A method and system for unmanned inspection task planning of a gas power plant

Abstract

This invention discloses a method and system for unmanned inspection tasks in gas-fired power plants, relating to the field of gas-fired power plant operation and maintenance technology. It addresses the problems of poor adaptability, low resource utilization, and slow fault response in existing planning methods. This invention, centered on "dynamic perception-collaborative planning-real-time adjustment," breaks away from the traditional separate approach. Its core is an improved ID-TPCO algorithm, combined with dynamic adjustment of task priorities, constructing a multi-objective optimization function and an improved fitness function. Adaptive genetic operations prevent premature convergence, and a dynamic adjustment mechanism adapts to complex scenarios. The system comprises seven modules, with the collaborative planning module at its core and incorporating the ID-TPCO algorithm. This invention significantly improves the intelligence and efficiency of inspections, reduces costs, ensures the safe operation of power plants, and demonstrates strong practicality and innovation.

Description

Technical Field

[0001] This invention relates to the field of operation and maintenance of gas-fired power plants, specifically to a method and system for planning unmanned inspection tasks in gas-fired power plants. Background Technology

[0002] As the core carrier of clean energy supply, the operating status of the internal equipment (such as gas turbines, steam turbines, transformers, pipelines, and valves) of gas-fired power plants directly determines the safety, stability, and reliability of power supply. Traditional gas-fired power plant inspections mainly rely on manual inspections, which suffer from problems such as large workload, high risk factor, large human error, low inspection efficiency, and non-standard data recording. Furthermore, it is difficult to achieve 24-hour all-weather inspections and cannot detect early potential equipment failures in a timely manner.

[0003] With the development of unmanned inspection technologies (inspection robots, drones, etc.), unmanned inspection is gradually replacing manual inspection and becoming the mainstream trend in the operation and maintenance of gas-fired power plants. However, current unmanned inspection task planning methods for gas-fired power plants still have many shortcomings: existing planning methods mostly adopt traditional fixed path planning algorithms (such as Dijkstra's algorithm, basic A* algorithm, etc.). Algorithms of this type can only achieve static path planning and cannot adapt to the dynamic changes in the operating status of gas-fired power plant equipment, the complexity of inspection scenarios (dense equipment, multiple obstacles, cross-regional access constraints), and the diversification of inspection tasks (routine inspection, fault inspection, special inspection). At the same time, existing methods do not consider the performance differences of inspection equipment, the priority differences of inspection tasks, and the load balancing of inspection resources, resulting in low utilization of inspection resources, many blind spots in inspection, untimely fault response, and even inspection redundancy or omissions.

[0004] Furthermore, existing unmanned inspection task planning methods mostly adopt a separate "path planning-task allocation" process, failing to achieve synergistic optimization between the two. This results in a disconnect between the planning results and actual inspection needs, making it difficult to balance inspection efficiency, quality, and cost. Therefore, developing an innovative unmanned inspection task planning method and system that can adapt to the dynamic scenarios of gas-fired power plants and meet multi-dimensional needs has become a pressing technical problem to be solved in the intelligent operation and maintenance of gas-fired power plants. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for planning unmanned inspection tasks in gas-fired power plants, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for planning unmanned inspection tasks in a gas-fired power plant, comprising the following steps: Step S1, Inspection Scene Modeling and Basic Information Collection: A three-dimensional spatial topology map of the inspection area is constructed using LiDAR and visual fusion perception technology. At the same time, basic information of the equipment to be inspected, performance parameters of the equipment to be inspected, and basic information of the inspection task are collected to form a basic information database. Step S2, Dynamic Adjustment of Inspection Task Priority: Based on the basic information collected in Step S1 and the real-time operating status parameters of the gas-fired power plant equipment, the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation method are used to calculate and sort the dynamic priority value of each inspection task. The formula for calculating the dynamic priority value is as follows: ,in For the first The dynamic priority value of each inspection task. For the first The weight of each primary evaluation indicator and , For the first The first inspection task was in the Scores under each primary evaluation indicator; Step S3: Improved dynamic task allocation and path planning collaborative optimization: Based on the basic information in step S1 and the dynamic value of task priority in step S2, the ID-TPCO algorithm is used to realize the collaborative linkage between inspection task allocation and path planning. Step S4: Inspection task execution and data feedback: The optimal planning scheme obtained in step S3 is sent to each inspection equipment to execute the inspection operation and provide real-time feedback of inspection data. The monitoring center analyzes the data and handles potential faults. Step S5: Planning Scheme Optimization and Iteration: Evaluate the optimization results of the ID-TPCO algorithm based on the inspection data, adjust the algorithm parameters, and realize the continuous iterative upgrading of the planning scheme.

