Inspection route optimization method for hydraulic power plant

Abstract

The invention belongs to the technical field of hydropower equipment inspection, and particularly provides an inspection route optimization method for a hydraulic power plant, which comprises the following steps: establishing an inspection item knowledge base; establishing a historical inspection route; generating an inspection point weight and a path weight; generating an inspection point weight and a path weight; acquiring a historical inspection task, and constructing a default inspection path according to the reserved path nodes and the inspection points; and obtaining the currently scheduled task content, obtaining the inspection points with the maximum similarity to generate an inspection point set, and matching the default inspection paths containing all the inspection points in the default inspection paths according to the inspection point set to serve as the task inspection path. Through the method, the inspection path can be optimized, and the inspection efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of hydropower equipment inspection technology, specifically, it relates to an inspection route optimization method for hydropower plants. Background Technology

[0002] The inspection of power equipment in hydropower stations is a crucial means of maintaining the stable operation of the power system. However, with the increasing demands for standardized data and reports, the need for multi-location offices, and stricter inspection supervision, current inspection methods are no longer sufficient to meet the current management and needs. They cannot be integrated with business data such as video and operational status during the inspection process, resulting in a limited range of inspection scenarios. Furthermore, fault information, start-up / shutdown events, and shift handover events rely heavily on manual processing, leading to a significant workload. Therefore, an automated, intelligent, and digitalized intelligent operation management system is needed to improve the efficiency of power equipment inspection and management.

[0003] CN116957535A proposes a knowledge-image-driven BIM emergency response automatic push method for hydropower projects. It establishes a knowledge graph by linking event names, causes, results, and locations through a hydropower project inspection report, creating a causal relationship. Semantic feature vectors are generated by extracting semantic features from the text and connecting them to the knowledge graph relationships. BILSTM is used to identify these vector features, obtaining causal scores for each event and its location. Based on these scores, corresponding inspection routes are established for each inspection. However, this technical solution only generates routes for unexpected and emergency situations during inspections. For routine inspections, it only considers the causal relationship of events, resulting in a limited range of factors and failing to dynamically schedule inspections based on real-time scenarios and inspection requirements.

[0004] CN113036906A proposes an automated inspection robot for a power dispatching system. By utilizing the power dispatching system's alarm window, power image displays, and front-end multi-channel images, the system extracts keywords from each image data using Tesseract-OCR to generate a set of points to be inspected. Based on this set of standby inspection points, the system generates inspection targets and notifies the robot to perform automated inspections. This technical solution can generate corresponding inspection targets using various information factors within the system; however, the robot has high maintenance costs and lacks proper planning for inspections, severely impacting work efficiency.

[0005] The current challenge in inspections lies in the difficulty for inspectors to dynamically plan and adjust inspection routes during the inspection process in order to improve the efficiency of power equipment inspection and management. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a method for optimizing inspection routes in hydropower plants, thereby optimizing the inspection path and improving inspection efficiency.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for optimizing inspection routes in hydropower plants, comprising the following steps: Step 1: Establish a knowledge base for inspection items; Step 2: Establish historical inspection routes: Establish inspection points based on the location of each device. Establish multiple historical inspection routes using the historical inspection items, historical inspection content, and inspection points in the inspection item knowledge base. Divide the historical inspection routes according to daily inspections and intensive inspections. Step 3: Generation of Inspection Point Weights and Path Weights: Based on the frequency of daily inspections and the frequency of intensive inspections, each inspection point is labeled with a frequency weight to obtain the frequency weights of the two types of inspection points: daily inspection points and intensive inspection points. For the historical inspection routes of daily inspections and intensive inspections, the corresponding path weights are obtained according to the frequency weights of the two types of inspection points. Step 4: Optimization of Inspection Point Weights and Path Weights: Obtain video surveillance equipment around each path node; determine whether the inspection points for routine and encrypted inspections are within the maximum monitoring range. If they are within the maximum monitoring range, add video weights to the inspection points; set the maximum safe distance and set the weights for dangerous inspection points based on hazardous targets around the equipment at each inspection point; optimize the inspection point weights based on the dangerous point weights and video weights to obtain the optimized inspection point weights; obtain the distance information between each inspection point and further optimize the path weights to obtain the initial path weights. Step 5: Default Inspection Path Generation: Obtain all path nodes of historical task routes, optimize the path nodes except for inspection points based on the monitoring range of video surveillance equipment and the danger range of danger points, obtain historical inspection tasks, and construct a default inspection path based on the retained path nodes and inspection points. Step 6: Generate inspection route: Obtain the currently scheduled task content, obtain the inspection points with the highest similarity to generate an inspection point set, and match the default inspection path containing all inspection points in the default inspection path as the task inspection path.

