffic status of the traffic and construction data, and the monitoring alarm status of the SCADA monitoring data. The task sorting module is used to cluster and sort multiple tasks to be inspected based on the work order task data and the comprehensive cost, and to determine the task access order. The path search module is used to generate inspection paths segment by segment based on the fusion graph model, the comprehensive cost, and the task access order, using a heuristic search algorithm that includes spatial distance estimation and business attribute estimation in the heuristic function.

[0016] (III) Beneficial Effects The beneficial effects of this invention are as follows: The shortest path planning method based on water pipeline network inspection of this invention adopts a multi-source data fusion to construct a fusion graph model containing road nodes, facility nodes, and virtual access edges. When calculating the comprehensive cost of each edge, it combines pipeline network risk attributes, task business attributes, traffic status, and monitoring alarm status to dynamically assign edge weights. Then, it uses a heuristic search algorithm that includes spatial distance estimation and business attribute estimation in the heuristic function to search and generate inspection paths segment by segment. Compared with the prior art, it can uniformly represent the spatial topological relationship and business attribute association between the road network and pipeline facilities. This allows path planning to simultaneously take into account spatial accessibility, pipeline network risk level, work order business priority, and real-time traffic status, thereby achieving the technical effect of improving the comprehensive rationality and on-site adaptability of inspection paths. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a shortest path planning method based on water pipeline network inspection according to an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a shortest path planning system based on water pipeline network inspection according to an embodiment of this application; Figure 3This is a flowchart illustrating the generation of a fusion graph model in a shortest path planning method based on water pipeline network inspection according to an embodiment of this application. Figure 4 This is a flowchart illustrating the process of obtaining the comprehensive cost in a shortest path planning method based on water pipeline network inspection according to an embodiment of this application; Figure 5 This is a flowchart illustrating the process of obtaining the optimal inspection sequence in a shortest path planning method based on water pipeline network inspection according to an embodiment of this application. Figure 6 This is a schematic diagram illustrating the subsequent development process of detecting dynamic events according to one embodiment of this application. Detailed Implementation

[0018] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] In related technologies, path planning for water pipeline network inspection mainly employs the following methods: Based on GIS road network data, the route of the inspection personnel is planned using the inspection facility points as waypoints and the shortest path algorithm or the traveling salesman problem solution method.

[0020] However, the above methods have the following drawbacks: the road network and pipeline facilities adopt independent topology models, and the spatial access relationship between the roadside and pipeline facility points is not established, which makes the path planning unable to accurately reflect the actual travel path from the road to the facility; the path cost only considers spatial distance or travel time, and does not incorporate multi-source dynamic factors such as pipeline risk level, work order business priority, real-time traffic construction status and SCADA monitoring alarms, which makes it difficult for the generated inspection path to take into account safety response, business urgency and on-site traffic conditions; the heuristic search only uses geometric distance as a heuristic estimate and does not consider the impact of business attributes on path selection, which causes the search direction to deviate from the actual optimal solution and the planning efficiency to be low.

[0021] In view of the above-mentioned shortcomings and deficiencies of the existing technology, the present invention provides a shortest path planning method and system based on water pipeline network inspection, which solves the technical problems of the disconnect between road network and pipeline facility topology, the difficulty in integrating multi-source dynamic factors into path cost calculation, and the lack of business attribute guidance in heuristic search in water pipeline network inspection scenarios. The method acquires road network data, pipeline facility data, work order task data, traffic and construction data, SCADA monitoring data, and inspection personnel status data. For each facility point to be inspected, it determines the access line segment and generates access nodes on the accessible road edges, establishes virtual access edges between the facility point to be inspected and the access nodes, and forms a fusion graph model containing facility nodes, road nodes, and virtual access edges. Based on the fusion graph model, it calculates the comprehensive cost of each edge by combining pipeline risk attributes, task business attributes, traffic status, and monitoring alarm status. Based on work order task data and comprehensive costs, it clusters and sorts multiple tasks to be inspected to determine the task access order. It uses a heuristic search algorithm with spatial distance estimation and business attribute estimation in the heuristic function to search segment by segment to generate inspection paths. Through the above technical solution, the present invention can uniformly represent the spatial topology and business relationship of road network and pipeline facilities, so that route planning can simultaneously take into account spatial accessibility, pipeline risk, business priority and real-time traffic status, thereby achieving the technical effect of improving the comprehensive rationality and on-site adaptability of inspection routes.

[0022] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0023] Figure 1 This is a flowchart illustrating a shortest path planning method based on water pipeline network inspection according to an embodiment of this application, as shown below. Figure 1 As shown, the shortest path planning method based on water pipeline network inspection includes: S100. Acquire road network data, pipeline facility data, work order task data, traffic and construction data, SCADA monitoring data, historical inspection trajectory data, and inspection personnel status data, wherein the road network data includes accessible roadside.

[0024] In the specific implementation process, the above seven types of data are accessed from the city GIS platform, water supply network management system, work order dispatch system, traffic management platform, SCADA monitoring system, historical trajectory database, and inspection personnel positioning terminals, respectively, and the multi-source data are unified into the same coordinate system, time base, and object coding system. The seven types of data are described in detail below: The road network data includes a road node table and a road edge table. The road node table contains the coordinates and node types of each road intersection, endpoint, and internal passage node. The road edge table contains the start node, end node, geometric segment, actual length, traffic direction, road grade, basic traffic speed, traffic mode, and turning restrictions for each passable road edge. The passable road edges include not only ordinary urban road edges, but also pedestrian walkway edges, internal passageway edges within residential areas, internal passageway edges within factory areas, and passable inspection passage edges supplemented by manual annotation or historical trajectories.

[0025] The pipeline network data includes a pipe segment attribute table and a facility point table. The pipe segment attribute table includes the pipe material, diameter, installation year, pressure zone, number of leaks in the past twelve months, number of alarms in the past thirty days, and importance level for each water supply or drainage pipe segment. The facility point table includes the coordinates, facility type, associated pipe segments, permitted access methods, facility importance, access difficulty, and operating status of facilities such as valve wells, water meter wells, fire hydrants, flow meters, pressure monitoring points, and pump stations. The access difficulty level reflects the difficulty of accessing facilities located in different areas such as within residential areas, factory areas, green belts, or beside roads. A higher access difficulty level indicates greater difficulty in reaching the facility from accessible roads.

[0026] The work order task data includes the number of each inspection task for the day, the associated facility point, the task type, the task priority, the earliest start time, the latest completion time, the on-site service duration, the task source, and the task status.

[0027] Traffic and construction data include real-time traffic speed ratios, congestion levels, data collection times and confidence levels for each roadside, as well as the impact range, start and end times, current passability status, and restriction level of construction closures or temporary road closures.