[0007] Preferably, in step S1, the basic information of the equipment to be inspected includes equipment number, equipment type, installation location coordinates, equipment importance level, regular inspection cycle, inspection items and inspection standards; the performance parameters of the equipment to be inspected include equipment number, equipment type, maximum inspection speed, battery life, sensing range, load capacity and passable area restrictions; the basic information of the inspection task includes task number, task type, task initiation time, associated equipment, required completion time and initial priority value.

[0008] Preferably, in step S2, the primary evaluation indicators include equipment importance weight, equipment operating status abnormality degree, and task urgency degree, with the corresponding weights preset as follows: , , ; and The range of values ​​is .

[0009] Preferably, the specific implementation steps of step S3 are as follows: Step S3.1: Determine the collaborative optimization objective and constraints: a. Collaborative Optimization Objective: Construct a multi-objective optimization function that takes into account four core objectives. Dynamic weighting coefficients are used to adjust the emphasis of the objectives in different scenarios. The optimization function is as follows: , The smaller the value of the overall objective function, the better the planning effect. , , , Dynamic weighting coefficients Dynamically adjust and set up routine inspection scenarios based on the inspection scenarios: , , , Fault inspection scenario: , , , ; The total inspection time for all inspection equipment reflects the inspection efficiency. , For the first Total inspection time for each inspection device; Inspection of equipment load balance reflects resource utilization. , The average inspection time for all inspection equipment. The smaller the value, the more balanced the load. The total length of the inspection route reflects the inspection cost. , For the first Total length of the inspection path for each inspection device; Inspection task delay rate reflects the timeliness of task completion. Delayed completion refers to a task being completed in a time that exceeds the required completion time. b. Constraints: Based on the inspection scenario and equipment performance of the gas-fired power plant, four types of constraints are set to ensure the feasibility of the planning results: Task allocation constraints: Each inspection task is assigned to only one suitable inspection device (e.g., high-altitude equipment is assigned only to drones, and equipment for confined indoor areas is assigned only to rail-mounted inspection robots), and high-priority tasks... Prioritize allocation and ensure no one is overlooked; Equipment performance constraints: The total inspection time of the inspection equipment must not exceed its operating time, the inspection path must not exceed its passable area, and the inspection speed must not exceed its maximum inspection speed. , , ,in , , The first The maximum battery life, maximum inspectable path length, and maximum inspection speed of each inspection device; Scenario constraints: The inspection path must avoid fixed obstacles (such as walls and equipment brackets) and temporary obstacles (such as construction areas and faulty equipment) within the inspection area. The spacing between path nodes must meet the sensing range requirements of the inspection equipment to ensure that there are no blind spots in the inspection. Time constraints: The actual completion time of all inspection tasks must not exceed their required completion time (the constraints may be appropriately relaxed for high-priority tasks, but the delay rate must not exceed 5%). Step S3.2, Initialize Parameters and Population: Based on the basic information from Step S1 and the task priority ranking from Step S2, initialize the relevant parameters and population of the ID-TPCO algorithm, specifically including: Algorithm parameters: Population size (Can be adjusted according to planning complexity), maximum number of iterations Crossover probability Probability of mutation Initial value of dynamic weight coefficient (set according to the current inspection scenario); Population initialization: Real-number encoding is used, with each individual corresponding to a set of task assignment-path planning combinations. The individual encoding length is [length missing]. ,in This represents the total number of inspection tasks. This represents the total number of equipment to be inspected. Number of path nodes for each inspection device; the first part of the code. The bit represents the task allocation scheme, i.e., the th bit. The value is 0. , representing the The task is assigned to the first Each inspection device, after The bit represents the path planning scheme, that is, each group The bit represents the coordinates of the path node of one inspection device; Initial population screening: Based on the constraints, the randomly generated initial population is screened to remove individuals that do not meet the constraints (such as unreasonable task allocation or paths exceeding the passable area) to ensure the feasibility of the initial population. Step 3.3, Improved Fitness Function Calculation: To avoid premature convergence and improve optimization accuracy, an improved fitness function is designed, combining the collaborative optimization objective function with constraint penalty terms. The larger the fitness function value, the better the individual (planning scheme). The calculation formula is as follows: ,in : No. The fitness function value of each individual; : No. The collaborative optimization objective function value of each individual; : Penalty coefficient, with a value of 10 (which can be adjusted according to the strictness of the constraints), used to penalize individuals who do not meet the constraints; : No. The constraint violation degree of each individual is calculated using the following formula: , For the first Individuals violated Article The degree of class constraint (when the constraint is satisfied, When the constraints are violated, The value is a quantitative measure of the degree of violation; the more severe the violation, the larger the value.