[0008] In the preferred embodiment, in step one, an inspection item knowledge base is established based on inspection items, inspection content, daily inspection frequency, encrypted inspection frequency, and encrypted inspection scenario content.

[0009] In the preferred embodiment, the formula for calculating the frequency weight in step three is: ; in, For the first i The first type of inspection n Frequency weight of each inspection point For the first iAverage frequency of inspection points for each type of inspection. For the first i The first type of inspection n Frequency of each inspection point; i =1 and 2 represent two types of inspections: routine inspection and frequent inspection.

[0010] In the preferred embodiment, the formula for calculating the path weight is: ; in, For the first i The first type of inspection n Each path weight For the first i The first type of inspection n +1 inspection point frequency weight.

[0011] In the preferred embodiment, step four, the process of adding video weights to the inspection points, is as follows: Get the video surveillance equipment around each path node and the maximum monitoring range of the video surveillance equipment. Determine whether the inspection points of daily inspection and encrypted inspection are within the maximum monitoring range. If they are within the maximum monitoring range, add video weights to the inspection points and associate them with video surveillance equipment. Generate the corresponding number matrix based on the video surveillance equipment number and the associated inspection point number. Based on the video surveillance equipment in the numbering matrix, image data information is obtained from different locations. The image data information includes images of inspection personnel and images of the inspection points.

[0012] In the preferred embodiment, in step four, the clarity data of the image data of the inspection personnel is used as the standard for the maximum monitoring range, specifically as follows: The YOLOv5 model is used to identify images of inspection personnel at different locations. The YOLOv5 model is used to determine whether the facial data of the inspection personnel can be matched with the inspection personnel information in the database. The maximum straight-line distance between the maximum location of the inspection personnel that can be matched and the video surveillance equipment is used as the maximum monitoring radius of the video surveillance equipment.

[0013] In the preferred embodiment, in step four, the optimized inspection point weights The expression is: ; In the formula: This is the video weight, set to 1 if it is within the maximum monitoring range, and 0 if it is outside the maximum monitoring range. This is the video weight score, with a default value of 1. Risk weight; Initial path weights The expression is: ; in, For the first i The (n+1)th optimized weight for each inspection type; For the first i The first type of inspection n Optimized weights; For the first i The first type of inspection n Path distance, This is the standard distance threshold.

[0014] In the preferred embodiment, in step five, the default weight of the path nodes of the historical inspection route, excluding the inspection points, is... Set to 5, the path nodes (excluding inspection points) are optimized based on the monitoring range of the video surveillance equipment and the danger range of the danger points. The expression is: ; in, Optimized path node weights; When the path nodes are other than the inspection points If the value is 0 or less than 0, then the path node is removed.

[0015] In the preferred embodiment, the process of constructing the default inspection path in step five is as follows: Obtain the weights of historical inspection tasks, extract semantic information based on the semantic feature model to match the same historical inspection tasks, generate multiple sets of historical inspection task weights, and use the minimum value of each historical inspection task weight set as the first default inspection path for each inspection task. Starting from the inspection point, a second default inspection path is generated based on the retained path nodes using a nonlinear regression algorithm. Based on the same historical inspection tasks, the corresponding inspection points are obtained. According to the path optimization algorithm, each inspection point is used as the starting point, and the path is optimized through the path nodes to generate multiple optimized paths. The minimum value of the optimized path is used as the third default inspection path. Compare the path weights of the first, second, and third default inspection paths for the same historical inspection task, and use the path weight with the smallest value as the default inspection path for each historical inspection task. If there is no historical inspection route, a path is generated based on a nonlinear regression algorithm. The path generated by the nonlinear regression algorithm is generated between two inspection points. Multiple path nodes that can be accessed by inspection personnel are pre-set, and the default inspection path is generated by connecting the various path nodes through the nonlinear regression algorithm.

[0016] In the preferred embodiment, in step six, if all default inspection paths cannot be matched, the default inspection path containing the most inspection points is selected as the first main path based on the matching results, and other default inspection paths that meet the inspection task content are selected to form an inspection path set. The inspection path with the smallest path weight in the inspection path set is generated as the second main path, and the first main path and the second main path are connected to generate the task path.