[0028] SCADA monitoring data includes the collection time, pressure value, flow rate, water quality index, alarm type, and alarm level for each monitoring point.

[0029] Historical inspection trajectory data includes the inspection trajectory sequence of each inspection personnel within a preset time period in the past. Each trajectory includes trajectory number, inspection personnel number, time series coordinates, movement speed sequence, stop point and associated task number.

[0030] The inspection personnel status data includes each inspection personnel's current location coordinates, mode of transportation, shift start and end times, current task number, and equipment status. The equipment status reflects the positioning accuracy and communication signal quality of the terminal held by the inspection personnel. When the positioning signal is weak or drifts, this information will be used for subsequent calculation of uncertainty costs.

[0031] Optionally, this embodiment also acquires weather status data, including current weather type, visibility level, rainfall level, wind force level, and weather-based traffic speed correction factor. The weather-based traffic speed correction factor is used to correct roadside travel time affected by weather. For example, the weather-based traffic speed correction factor is set to 0.6 to 0.8 in heavy rain or low visibility scenarios, and 1 in normal weather.

[0032] Through the above steps, this embodiment accesses seven types of data from the city GIS platform, water supply network management system, work order scheduling system, traffic management platform, SCADA monitoring system, historical trajectory database, and inspection personnel positioning terminal, and unifies them into the same coordinate system, time base, and object coding system. This provides a complete and consistent data foundation for subsequent fusion map construction and path planning, avoiding matching errors and information loss caused by heterogeneous multi-source data.

[0033] S200. For each facility point to be inspected, determine the access line segment and generate the access node along the accessible roadside, establish a virtual access edge between the facility point to be inspected and the access node, and form a fusion graph model containing the facility node, road node, and the virtual access edge. (e.g.) Figure 3 (As shown) In step S200, determining the access segment and generating the access node along the accessible roadside for each facility point to be inspected includes: S210. For each facility point to be inspected, a set of candidate edges is selected from the accessible road edges within a preset search radius. That is, a spatial index is established on the geometric line segments of the road edge table with the facility point's coordinates as the center and the preset search radius as the index, and road edges that meet the distance conditions are selected as the initial set of candidate edges. The recommended initial value for the preset search radius is 80 meters, with a range of 30 meters to 200 meters.

[0034] It should be noted that the candidate edge selection target is the passable road edge, not the pipeline segment; the pipeline segment is only used for subsequent pipeline risk cost calculation.

[0035] S220. Filter each candidate edge in the candidate edge set step by step according to the preset filtering rules to obtain the filtered candidate edge set.

[0036] The step-by-step filtering based on preset filtering rules is as follows: The first level eliminates candidate edges that do not match the traffic mode and the inspection mode. For example, when the inspection mode is vehicle inspection, narrow internal passages that only allow walking are eliminated; when the inspection mode is pedestrian inspection, roadside and pedestrian passage edges are retained.

[0037] The second level removes candidate edges that are restricted by travel direction, turning restrictions, or management permissions.

[0038] The third level involves eliminating candidate edges that are currently impassable based on the current accessibility indicators in the construction lockdown data, while retaining candidate edges that are temporarily restricted but still passable and increasing their accessibility penalty.

[0039] Level 4: When the difficulty of accessing a facility indicates that the facility is located inside a residential area, factory area, or green belt, priority should be given to reserving access to internal passageways that are in the same management area as the facility or that have been traversed by historical inspection routes.

[0040] For example, for a valve well facility located within a residential area, three candidate edges are selected within an 80-meter search radius. The first level of filtering eliminates one narrow passage that only allows pedestrian access. The second level of filtering does not eliminate any edges. The third level of filtering eliminates one road edge that is currently completely closed due to construction, leaving one candidate edge. If the final set of candidate edges is empty, the search radius is expanded by a factor of 1.5 up to 120 meters for re-selection until a candidate edge is found or the maximum search radius limit is reached.

[0041] After filtering, if the candidate edge set is empty, the search radius is gradually expanded by a preset multiple of the preset search radius until the maximum search radius limit is reached. The recommended value for the preset multiple is 1.5 times, and the recommended value for the maximum search radius limit is twice the preset search radius but not exceeding 300 meters. If the candidate edge set is still empty after expanding the search radius, the facility is marked as requiring manual confirmation for access and will not directly participate in automatic path planning.

[0042] S230. For each candidate edge in the filtered candidate edge set, calculate the access cost based on access distance, reachability penalty, historical trajectory deviation penalty, and travel mode matching penalty; In the specific implementation process, the access cost is obtained by weighting and summing the access distance, the reachability penalty, the historical trajectory deviation penalty, and the travel mode matching penalty according to their respective preset weights.

[0043] Wherein, the access distance is the vertical projection distance from the facility point to be inspected to the candidate side; The reachability penalty is determined as follows: if a candidate edge is marked as restricted after being filtered step by step, a corresponding penalty value is assigned according to the degree of restriction; if the candidate edge is normally accessible, the penalty value is zero. The historical trajectory deviation penalty is determined based on the frequency with which the historical inspection trajectory passes through the candidate edge; The method for determining the penalty for matching the passage mode is as follows: if the passage mode allowed by the facility point to be inspected matches the passage mode allowed by the candidate edge, the penalty value is zero; otherwise, a corresponding penalty value is assigned according to the degree of mismatch.

[0044] The recommended initial values ​​for each preset weight are as follows: access distance weight 0.45, reachability penalty weight 0.25, historical trajectory deviation penalty weight 0.20, and travel mode matching penalty weight 0.10.

[0045] For example, if the access distance of a candidate edge is normalized to 0.3, and this edge is marked as restricted access with a reachability penalty of 0.6 in the filtering, the historical inspection trajectory passes through this edge frequently with a deviation penalty of 0.2, and the access method is a perfect match with a matching penalty of 0, then the access cost is 0.45×0.3+0.25×0.6+0.20×0.2+0.10×0=0.325.

[0046] S240. Select the candidate edge with the minimum access cost as the optimal access line segment, and generate a projection access point on the optimal access line segment as the access node.

[0047] It is important to note that when generating projection access points, the original roadside is not physically severed. Instead, a new projection access point is added as a node, and this projection access point is connected to the two endpoints of the optimal access segment with child edges. The child edges inherit the attributes of the original candidate edges, and their lengths are recalculated based on the actual distance from the projection access point to the corresponding endpoints, ensuring that the projection access point can be truly reached by subsequent path searches. Then, a virtual access edge connects the facility point and the projection access point; the length of the virtual access edge is the distance from the facility point to the projection access point.