[0010] For example, if in a certain entity (planning scheme), the inspection time of a certain inspection device exceeds its endurance by 1 hour (violating equipment performance constraints), then All other constraints are satisfied. Then the degree of constraint violation By reducing the fitness function value of an individual through a penalty term, the algorithm ensures that high-quality individuals that meet the constraints are selected first. Step S3.4, Improved Genetic Operations: Traditional genetic algorithms' crossover and mutation operations easily lead to a decrease in population diversity, premature convergence, and failure to find the global optimum. This invention improves genetic operations by designing an adaptive crossover and mutation mechanism, combining inspection task priority and equipment performance to improve the algorithm's optimization accuracy and convergence speed.

[0011] Adaptive crossover operation: A two-point crossover method is used, with the crossover probability dynamically adjusted based on the individual fitness function value. The higher the fitness function value (the better the individual), the lower the crossover probability, avoiding the destruction of high-quality individuals; the lower the fitness function value (the worse the individual), the higher the crossover probability, accelerating the elimination and optimization of inferior individuals. Subsequently, a single-point mutation method is used, with the mutation probability also dynamically adjusted based on the individual fitness function value. The higher the fitness function value, the lower the mutation probability; the lower the fitness function value, the higher the mutation probability. At the same time, combined with the performance constraints of the inspection equipment, the mutation range is limited to avoid individuals violating the constraints after mutation. Step S3.5, Population Iterative Update and Optimal Solution Selection: Population iteration: Based on improved genetic operations, the current population is crossovered and mutated to generate a new generation of population; at the same time, an elite retention strategy is adopted to directly retain the top 10% of high-quality individuals in the current population with the highest fitness function value into the new generation of population, so as to avoid the loss of high-quality individuals and accelerate the convergence of the algorithm.

[0012] Convergence Criterion: Calculate the average fitness function value and the maximum fitness function value of the new generation population. If one of the following two conditions is met, the iteration terminates, and the individual with the largest fitness function value in the current population is output as the optimal task assignment-path planning combination scheme: The maximum number of iterations has been reached. ; The change in the maximum fitness function value over 10 consecutive iterations is less than 10⁻ 4 ,Right now ,in For the first The maximum fitness function value in the next iteration; Optimal solution verification: Verify the constraints of the selected optimal planning scheme to ensure that the scheme meets all constraints; if not, return to step S3.4, adjust the genetic operation parameters, and iterate again until the optimal scheme that meets the constraints is obtained. Step 3.6, Dynamic Adjustment Mechanism: To adapt to the dynamic changes in gas-fired power plant inspection scenarios (such as the temporary appearance of obstacles, sudden changes in equipment operating status, and the addition of new inspection tasks), a dynamic adjustment mechanism is designed to update the planning scheme in real time, specifically including: Dynamic scene adjustment: By monitoring the changes in obstacles and equipment status in the inspection area through real-time sensing data of the inspection equipment, if a temporary obstacle or sudden change in equipment status is found (such as a new fault point), the path replanning is immediately triggered to adjust the path nodes of the affected inspection equipment to avoid obstacles or add fault inspection tasks. Dynamic task adjustment: If a new inspection task (such as a special inspection) is added or an existing task is canceled, the priority dynamic value of all tasks is recalculated, triggering a re-optimization of task allocation and path planning to ensure the reasonable allocation of inspection resources. Dynamic equipment adjustment: If a certain inspection equipment malfunctions (such as insufficient battery life or abnormal sensing), the unfinished tasks assigned to that equipment will be reassigned to other suitable inspection equipment, and the route will be replanned to ensure that the inspection tasks are completed on time.