[0017] The present invention provides a method for optimizing inspection routes in hydropower plants, which has the following beneficial effects: 1. Inspection point weights and path weights can perform corresponding feature calculations for hazards, inspection requirements, and monitoring requirements encountered during actual inspections, enhancing the dimensionality of data features. They can also be used to add further feature dimensions as requirements increase, improving the robustness of inspection routes. In addition, these data, such as inspection point weights and path weights, can visualize subsequent route information, providing a data foundation for subsequent review, evaluation, and ranking work, and improving the efficiency of inspection review and route generation.

[0018] 2. During the inspection process, the system can establish personalized inspection points and routes based on the inspection reports of actual inspectors, thereby improving individual inspection efficiency. Simultaneously, based on the actual paths taken by inspectors during the inspection process, the system can provide corresponding feedback on the weight of path distances, replacing higher-weighted paths with lower-weighted paths to achieve dynamic path weight data updates and further improve inspection efficiency.

[0019] 3. Based on actual emergency tasks, temporary tasks, etc., and according to path weights, inspection point weights, and stored path data, special task paths are automatically generated according to the task content. Furthermore, new task paths can be generated based on the current location of the inspection personnel. These path generation methods based on inspection point weights and path weights can be used not only for path generation of daily tasks but also for path generation of emergency tasks. This enhances the usability of data and path methods and improves the efficiency of path calculation. Attached Figure Description

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of the present invention; Figure 2 Generate a flowchart for inspection points and path weights; Figure 3 Generate a flowchart for the default inspection path; Figure 4 Generate a flowchart for the task path. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0022] Example 1: A method for optimizing inspection routes in hydropower plants, such as Figure 1 As shown, it includes the following steps: Step 1: Establish a knowledge base for inspection items.

[0023] Establish an inspection item knowledge base based on inspection items, inspection content, daily inspection frequency, encrypted inspection frequency, and encrypted inspection scenario content.

[0024] Step 2: Establish historical inspection routes.

[0025] Inspection points are established based on the location of each device. Multiple historical inspection routes are established using the historical inspection items, historical inspection content, and inspection points in the inspection item knowledge base. These historical inspection routes are then divided into routine inspections and intensive inspections.

[0026] Step 3: Generate inspection point weights and path weights. like Figure 2 As shown, each inspection point is assigned a frequency weight based on the frequency of routine inspections and the frequency of intensive inspections. For historical inspection routes of both routine and intensive inspections, the frequency weights of the two types of inspection points are obtained according to their respective weights. An inspection point is any node on the historical inspection route where the inspector checks the equipment. The number of inspection points for the same piece of equipment being checked includes one or more inspection points.

[0027] The formula for calculating frequency weight is: ; in, For the first i The first type of inspection n Frequency weight of each inspection point For the first i Average frequency of inspection points for each type of inspection. For the first i The first type of inspection n Frequency of each inspection point; i =1 and 2 represent two types of inspections: routine inspection and frequent inspection.

[0028] For historical inspection routes of routine and intensive inspections, the corresponding path weights are obtained according to the frequency weights of the two types of inspection points.

[0029] The formula for calculating the path weight is: ; in, For the first i The first type of inspection n Each path weight For the first i The first type of inspection n +1 inspection point frequency weight.

[0030] Step 4: Optimize the weights of inspection points and paths.

[0031] The system acquires video surveillance equipment around each path node and determines the maximum monitoring range of the equipment. It then determines whether the inspection points for routine and intensive inspections are within the maximum monitoring range. If they are, the system adds video weights to the inspection points and associates them with video surveillance equipment. A corresponding number matrix is ​​generated based on the video surveillance equipment number and the associated inspection point number. Based on the video surveillance equipment in the number matrix, the system acquires image data information from different locations, including images of inspection personnel and images of the inspection point targets.

[0032] The image clarity of the inspection personnel is used as the standard for the maximum monitoring range, specifically: The YOLOv5 model is used to identify images of inspection personnel at different locations. The YOLOv5 model is used to determine whether the facial data of the inspection personnel can be matched with the inspection personnel information in the database. The maximum straight-line distance between the maximum location of the inspection personnel that can be matched and the video surveillance equipment is used as the maximum monitoring radius of the video surveillance equipment.