[0048] When multiple facilities to be inspected are projected onto the same candidate edge, each facility independently generates its own projection access point and establishes a sub-edge connection with the two endpoints of the candidate edge, thus avoiding the cascading splitting problem when multiple facilities are projected onto the same edge.

[0049] For the current location of the inspection personnel, the projection access method is also adopted. Within the preset search radius, the optimal access line segment of the current location of the inspection personnel is determined from the sides of the passable road. The projection access point of the inspection personnel is generated on the optimal access line segment, and a virtual access edge is established between the current location of the inspection personnel and the projection access point of the inspection personnel. This enables the subsequent path search to truly access the fusion graph model from the current location of the inspection personnel, rather than simply taking the nearest road node.

[0050] For road edges with bidirectional traffic, two directed edges sharing the original attributes are generated (one forward and one reverse). For road edges with unidirectional traffic, only one directed edge is generated, allowing traffic in that direction, while the reverse direction is unreachable. Edges with turning restrictions set to true are used in subsequent path searches to trigger permission pruning or turning penalties based on the entry direction.

[0051] After the above processing, the original nodes and edges of the road network, the access nodes and virtual access edges generated by each facility point, and the access nodes and virtual access edges of the current location of the inspection personnel are unified into the same directed weighted graph to form a fusion graph model.

[0052] Through step S200, this embodiment generates projection access points and establishes virtual access edges on the sides of passable roads, truly connecting water pipeline facilities to the road network map, avoiding the access deviation caused by simply taking the nearest road node in the traditional method; at the same time, the current location of the inspection personnel is also projected to ensure that the starting point of the path planning corresponds precisely to the actual location, ensuring that the path planning results can accurately reflect the actual reachable path from the road to the facility.

[0053] S300. Based on the fusion graph model, and combining the pipeline risk attributes of the pipeline facility data, the task business attributes of the work order data, the traffic status of the traffic and construction data, and the monitoring alarm status of the SCADA monitoring data, calculate the comprehensive cost of each edge in the fusion graph model.

[0054] Specifically, calculating the comprehensive cost of each edge in the fusion graph model includes: S310. Obtain the distance cost, travel time cost, network risk cost, business preference cost, turning penalty cost, and uncertainty cost (e.g., ...) for each edge in the fusion graph model. Figure 4 (As shown).

[0055] Specifically, the distance cost is determined based on the actual length of the roadside or the geometric length of the virtual access edge. For a regular roadside, the distance cost is equal to the actual length of the edge; for a virtual access edge, the distance cost is equal to the geometric distance from the facility point to the projected access point.

[0056] The travel time cost is calculated based on the roadside length and basic travel speed, combined with traffic and construction data correction coefficients for traffic status, weather, and construction. The basic travel speed is derived from roadside data or historical inspection trajectory statistics; the traffic status correction coefficient is derived from the travel speed ratio in real-time traffic data, with a default value of 1; the weather correction coefficient is derived from the weather travel speed correction factor in weather status data, with a value of 1 for normal weather and 0.6 to 0.8 for heavy rain or low visibility scenarios; the construction correction coefficient is derived from the degree of restriction in construction control data, with a value of 1 for unaffected areas and 0.5 to 0.9 for restricted traffic. Completely closed roadside sections are directly marked as impassable and are not included in the cost calculation. It is necessary to ensure consistency between length and speed units; during implementation, unit conversion should be performed to output travel time in minutes.

[0057] The pipeline risk cost is calculated by weighting the number of leaks, pipe age, alarm frequency, and importance level of pipe sections near or associated with the roadside. Specifically, the number of leaks is normalized as the ratio of the number of leaks in the vicinity within twelve months to the maximum number of leaks in the region; pipe age is normalized as the ratio of pipe age to a preset upper limit, and for multi-material pipe sections, it is weighted by pipe length; the number of alarms is normalized as the ratio of the number of alarms in the past thirty days to the maximum number of alarms in the region; and the importance level is determined based on the importance of the associated pipe sections, with a value of 1 for key protection areas, 0 for ordinary areas, and 0.7 for construction-sensitive areas. The pipeline risk cost is obtained by weighting and summing each indicator according to its preset weight.

[0058] The service preference cost is determined based on the priority intensity of tasks awaiting inspection around the roadside, the coverage contribution of key inspection areas, and the degree of matching between pressure zones. Specifically, the priority intensity is calculated based on the priority of tasks awaiting inspection within a preset distance around the roadside and distance attenuation; edges closer to high-priority tasks have higher priority intensity. The coverage contribution is determined based on the importance level of the associated pipe segments along the roadside; virtual access edges without associated pipe segments have a value of zero. The degree of matching between pressure zones is an indicator; it is set to 1 when there are tasks awaiting inspection within the pressure zone where the roadside is located, and 0 otherwise. The calculation of the service preference cost ensures that edges closer to high-priority tasks, key coverage areas, or tasks within the same zone have lower service preference costs, making them more likely to be selected by path search.

[0059] The turning penalty cost is determined based on the turning angle, turning type, and turning restrictions of the current edge, including at least one of the following: ordinary turning penalty, U-turn penalty, and penalty for complex intersections or crossing main roads. Since the turning penalty cost depends on the direction of entry into the current edge, i.e., it is related to the previous edge, and is a path-dependent quantity, it cannot be uniquely determined for a single edge during the graph construction phase. Therefore, it is calculated in real time based on the actual entry direction when the search algorithm expands each edge.

[0060] The uncertainty cost is calculated by weighting the probability of temporary lockdown, map reliability error, positioning signal stability, historical inspection trajectory deviation, and facility access difficulty. Specifically, the probability of temporary lockdown is derived from the restriction level of the construction lockdown data; map reliability error is derived from the map reliability in the roadside data; positioning signal stability is derived from the equipment status in the inspection personnel status data, with lower values ​​when the equipment is normal and higher values ​​when the signal is weak or the positioning is drifting; the historical inspection trajectory deviation is obtained by inverse normalization based on the number of times the candidate edge has been traversed by historical inspection trajectories; and facility access difficulty is determined based on the facility access difficulty level.

[0061] For example, if a roadside section is 200 meters long, has a basic traffic speed of 30 kilometers per hour, a real-time traffic speed ratio of 0.7, a weather-related traffic speed correction factor of 1, and no construction impact, then the travel time cost for this section is approximately 5.7 minutes. If a nearby pipe section has experienced two leaks within the past twelve months, is 15 years old, has triggered one alarm in the past thirty days, and belongs to a key protection area, then the pipeline risk cost will be calculated with a higher value after normalization and weighting, causing the path search to favor detouring through pipe sections with lower risk.