[0013] Preferably, the crossover probability calculation formula in step S3.4 is as follows: The formula for calculating the mutation probability is: in , These are the maximum and minimum crossover probabilities, respectively. , These are the maximum and minimum mutation probabilities, respectively. This represents the current maximum fitness function value of the population. This represents the current average fitness function value of the population.

[0014] Preferably, an unmanned inspection task planning system for a gas-fired power plant is characterized by comprising a basic perception module, a data storage module, a task priority adjustment module, a collaborative planning module, a task execution module, a monitoring feedback module, and an iterative optimization module. The collaborative planning module incorporates an improved dynamic task allocation and path planning collaborative optimization algorithm, namely the ID-TPCO algorithm, used to achieve collaborative optimization and dynamic adjustment of inspection task allocation and path planning. The basic perception module consists of a lidar, a visual camera, various sensors, and a wireless communication module, used for inspection scene modeling, basic information collection, and real-time status perception. The data storage module uses a distributed database to store all basic data and inspection-related data. The task priority adjustment module calculates and sorts the dynamic values ​​of inspection task priorities. The task execution module consists of various inspection devices used to execute inspection operations and provide data feedback. The monitoring feedback module monitors the inspection process in real time, analyzes data, and handles faults. The iterative optimization module adjusts the ID-TPCO algorithm parameters to achieve iterative upgrades of the planning scheme.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention designs an improved dynamic task allocation and path planning collaborative optimization algorithm (ID-TPCO algorithm), which integrates task allocation and path planning into a unified optimization objective. It introduces dynamic weight coefficients and adaptive genetic operations to avoid premature convergence of the algorithm and solves the problems of poor adaptability, uneven resource load, and slow fault response of traditional separate planning and basic collaborative planning. At the same time, a dynamic adjustment mechanism is designed to realize the real-time updating of the planning scheme and adapt to the needs of dynamic inspection scenarios of gas-fired power plants.

[0016] Significantly improved inspection efficiency and quality: This invention dynamically adjusts task priorities to ensure that fault-related and high-risk tasks are executed first, reducing the rate of missed fault detections; through collaborative optimization algorithms, it achieves load balancing of inspection resources, improves the utilization rate of inspection resources, shortens the total inspection time, avoids path redundancy and task omissions, and ensures that inspection tasks are completed on time and with high quality.

[0017] Highly adaptable and practical: This invention adopts 3D scene modeling and real-time perception technology to adapt to the inspection needs of multiple scenarios and equipment such as boiler rooms, turbine rooms, and power distribution rooms in gas-fired power plants; the dynamic adjustment mechanism can cope with emergencies such as temporary obstacles, sudden changes in equipment status, and new inspection tasks; the algorithm parameters can be adjusted according to the actual operation and maintenance needs of the power plant, without the need for large-scale modification of existing inspection equipment, making it easy to promote and apply.

[0018] Reduce inspection costs and ensure power plant safety: This invention achieves full automation of unmanned inspection, reducing the workload and safety risks of manual inspection, and lowering labor and equipment maintenance costs; through real-time fault detection and alarm, it can promptly capture early potential equipment faults, prevent faults from escalating, and ensure the safe and stable operation of gas-fired power plants. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the method flow structure of the present invention; Figure 2 This is a schematic diagram of the improved dynamic task allocation and path planning collaborative optimization workflow of the present invention; Figure 3 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1-3 This invention provides a technical solution: a method for planning unmanned inspection tasks in a gas-fired power plant, comprising the following steps: Step S1, Inspection Scene Modeling and Basic Information Collection: A three-dimensional spatial topology map of the inspection area is constructed using LiDAR and visual fusion perception technology. At the same time, basic information of the equipment to be inspected, performance parameters of the equipment to be inspected, and basic information of the inspection task are collected to form a basic information database. Step S2, Dynamic Adjustment of Inspection Task Priority: Based on the basic information collected in Step S1 and the real-time operating status parameters of the gas-fired power plant equipment, the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation method are used to calculate and sort the dynamic priority value of each inspection task. The formula for calculating the dynamic priority value is as follows: ,in For the first The dynamic priority value of each inspection task. For the first The weight of each primary evaluation indicator and , For the first The first inspection task was in the Scores under each primary evaluation indicator; Step S3: Improved dynamic task allocation and path planning collaborative optimization: Based on the basic information in step S1 and the dynamic value of task priority in step S2, the ID-TPCO algorithm is used to realize the collaborative linkage between inspection task allocation and path planning. Step S4: Inspection Task Execution and Data Feedback: The optimal "task allocation-path planning" combination scheme obtained in Step S3 is distributed to each inspection device (inspection robot, drone). Each inspection device performs unmanned inspection operations according to the planned path and assigned tasks. During the inspection, the operating status data (temperature, pressure, vibration, gas concentration, etc.) and inspection path data of the equipment to be inspected are collected in real time and fed back to the monitoring center through the wireless communication module.