[0033] Based on the hazardous targets surrounding the equipment at each inspection point, a maximum safe distance is set, and a weight is assigned to each hazardous inspection point. Simultaneously, the system generates corresponding measurement point information for each inspection point; the equipment at each inspection point is linked to the system, allowing for real-time viewing of operational status and receiving equipment warnings, while also adding observation records.

[0034] The corresponding management system includes an inspection task dashboard, which is divided according to department and shift task type. It can realize task arrangement such as pending tasks, task execution tracking, encrypted inspection task scheduling, and video-assisted inspection. It also includes linkage push notifications of key events such as current residual defects of equipment, early warning information, accident information, and equipment operation status. All system devices can connect and transmit data via 5G signals. The system configuration information implemented by each device also includes video equipment connection, measurement point connection, and hazard point configuration, so as to provide inspectors with important information on the implementation of each inspection device.

[0035] The inspection point weights are optimized based on the hazard point weights and video weights to obtain the optimized inspection point weights; the distance information between each inspection point is obtained, the path weights are optimized a second time to obtain the initial path weights, and the initial path weights and inspection point weights are stored.

[0036] The optimized inspection point weights The expression is: ; In the formula: This is the video weight, set to 1 if it is within the maximum monitoring range, and 0 if it is outside the maximum monitoring range. This is the video weight score, which defaults to 1 but can be set manually. This is the hazard weight, which increases based on the distance between the inspector and the hazard point. Different hazard weight scores are set according to the maximum safe distance, with a default range of 0-5, a minimum of 0, and a maximum of 5. If new factors affecting inspection points are added, the weights of the inspection points will be adjusted accordingly. The expression is optimized to: ; In the formula, To extend the weighting function, if new influencing factors at inspection points are added as subsequent parameters, this can be achieved by defining... The specific functional form is used to incorporate these new factors into the calculation model of the inspection point weights.

[0037] Initial path weights The expression is: ; in, For the first i The (n+1)th optimized weight for each inspection type; For the first i The first type of inspection n Optimized weights; For the first i The first type of inspection n Path distance, The standard distance threshold is 10 meters by default, but you can set it yourself. The lower the weight, the better the path efficiency.

[0038] Step 5: Generate the default inspection path.

[0039] like Figure 3As shown, the historical inspection routes of historical inspection tasks are matched according to the initial path weight. Multiple path nodes are generated according to the historical inspection routes of the same historical inspection tasks, excluding inspection points. The path nodes excluding inspection points are optimized based on the monitoring range of the video surveillance equipment and the danger range of the danger points.

[0040] The default weight of path nodes in historical inspection routes, excluding inspection points, is... Set to 5, the path nodes (excluding inspection points) are optimized based on the monitoring range of the video surveillance equipment and the danger range of the danger points. The expression is: ; in, Optimized path node weights; When the path nodes are other than the inspection points If the value is 0 or less than 0, then the path node is removed.

[0041] Retrieve historical inspection tasks, and construct a default inspection path based on the retained path nodes and inspection points. The construction process is as follows: Obtain the weights of historical inspection tasks, extract semantic information based on the semantic feature model to match the same historical inspection tasks, and generate multiple sets of weights for historical inspection tasks. The semantic feature model can be Transformer, chatgpt, etc.

[0042] The minimum value of the weight set of each historical inspection task is used as the first default inspection path for each inspection task.

[0043] Starting from the inspection point, a second default inspection path is generated using a nonlinear regression algorithm based on the retained path nodes.

[0044] Based on the same historical inspection tasks, the corresponding inspection points are obtained. Each inspection point is used as the starting point according to the path optimization algorithm. The path optimization algorithm uses Dijkstra's algorithm, ant colony optimization algorithm, etc. to optimize the path through the path nodes and generate multiple optimized paths. The minimum value of the optimized path is used as the third default inspection path.

[0045] Compare the path weights of the first, second, and third default inspection paths for the same historical inspection task, and use the path weight with the smallest value as the default inspection path for each historical inspection task.

[0046] The path weights of the first, second, and third default inspection paths are calculated using steps three and four. Specifically, the distance weights of each inspection point along the inspection path are calculated and combined with the inspection point weights to generate the path weights.

[0047] In actual implementation, if there is no historical inspection route, the path is generated based on the nonlinear regression algorithm. The nonlinear regression algorithm generates the path between two inspection points, determines multiple path nodes that can be accessed by inspection personnel, and connects each path node through the nonlinear regression algorithm to generate the default inspection path.

[0048] Step 6: Generate inspection route.