[0062] S320. The distance cost, travel time cost, pipeline risk cost, business preference cost, and uncertainty cost are weighted and summed according to their respective preset weights to obtain the basic comprehensive cost of each side.

[0063] In the specific implementation process, the distance cost, travel time cost, pipeline risk cost, business preference cost, and uncertainty cost are normalized to ensure that each cost item falls within the same numerical range. The normalization method is as follows: normalize each cost item in the current dataset by subtracting the maximum value from the minimum value, and add a very small positive number to the denominator to prevent division by zero.

[0064] After normalization, the weighted sums are calculated according to their respective preset weights to obtain the basic comprehensive cost of each side. The preset weights satisfy the normalization constraint, that is, the sum of all weights is 1.

[0065] In the specific implementation process, different preset weight combinations are set according to different inspection scenarios: in daily inspection scenarios, distance and time weights are higher to balance efficiency and coverage; in leakage verification scenarios, pipeline risk cost weights are higher to prioritize coverage of high-risk areas and alarm points; in emergency repair scenarios, travel time cost weights are higher to prioritize reaching emergency tasks; and in nighttime inspection scenarios, uncertainty cost weights are higher to increase safety factors.

[0066] S330. When expanding each edge using the heuristic search algorithm, the turning penalty cost is calculated based on the actual entry direction, normalized, and then superimposed on the basic comprehensive cost to obtain the final comprehensive cost of each edge.

[0067] In other words, when the search algorithm expands an edge from a node, it calculates the turning penalty cost in real time based on the actual turning angle from the previous edge to the current edge, normalizes it to the theoretical maximum value so that it falls within the range of zero to 1, and then adds it to the basic comprehensive cost of the edge with the corresponding preset weights to obtain the complete comprehensive cost used for this expansion. When the first edge is the starting point of the path, i.e., there is no previous edge, the turning penalty cost is zero.

[0068] This approach ensures the completeness of the comprehensive cost definition and solves the problem that the turning penalty cost cannot be pre-calculated and normalized in static edge weights.

[0069] In the specific implementation process, this embodiment also calculates the node service cost and task breach penalty for each task node. The node service cost includes on-site dwell time, task processing time, on-site preparation time, security check time, and end-point approach time from the access point to the facility body; wherein the task processing time is directly taken from the on-site service duration in the work order task data, and the end-point approach time is determined by the virtual access edge length combined with the facility access difficulty.

[0070] The task breach penalties include lateness penalties, early arrival penalties, priority inversion penalties, and task skipping penalties. Lateness penalties are calculated based on the timeout amount when the expected arrival time is later than the latest task completion time; early arrival penalties are calculated based on the pre-arrival waiting time when the expected arrival time is earlier than the earliest task start time; priority inversion penalties indicate a breach where a low-priority task is prioritized over a high-priority urgent task; and task skipping penalties indicate tasks that are not yet accessed in this planning round but have not yet been cancelled.

[0071] Since the overall edge cost is a normalized dimensionless quantity, while the node service cost and task default penalty are measured in minutes, and the two are of different magnitudes, when accumulating them into the cumulative cost in subsequent path searches, the node service cost and task default penalty need to be multiplied by a conversion factor to make them of the same magnitude as the overall edge cost.

[0072] Optionally, the preset weights are calibrated every preset period based on the actual inspection trajectory and task completion records. Calibration methods include: increasing the weight of passage time cost if the arrival time error is consistently large; increasing the weight of pipeline risk cost and business preference cost if high-risk point coverage is insufficient; and increasing the weight of uncertainty cost if frequent passage failures or positioning deviations occur. Each adjustment does not exceed a preset step size, and the weight sum is kept to 1. The recommended preset period is 30 days, and the recommended preset step size is 0.05.

[0073] Through the above steps, this embodiment comprehensively considers six cost indicators: distance, travel time, pipeline risk, business preference, turning penalty, and uncertainty. During the search process, the turning penalty is dynamically superimposed based on the actual entry direction. This allows the comprehensive cost to reflect both static road attributes and dynamic traffic conditions, avoiding the shortcomings of traditional single distance cost in distinguishing different traffic conditions. This provides an accurate cost assessment basis for path search.

[0074] S400. Based on the work order task data and the comprehensive cost, cluster and sort multiple tasks to be inspected to determine the task access order.

[0075] In the specific implementation process, see Figure 5 Step S400 includes: S410. Obtain task characteristic information for each inspection task based on the work order task data. The task characteristic information includes task location, task priority, task time window, task service duration, task pressure zone, and task inspection area.

[0076] S420. Calculate the clustering distance between any two inspection tasks based on the task characteristic information of each inspection task; the clustering distance is obtained by weighted summation of the spatial distance between the two tasks, the path cost calculated based on the comprehensive cost, the affiliation difference of the inspection area, the affiliation difference of the pressure zone, the task priority difference, and the task time window cutoff difference.

[0077] Wherein, the spatial distance is the geometric distance between the coordinates of the two task facilities; the path cost is the comprehensive cost of the shortest path between the two tasks calculated based on the comprehensive cost of each edge in the fusion graph, used to reflect the actual travel cost and avoid situations where two task points are geometrically close but inaccessible by road; the affiliation difference is zero if they are of the same type and one if they are different; the task priority difference is obtained by normalizing the priority difference between the two tasks; the task time window cutoff difference is obtained by normalizing the ratio of the difference in the latest completion time of the two tasks to the daily inspection work duration.

[0078] S430. Divide all tasks to be inspected into several task clusters according to the clustering distance.

[0079] It is important to note that during implementation, when the number of tasks to be inspected is less than a preset task threshold, a constrained hierarchical clustering method is used: initially, each task is a cluster, and the two clusters with the smallest clustering distance are repeatedly selected as candidate merging objects. Merging constraints include: if the number of tasks within a merged cluster exceeds a preset cluster limit, merging is prohibited and the next set of candidate clusters is selected; if there is an unsatisfactory time window conflict, merging is prohibited. The time window conflict is determined as follows: tasks within the merged cluster are simulated in ascending order of their latest completion time, and the inter-task travel time and on-site service time are accumulated sequentially from the current time to obtain the expected completion time of each task. If the expected completion time of any task is later than its latest completion time, a time window conflict is determined. This process continues until there are no more merging clusters, or the minimum clustering distance between all merging clusters is greater than a preset clustering cutoff threshold. The clustering distance between two clusters is calculated using the average connectivity method, i.e., the average of the clustering distances between all task pairs in the two clusters is taken.

[0080] When the number of tasks to be inspected is not less than the preset task threshold, the DBSCAN clustering method is adopted. The clustering distance is used as the pre-calculated distance matrix and input into DBSCAN. The generated noisy tasks are assigned according to the following rules: For each noisy task, the minimum clustering distance to each existing cluster is calculated. If the minimum value is not greater than the preset neighborhood threshold, it is merged into the corresponding cluster. If it is still greater than the preset neighborhood threshold, it is formed into a separate cluster as an independent inspection cluster to participate in the subsequent sorting, ensuring that all tasks are included in a certain inspection cluster.

[0081] For example, on a certain day, 25 inspection tasks are clustered into 6 task clusters. The first task cluster contains 5 fire hydrant inspection tasks located in the same inspection area. Two tasks in the cluster meet the emergency conditions and are prioritized, while the remaining 3 are inserted according to the minimum incremental cost. The order of the remaining 5 task clusters is determined in a similar way. The 6 clusters are sorted by comprehensive score and concatenated to obtain the global task access sequence.

[0082] For each task cluster, the cluster center task is selected from the task node with the lowest average path cost to other tasks within the cluster, rather than the geometric center, because the inspection path depends on road and passage costs, and the geometric center may fall in an impassable location or not be the actual task point. The cluster priority level is the highest priority task within the cluster.

[0083] S440. Within each task cluster, tasks that meet preset urgency conditions are prioritized, and the access order within the cluster is determined in a way that minimizes the incremental cost of the path between adjacent tasks.

[0084] The preset emergency conditions include a task priority reaching a preset emergency level, or the difference between the task's time window deadline and the current time being less than a preset time threshold. Emergency tasks are arranged in ascending order of deadline and descending order of priority.

[0085] For the remaining non-urgent tasks, the minimum incremental cost insertion method is used to insert them into the current sequence one by one. For each task to be inserted, all possible insertion positions are enumerated, positions that would inevitably lead to a breach of the time window are eliminated, and the position with the minimum incremental path cost is selected for insertion. The incremental path cost is the additional path cost after insertion minus the path cost before insertion, plus the cost corresponding to the on-site service time of the task.

[0086] S450. Sort each task cluster according to its cluster center location, highest priority within the cluster, and earliest deadline within the cluster. Concatenate the access order within each task cluster to obtain the global task access order.

[0087] In the specific implementation process, starting from the current location of the inspector, the inter-cluster access order is determined by a comprehensive score based on the path cost to the cluster center, the highest priority within the cluster, and the earliest deadline within the cluster. The comprehensive score includes the path cost to the cluster center, which is the comprehensive cost of the shortest path from the inspector's current location to the task at the cluster center. This cost is pre-calculated on the fusion graph, normalized, and then weighted and summed with two other indicators of consistent dimensions. Clusters are arranged in ascending order of comprehensive score, and the intra-cluster access order of each cluster is then concatenated to obtain the global task access sequence.

[0088] Through the above step S400, this embodiment performs clustering based on six-dimensional task characteristics: spatial distance, path cost, area affiliation, pressure partition affiliation, priority, and time window deadline. Tasks that are spatially close, business-related, and time-sensitive are grouped together. Within each cluster, the access order is determined by prioritizing urgent tasks and inserting them with the minimum incremental cost. After sorting the clusters based on multiple factors, the sequences are spliced ​​into a global sequence, avoiding the combinatorial explosion problem of global search. At the same time, it ensures the priority completion of urgent tasks and the optimal path cost between adjacent tasks.

[0089] S500. Based on the fusion graph model, the comprehensive cost, and the task access order, a heuristic search algorithm containing spatial distance estimation and business attribute estimation in the heuristic function is used to generate an inspection path segment by segment.

[0090] Specifically, step S500 includes: S510. In accordance with the task access order, perform heuristic search on the fusion graph model, starting from the current inspection position or the previous task node and ending at the target task node.

[0091] If the starting point or ending point is a facility point rather than a road map node, the facility access node generated in step S200 is used as the starting point or ending point in the search, and the virtual access edge from the facility body to the access node is still included in the path cost.

[0092] S520. During the heuristic search process, candidate edges and candidate nodes are pruned based on the passability status and access permissions of the traffic and construction data, as well as the task time window of the work order task data, and search branches that do not meet the constraints are discarded.

[0093] Specifically, constrained pruning includes the following types: The blocking and pruning mechanism excludes candidate edges when the passability status of a candidate edge is marked as impassable or the restriction level reaches the prohibition level. Permission pruning: When the passage method of a candidate edge does not match the inspection method, the branch is abandoned to prevent vehicles from entering pedestrian or restricted areas of the factory. Time window pruning: If the current estimated arrival time plus the minimum remaining service time is still later than the latest completion time of the target task, the search branch that will inevitably default will be excluded in advance. Upper bound pruning: When the cumulative cost of the current search path exceeds the cumulative cost of the known optimal feasible path, abandon the branch to reduce search complexity. Loop pruning reduces detours and backtracking when repeatedly visiting the same node within a short period of time without increasing risk or reward.

[0094] S530. When a feasible path has been found, the cumulative cost of the current feasible path is used as the upper bound. Search branches with a cumulative cost exceeding this upper bound are pruned. The cumulative cost is the sum of the combined costs of all edges along the traversed path. For example, in a certain search segment, if a feasible path with a cumulative cost of 12.5 has been found from the starting point to the target node, all branches with a cumulative cost exceeding 12.5 in subsequent searches are pruned. As the search progresses, if a feasible path with a lower cost is found, the upper bound is updated, allowing subsequent searches to further prune branches with higher costs, thus improving search efficiency.

[0095] S540. When expanding to the target task node, return to the starting point level by level from the target task node along the recorded backtracking path to generate the current segment path. Specifically, when the target node is selected as the current expansion node, backtrack to the starting point level by level from the target node along the recorded parent node relationship to obtain the node sequence, and then generate the edge sequence according to the edge corresponding to the adjacent node. At the same time, output the path cost, estimated arrival time, path length and travel time of the segment.

[0096] S550. Concatenate the path segments according to the task access order to obtain the complete inspection path.

[0097] Specifically, the local search described above is performed sequentially according to the task access order, and the path segments are concatenated in the access order. If there are duplicate access edges or duplicate nodes between adjacent segments, duplicates are removed during concatenation. The final output includes the task access order, global path edge sequence, path geometry, estimated arrival time of each task, cumulative comprehensive cost, total distance, and total travel time.

[0098] The heuristic function for the heuristic search includes a spatial distance estimation term and a business attribute estimation term.

[0099] The spatial distance estimation term is determined based on the geometric distance between the current node and the target task node. Specifically, it is normalized according to the minimum and maximum distances from all nodes to the target node within the current segment search subgraph. The normalization boundary is determined once before the search begins and is not updated during the search process to ensure the stability of the heuristic value within the same segment.