[0049] like Figure 4 As shown, during the inspection process, the current scheduled task content is obtained, semantic information is extracted based on the semantic feature model to obtain the inspection points with the highest similarity and generate an inspection point set. Based on the inspection point set, the default inspection path containing all inspection points is matched as the current inspection path, including daily inspection paths and encrypted inspection paths, and the inspection route is automatically matched.

[0050] If no match can be found, a new inspection route will be generated: If all default inspection paths cannot be matched, the default inspection path with the most inspection points will be selected as the first main path based on the matching results, and other default inspection paths that meet the inspection task content will be selected to form an inspection path set. The inspection path with the smallest path weight in the inspection path set will be generated as the second main path. The first main path and the second main path will be connected to generate the task path and stored.

[0051] During personnel inspections, key inspection reminders are given based on the inspection information, and data at each inspection point is updated in real time. Inspection reports of deficiencies are fed back based on the inspection results; if a deficiency is reported, the corresponding inspection point information is recorded, the relevant inspection personnel are associated with it, and the deficiency report weight is increased in the inspection point's weighting. , The default setting is 5, but you can adjust the setting based on your specific needs and update the inspection point weights. The system updates the path weights and records the updated path weights for the corresponding inspection personnel via the app; it also records the missing inspection data into the database to facilitate subsequent path optimization and updates.

[0052] If a temporary task, emergency task, or early warning information occurs, semantic information is extracted based on the semantic feature model to obtain the corresponding inspection point. An emergency inspection route is generated according to the method of "generating a new inspection route if no match can be found". The location of the current inspection personnel is obtained through the mobile app and video surveillance equipment, and the location information of the nearest inspection point is obtained. A new inspection route is generated by combining the emergency inspection route.

[0053] Example 2: In this embodiment, an intelligent inspection management system for hydropower plants is provided to execute the inspection route optimization method in Embodiment 1.

[0054] This includes an inspection task dashboard, which is divided according to department and shift task type. It can arrange tasks such as pending tasks, task execution tracking, encrypted inspection task scheduling, and video-assisted inspections. It also includes linked push notifications of key events such as current equipment defects, early warning information, accident information, and equipment operation status. All system devices can connect and transmit data via 5G signals. The system configuration information implemented by each device also includes video equipment connection, measurement point connection, and hazard point configuration, so as to provide inspectors with important information on the implementation of each inspection device.

[0055] During the inspection, the mobile app connects to the system, enabling real-time viewing of task progress, reports, equipment information, and equipment defect registration to assist in the inspection; it also pushes key events, risk warning information, alarm alerts, and to-do items, and can initiate related tasks; the app also connects to the system and communicates with data via 5G signals.

[0056] During the inspection, the personnel's route is recorded based on the mobile app and video surveillance equipment. The personnel inspection path is generated based on the inspection target, and its distance weight is calculated. If the distance weight is less than the currently recorded weight, the path weight is updated and the default inspection path is optimized. If a new path node appears, it is stored and registered, and the node weight is calculated and optimized in the same way.

[0057] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing inspection routes in hydropower plants, characterized in that, Includes the following steps: Step 1: Establish a knowledge base for inspection items; Step 2: Establish historical inspection routes: Establish inspection points based on the location of each device. Establish multiple historical inspection routes using the historical inspection items, historical inspection content, and inspection points in the inspection item knowledge base. Divide the historical inspection routes according to daily inspections and intensive inspections. Step 3: Generation of Inspection Point Weights and Path Weights: Based on the frequency of daily inspections and the frequency of intensive inspections, each inspection point is labeled with a frequency weight to obtain the frequency weights of the two types of inspection points: daily inspection points and intensive inspection points. For the historical inspection routes of daily inspections and intensive inspections, the corresponding path weights are obtained according to the frequency weights of the two types of inspection points. Step 4: Optimize inspection point weight and path weight: Obtain video surveillance equipment around each path node, determine whether the inspection points of daily inspection and encrypted inspection are within the maximum monitoring range, and add video weight to the inspection points if they are within the maximum monitoring range. Based on the dangerous targets around the inspection point equipment, set the maximum safe distance and set the weight of the dangerous inspection point; The inspection point weights are optimized based on the hazard point weights and video weights to obtain the optimized inspection point weights. Obtain the distance information between each inspection point, optimize the path weights a second time, and obtain the initial path weights; Step 5: Default Inspection Path Generation: Obtain all path nodes of historical task routes, optimize the path nodes except for inspection points based on the monitoring range of video surveillance equipment and the danger range of danger points, obtain historical inspection tasks, and construct a default inspection path based on the retained path nodes and inspection points. Step 6: Generate inspection route: Obtain the currently scheduled task content, obtain the inspection points with the highest similarity to generate an inspection point set, and match the default inspection path containing all inspection points in the default inspection path as the task inspection path.