[0100] The business attribute estimation items include at least one of the following: risk area crossing estimation item, time window urgency estimation item, and facility access reachability estimation item.

[0101] The risk area crossing estimation term is determined based on the high-risk pipe section areas, construction-sensitive areas, or pressure zone boundaries traversed by the passageway from the current node to the target task node. Specifically, the judgment range is the straight buffer zone or rough passageway from the current node to the target node. If it crosses a high-risk area, a construction-sensitive area, or a pressure zone, the value is increased. The terms are then weighted and summed according to the proportion of high-risk areas, the proportion of pressure zones crossed, and the proportion of construction-sensitive areas.

[0102] The time window urgency estimation term is determined based on the relationship between the estimated arrival time from the current node to the target task node and the target task's time window. Specifically, the estimated arrival time is obtained by adding the estimated arrival time of the current node to the ratio of the geometric distance from the current node to the target node to the estimated average speed. If the estimated arrival time is not later than the latest completion time of the target task, the time window urgency estimation term is set to zero. If the estimated arrival time is later than the latest completion time, it is determined by the ratio of the excess amount to the time window width, with an upper limit of 1 to avoid abnormal amplification of this term in cases of severe delays.

[0103] The facility access reachability estimate is determined based on the access difficulty of the facility point corresponding to the target task node. Priority is given to the access cost calculated in step S200 and normalized; if there is no access cost, it is calculated according to the facility's access difficulty level; if the target access edge is affected by construction, a construction penalty is added.

[0104] It should be noted that, since the heuristic function introduces business heuristics such as risk area crossing, time window urgency, and facility access reachability in addition to geometric distance estimation, the overall heuristic function does not strictly satisfy the lower bound condition of cost. Therefore, the path obtained in this step is approximately optimal in terms of comprehensive business cost. If a strictly mathematically optimal solution is required, the weight of the business attribute estimation term can be set to zero, so that the heuristic function degenerates into an acceptable heuristic containing only spatial distance estimation.

[0105] The spatial distance estimation term, risk area crossing estimation term, time window urgency estimation term, and facility access reachability estimation term are respectively weighted and summed according to their corresponding preset weights to obtain the heuristic function; the preset weights satisfy the normalization constraint, and the sum of the weights in each group is 1.

[0106] Through the above steps, this embodiment uses a heuristic function that integrates spatial distance estimation and business attribute estimation to guide the search direction. Spatial distance estimation provides basic directional guidance, while the estimation of three business attributes—risk area crossing, time window urgency, and facility access accessibility—further incorporates pipeline network risks and business constraints into the search guidance. At the same time, constraint pruning eliminates infeasible branches and upper bound pruning removes suboptimal branches, which greatly reduces the amount of computation while ensuring optimality of the path search, and can quickly generate inspection paths that take into account both traffic efficiency and business constraints.

[0107] In specific implementation, the shortest path planning method based on water pipeline network inspection in this application embodiment also includes: S600. During the inspection process, dynamic events are detected based on the traffic and construction data, SCADA monitoring data, and inspection personnel status data.

[0108] In the specific implementation process, we continuously monitor real-time data changes during the inspection process. The types of dynamic events include: changes in the overall cost of the roadside, the deviation distance of the current position of the inspection personnel from the original planned route, the emergence of new emergency work orders, and changes in the construction control level.

[0109] S700. When the dynamic event meets the preset triggering conditions, determine the affected edges and affected nodes in the fusion graph model, lock the completed tasks, construct a local subgraph based on the affected edges and affected nodes, and replan the remaining paths (see...). Figure 6 ).

[0110] Specifically, the preset triggering conditions are that the maximum value among the following—the edge weight change rate, the ratio of the patrol personnel's yaw distance to the yaw threshold, the urgency of the new task, and the road closure level—exceeds the preset threshold.

[0111] Wherein, the edge weight change rate is the maximum value of the ratio of the change in comprehensive cost of each edge before and after the event to the original comprehensive cost in the fusion graph; the yaw distance of the inspection personnel is the shortest distance from the current position of the inspection personnel to the original planned path; the urgency of the new task is obtained by normalizing the priority of the new task; the road closure level is derived from the construction closure data, with 1 for complete closure and 0.5 to 0.9 for restricted passage.

[0112] Upon triggering, the system determines the set of affected edges, which includes: edges where the change in overall cost exceeds a preset change ratio, edges affected by construction lockdown, and edges within a preset influence radius of the added or modified task. Centered on the set of affected edges, the system searches for neighboring nodes according to the preset influence radius to obtain the set of affected nodes, forming a local subgraph. On the local subgraph, the system re-executes the task sorting in step S400 and the path search in step S500 for the remaining unfinished tasks, locking completed tasks out of the replanning process.

[0113] Optionally, during the replanning process, the system will output the reasons for the replanning, the scope of impact, and the updated path for inspection personnel and dispatch managers to view and confirm.

[0114] The recommended initial value for the above preset threshold is 0.35, the recommended initial value for the yaw threshold is 80 meters, the recommended initial value for the local submap influence radius is 1000 meters, and this can be extended to 3000 meters in emergency repair scenarios.

[0115] Through the above steps, this embodiment continuously monitors dynamic events such as road closures, new emergency tasks, and deviations during the inspection process. When the comprehensive triggering conditions exceed the preset threshold, the completed tasks are locked and only a local sub-graph is constructed for replanning in the affected area. This avoids the huge computational overhead of global replanning and ensures that the inspection path can respond to sudden changes in a timely manner, thereby improving the robustness and practical usability of the path planning.

[0116] The above embodiments will be further illustrated by a complete application example below.

[0117] A water supply company needs to inspect 20 locations (12 valve wells, 5 fire hydrants, and 3 pressure monitoring points) on a given day. Inspector P1 will travel by electric vehicle, working from 09:00 to 17:00. The system accesses 3680 roadside locations (approximately 142 kilometers in total), 20 facility points, 20 work orders (3 with priority level 5 and 5 with priority level 4), real-time traffic data (speed ratio less than 0.5 for 12 roadside locations), construction closure data (2 roads closed from 10:00 to 16:00), SCADA monitoring data (abnormal fluctuations at a pressure monitoring point), historical inspection routes for the past 30 days, inspector status, and light rain weather data (speed correction factor 0.85).

[0118] Access Calculation: Taking valve well F003 as an example, three candidate edges were found within a search radius of 80 meters, with access distances of 12 meters, 18 meters, and 9 meters, respectively. The candidate edge with the shortest access distance was eliminated due to construction closure. Among the remaining candidate edges, the edge with the best matching traffic mode and the highest historical trajectory fit had the lowest access cost and was selected as the optimal access line segment. A projection access point was generated and a virtual access edge was established.

[0119] Fusion Graph Construction: After all 20 facility points are connected, a fusion graph is constructed, and the comprehensive cost is calculated for each edge. Edges close to old pipelines and with frequent alarms have higher pipeline risk costs. Edges closed for construction are marked as impassable. Edges close to high-priority tasks have lower service preference costs. Light rain increases the travel time cost of affected edges.

[0120] Clustering and Ranking: After pre-calculating the path cost for each pair of the 20 tasks, the cluster distance is calculated. Since the number of tasks is less than 200, hierarchical clustering is used to divide them into 4 inspection clusters. Within each cluster, urgent tasks are prioritized and inserted in the order of minimum incremental cost. Between clusters, tasks are sorted and concatenated according to comprehensive score to obtain the global access sequence: P1→T07→T03→T11→T02→T15→T18→...→T20.

[0121] Path search: Taking the T07 to T03 section as an example, a heuristic search is performed with the access nodes at both ends as the starting and ending points. Control and pruning are applied to the construction closure edges. The search direction is guided by spatial distance and business attribute estimation terms, outputting the passable route with the lowest overall cost. The complete inspection path is obtained by splicing together the segments.

[0122] Incremental Replanning: At 10:35, during the inspection, SCADA generated a new emergency task T21. Simultaneously, the inspection personnel deviated 95 meters, exceeding the threshold of 80 meters, with a yaw ratio of 1.19. The urgency level of T21 was 1, and the maximum value of the four trigger indicators exceeded the preset threshold of 0.35, triggering replanning. The system locked completed tasks, constructed a local subgraph within a 1000-meter influence range, reordered and searched the remaining tasks, inserted T21 as the second access point after the current location, and output the updated path and influence range.

[0123] In the above case, the sample road network is relatively small, and when the radius of influence is 1000 meters, the local submap may cover a large proportion of the road network; in a real city-level road network, the radius of influence only covers a local area around the emergency, which is still considered a local incremental replanning.

[0124] In summary, this embodiment achieves accurate projection access of water pipeline network facilities to the road network by constructing a fusion graph model that includes facility nodes, road nodes, and virtual access edges. By integrating six cost indicators—distance, travel time, pipeline risk, business preference, turning penalty, and uncertainty—and dynamically adding turning penalties during the search process, edge weights comprehensively reflect static pipeline attributes and real-time travel conditions. Global path planning is decomposed into intra-cluster and inter-cluster two-level sorting based on multi-dimensional task features, effectively reducing solution complexity and ensuring priority completion of urgent tasks. A heuristic search algorithm that integrates spatial distance estimation and business attribute estimation, combined with a multi-layer pruning strategy, significantly improves search efficiency while ensuring path optimality. Continuous monitoring of dynamic events during inspections and locking completed tasks for incremental replanning of local subgraphs when trigger conditions are met avoids the overhead of global recalculation, enabling paths to respond promptly to sudden changes such as road closures and new urgent tasks. The coordinated efforts of all these links ultimately achieved a comprehensive and optimal inspection route planning that takes into account spatial accessibility, pipeline risks, business priorities, and real-time traffic conditions, significantly improving the efficiency of water pipeline inspections and emergency response capabilities.

[0125] Furthermore, see Figure 2 This application also includes a shortest path planning system based on water pipeline network inspection, comprising: The data acquisition module is used to acquire road network data, pipeline facility data, work order task data, traffic and construction data, SCADA monitoring data, historical inspection trajectory data, and inspection personnel status data. The road network data includes the sides of passable roads. The fusion graph construction module is used to determine the access line segment and generate the access node on the accessible roadside for each facility point to be inspected, establish the virtual access edge between the facility point to be inspected and the access node, and form a fusion graph model containing facility nodes, road nodes and the virtual access edge. The comprehensive cost calculation module is used to calculate the comprehensive cost of each edge in the fusion graph model based on the fusion graph model, combined with the pipeline risk attributes of the pipeline facility data, the task business attributes of the work order data, the traffic status of the traffic and construction data, and the monitoring alarm status of the SCADA monitoring data. The task sorting module is used to cluster and sort multiple tasks to be inspected based on the work order task data and the comprehensive cost, and to determine the task access order. The path search module is used to generate inspection paths segment by segment based on the fusion graph model, the comprehensive cost, and the task access order, using a heuristic search algorithm that includes spatial distance estimation and business attribute estimation in the heuristic function.

[0126] Through the above system, this embodiment organically integrates five modules: data acquisition, fusion graph construction, comprehensive cost calculation, task sorting, and path search, forming a complete automated link from multi-source data access to final inspection path output. The data flow between each module is closely connected, and the entire process from facility access modeling, multi-dimensional cost assessment, task priority sorting to heuristic path search can be completed without manual intervention. This significantly reduces the manual cost and computational complexity of inspection path planning, while enabling the planning results to comprehensively consider multiple factors such as spatial accessibility, pipeline network risk, business priority, and real-time traffic conditions, providing efficient and intelligent path planning support for water pipeline network inspection.

[0127] In the description of this invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0128] In the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0129] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make modifications, alterations, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A shortest path planning method based on water pipeline network inspection, characterized in that, include: Acquire road network data, pipeline facility data, work order task data, traffic and construction data, SCADA monitoring data, historical inspection trajectory data, and inspection personnel status data. The road network data includes accessible roadside areas. For each facility point to be inspected, an access line segment is determined and an access node is generated along the accessible road. A virtual access edge is established between the facility point to be inspected and the access node, forming a fusion graph model that includes facility nodes, road nodes and the virtual access edge. Based on the fusion graph model, and combining the pipeline risk attributes of the pipeline facility data, the task business attributes of the work order data, the traffic status of the traffic and construction data, and the monitoring alarm status of the SCADA monitoring data, the comprehensive cost of each edge in the fusion graph model is calculated. Based on the work order task data and the comprehensive cost, multiple tasks to be inspected are clustered and sorted to determine the task access order. Based on the fusion graph model, the comprehensive cost, and the task access order, a heuristic search algorithm that includes spatial distance estimation and business attribute estimation in the heuristic function is used to generate inspection paths segment by segment.

2. The shortest path planning method based on water pipeline network inspection according to claim 1, characterized in that, The method further includes: During the inspection process, dynamic events are detected based on the traffic and construction data, SCADA monitoring data, and inspection personnel status data. When the dynamic event meets the preset triggering conditions, the affected edges and affected nodes in the fusion graph model are determined, the completed tasks are locked, a local subgraph is constructed based on the affected edges and affected nodes, and the remaining paths are replanned.