2. The method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step one, an inspection item knowledge base is established based on inspection items, inspection content, daily inspection frequency, encrypted inspection frequency, and encrypted inspection scenario content.

3. The method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step three, the formula for calculating the frequency weight is: ； in, For the first i The first type of inspection n Frequency weight of each inspection point For the first i Average frequency of inspection points for each type of inspection. For the first i The first type of inspection n Frequency of each inspection point; i =1 and 2 represent two types of inspections: routine inspection and frequent inspection.

4. The method for optimizing inspection routes in a hydropower plant according to claim 3, characterized in that, The formula for calculating the path weight is: ； in, For the first i The first type of inspection n Each path weight For the first i The first type of inspection n +1 inspection point frequency weight.

5. The method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step four, the process of adding video weights to the inspection points is as follows: Get the video surveillance equipment around each path node and the maximum monitoring range of the video surveillance equipment. Determine whether the inspection points of daily inspection and encrypted inspection are within the maximum monitoring range. If they are within the maximum monitoring range, add video weights to the inspection points and associate them with video surveillance equipment. Generate the corresponding number matrix based on the video surveillance equipment number and the associated inspection point number. Based on the video surveillance equipment in the numbering matrix, image data information is obtained from different locations. The image data information includes images of inspection personnel and images of the inspection points.

6. The method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step four, the clarity data of the image data of the inspection personnel is used as the standard for the maximum monitoring range, specifically: The YOLOv5 model is used to identify images of inspection personnel at different locations. The YOLOv5 model is used to determine whether the facial data of the inspection personnel can be matched with the inspection personnel information in the database. The maximum straight-line distance between the maximum location of the inspection personnel that can be matched and the video surveillance equipment is used as the maximum monitoring radius of the video surveillance equipment.

7. The method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step four, the optimized inspection point weights The expression is: ； In the formula: This is the video weight, set to 1 if it is within the maximum monitoring range, and 0 if it is outside the maximum monitoring range. This is the video weight score, with a default value of 1. Risk weight; Initial path weights The expression is: ； in, For the first i The (n+1)th optimized weight for each inspection type; For the first i The first type of inspection n Optimized weights; For the first i The first type of inspection n Path distance, This is the standard distance threshold.

8. The method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step five, the default weight of the path nodes of the historical inspection route, excluding inspection points, is... Set to 5, the path nodes (excluding inspection points) are optimized based on the monitoring range of the video surveillance equipment and the danger range of the danger points. The expression is: ； in, Optimized path node weights; When the path nodes are other than the inspection points If the value is 0 or less than 0, then the path node is removed.

9. A method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step five, the process of constructing the default inspection path is as follows: Obtain the weights of historical inspection tasks, extract semantic information based on the semantic feature model to match the same historical inspection tasks, generate multiple sets of historical inspection task weights, and use the minimum value of each historical inspection task weight set as the first default inspection path for each inspection task. Starting from the inspection point, a second default inspection path is generated based on the retained path nodes using a nonlinear regression algorithm. Based on the same historical inspection tasks, the corresponding inspection points are obtained. According to the path optimization algorithm, each inspection point is used as the starting point, and the path is optimized through the path nodes to generate multiple optimized paths. The minimum value of the optimized path is used as the third default inspection path. Compare the path weights of the first, second, and third default inspection paths for the same historical inspection task, and use the path weight with the smallest value as the default inspection path for each historical inspection task. If there is no historical inspection route, a path is generated based on a nonlinear regression algorithm. The path generated by the nonlinear regression algorithm is generated between two inspection points. Multiple path nodes that can be accessed by inspection personnel are pre-set, and the default inspection path is generated by connecting the various path nodes through the nonlinear regression algorithm.

10. A method for optimizing inspection routes in a hydropower plant according to claim 1, characterized in that, In step six, if all default inspection paths cannot be matched, the default inspection path with the most inspection points is selected as the first main path based on the matching results, and other default inspection paths that meet the inspection task content are selected to form an inspection path set. The inspection path with the smallest path weight in the inspection path set is generated as the second main path, and the first main path and the second main path are connected to generate the task path.

Images