3. The shortest path planning method based on water pipeline network inspection according to claim 2, characterized in that, The preset triggering conditions are that the maximum value among the following—the edge weight change rate, the ratio of the inspection personnel's yaw distance to the yaw threshold, the urgency of the new task, and the road closure level—exceeds the preset threshold.

4. The shortest path planning method based on water pipeline network inspection according to claim 1, characterized in that, The step of determining the access segment and generating the access node along the accessible roadside for each facility point to be inspected includes: For each facility point to be inspected, a set of candidate edges is selected from the accessible road edges within a preset search radius; Each candidate edge in the candidate edge set is filtered step by step according to a preset filtering rule to obtain a filtered candidate edge set. For each candidate edge in the filtered candidate edge set, the access cost is calculated based on the access distance, reachability penalty, historical trajectory deviation penalty, and travel mode matching penalty; Select the candidate edge with the minimum access cost as the optimal access line segment, and generate a projection access point on the optimal access line segment as the access node.

5. The shortest path planning method based on water pipeline network inspection according to claim 4, characterized in that, The access cost is obtained by weighting and summing the access distance, the reachability penalty, the historical trajectory deviation penalty, and the travel mode matching penalty according to their respective preset weights. Wherein, the access distance is the vertical projection distance from the facility point to be inspected to the candidate side; The reachability penalty is determined as follows: if a candidate edge is marked as restricted after being filtered step by step, a corresponding penalty value is assigned according to the degree of restriction; if the candidate edge is normally accessible, the penalty value is zero. The historical trajectory deviation penalty is determined based on the frequency with which the historical inspection trajectory passes through the candidate edge; The method for determining the penalty for matching the passage mode is as follows: if the passage mode allowed by the facility point to be inspected matches the passage mode allowed by the candidate edge, the penalty value is zero; otherwise, a corresponding penalty value is assigned according to the degree of mismatch.

6. The shortest path planning method based on water pipeline network inspection according to claim 1, characterized in that, Calculating the comprehensive cost of each edge in the fusion graph model specifically includes: The distance cost, travel time cost, pipeline risk cost, business preference cost, turning penalty cost, and uncertainty cost of each edge in the fusion graph model are obtained respectively. The distance cost, travel time cost, pipeline risk cost, business preference cost, and uncertainty cost are weighted and summed according to their respective preset weights to obtain the basic comprehensive cost of each side. When the heuristic search algorithm expands each edge, the turning penalty cost is calculated based on the actual entry direction, normalized, and then superimposed on the basic comprehensive cost to obtain the final comprehensive cost of each edge.

7. The shortest path planning method based on water pipeline network inspection according to claim 1, characterized in that, The step of clustering and sorting multiple tasks to be inspected based on the work order task data and the comprehensive cost to determine the task access order specifically includes: Based on the work order task data, obtain the task characteristic information of each inspection task. The task characteristic information includes task location, task priority, task time window, task service duration, task pressure zone, and task inspection area. Based on the task characteristic information of each inspection task, the clustering distance between any two inspection tasks is calculated; the clustering distance is obtained by weighted summation of the spatial distance between the two tasks, the path cost calculated based on the comprehensive cost, the affiliation difference of the inspection area, the affiliation difference of the pressure zone, the task priority difference, and the task time window cutoff difference. All tasks to be inspected are divided into several task clusters based on the clustering distance; Within each task cluster, tasks that meet preset urgency conditions are prioritized, and the access order within the cluster is determined in a way that minimizes the incremental cost of the path between adjacent tasks. The task clusters are sorted according to their cluster center location, highest priority within the cluster, and earliest deadline within the cluster. The access order within each task cluster is then concatenated to obtain the global task access order.

8. The shortest path planning method based on water pipeline network inspection according to claim 1, characterized in that, A heuristic search algorithm, employing a heuristic function that includes spatial distance estimation and business attribute estimation, is used to generate inspection paths segment by segment. Specifically, this includes: According to the task access order, heuristic search is performed on the fusion graph model, starting from the current inspection position or the previous task node and ending at the target task node. During the heuristic search process, candidate edges and nodes are pruned based on the passability status and access permissions of the traffic and construction data, as well as the task time window of the work order task data, and search branches that do not meet the constraints are discarded. When a feasible path has been found, the cumulative cost of the current feasible path is used as the upper bound, and search branches whose cumulative cost exceeds the upper bound are pruned; the cumulative cost is the sum of the comprehensive costs of each edge on the traversed path. When expanding to the target task node, the path returns to the starting point level by level from the target task node along the recorded backtracking path to generate the current segment path; By concatenating the various path segments according to the order in which the tasks are accessed, a complete inspection path is obtained.

9. The shortest path planning method based on water pipeline network inspection according to claim 8, characterized in that, The heuristic function for the heuristic search includes a spatial distance estimation term and a business attribute estimation term; The spatial distance estimation term is determined based on the geometric distance between the current node and the target task node; the business attribute estimation term includes at least one of the following: risk area crossing estimation term, time window urgency estimation term, and facility access reachability estimation term; The risk area crossing estimation term is determined based on the high-risk pipe section area, construction sensitive area, or pressure zone boundary traversed by the passageway from the current node to the target task node; the time window urgency estimation term is determined based on the relationship between the expected arrival time from the current node to the target task node and the target task time window; and the facility access reachability estimation term is determined based on the access difficulty of the facility point corresponding to the target task node.

10. A shortest path planning system based on water pipeline network inspection, characterized in that, include: The data acquisition module is used to acquire road network data, pipeline facility data, work order task data, traffic and construction data, SCADA monitoring data, historical inspection trajectory data, and inspection personnel status data. The road network data includes the sides of passable roads. The fusion graph construction module is used to determine the access line segment and generate the access node on the accessible roadside for each facility point to be inspected, establish the virtual access edge between the facility point to be inspected and the access node, and form a fusion graph model containing facility nodes, road nodes and the virtual access edge. The comprehensive cost calculation module is used to calculate the comprehensive cost of each edge in the fusion graph model based on the fusion graph model, combined with the pipeline risk attributes of the pipeline facility data, the task business attributes of the work order data, the traffic status of the traffic and construction data, and the monitoring alarm status of the SCADA monitoring data. The task sorting module is used to cluster and sort multiple tasks to be inspected based on the work order task data and the comprehensive cost, and to determine the task access order. The path search module is used to generate inspection paths segment by segment based on the fusion graph model, the comprehensive cost, and the task access order, using a heuristic search algorithm that includes spatial distance estimation and business attribute estimation in the heuristic function.

Images