[0022] The monitoring center analyzes the feedback data in real time to determine whether there are potential faults in the equipment to be inspected. If there are potential faults, the alarm mechanism is immediately triggered, and a fault handling instruction is generated and sent to the relevant maintenance personnel and the equipment to be inspected, and a special re-inspection is arranged. If there are no potential faults, the inspection data is recorded, an inspection report is generated, and the inspection task is completed. Step S5: Planning Scheme Optimization and Iteration: After each inspection task is completed, the monitoring center collects relevant data during the inspection process (inspection task completion status, inspection time, path redundancy, resource load, fault detection accuracy, etc.) and evaluates the optimization results of the ID-TPCO algorithm. Based on the evaluation results, the relevant parameters of the algorithm are adjusted (such as dynamic weight coefficient, population size, crossover probability, mutation probability), and the co-optimization objective function and fitness function are optimized to achieve continuous iterative upgrading of the planning scheme and improve the rationality and adaptability of subsequent inspection task planning.

[0023] Furthermore, in step S1, the basic information of the equipment to be inspected includes the equipment number, equipment type, installation location coordinates, equipment importance level, regular inspection cycle, inspection items, and inspection standards; the performance parameters of the equipment to be inspected include the equipment number, equipment type, maximum inspection speed, battery life, sensing range, load capacity, and traversable area restrictions; the basic information of the inspection task includes the task number, task type, task initiation time, associated equipment, required completion time, and initial priority value.

[0024] Furthermore, in step S2, the primary evaluation indicators include the weight of equipment importance, the degree of abnormality in equipment operating status, and the degree of urgency of the task, with the corresponding weights preset as follows: , , ; and The range of values ​​is Furthermore, the specific implementation steps of step S3 are as follows: Step S3.1: Determine the collaborative optimization objective and constraints: a. Collaborative Optimization Objective: Construct a multi-objective optimization function that takes into account four core objectives. Dynamic weighting coefficients are used to adjust the emphasis of the objectives in different scenarios. The optimization function is as follows: , The smaller the value of the overall objective function, the better the planning effect. , , , Dynamic weighting coefficients Dynamically adjust and set up routine inspection scenarios based on the inspection scenarios: , , , Fault inspection scenario: , , , ; The total inspection time for all inspection equipment reflects the inspection efficiency. , For the first Total inspection time for each inspection device; Inspection of equipment load balance reflects resource utilization. , The average inspection time for all inspection equipment. The smaller the value, the more balanced the load. The total length of the inspection route reflects the inspection cost. , For the first Total length of the inspection path for each inspection device; Inspection task delay rate reflects the timeliness of task completion. Delayed completion refers to a task being completed in a time that exceeds the required completion time. b. Constraints: Based on the inspection scenario and equipment performance of the gas-fired power plant, four types of constraints are set to ensure the feasibility of the planning results: Task allocation constraints: Each inspection task is assigned to only one suitable inspection device (e.g., high-altitude equipment is assigned only to drones, and equipment for confined indoor areas is assigned only to rail-mounted inspection robots), and high-priority tasks... Prioritize allocation and ensure no one is overlooked; Equipment performance constraints: The total inspection time of the inspection equipment must not exceed its operating time, the inspection path must not exceed its passable area, and the inspection speed must not exceed its maximum inspection speed. , , ,in , , The first The maximum battery life, maximum inspectable path length, and maximum inspection speed of each inspection device; Scenario constraints: The inspection path must avoid fixed obstacles (such as walls and equipment brackets) and temporary obstacles (such as construction areas and faulty equipment) within the inspection area. The spacing between path nodes must meet the sensing range requirements of the inspection equipment to ensure that there are no blind spots in the inspection. Time constraints: The actual completion time of all inspection tasks must not exceed their required completion time (the constraints may be appropriately relaxed for high-priority tasks, but the delay rate must not exceed 5%). Step S3.2, Initialize Parameters and Population: Based on the basic information from Step S1 and the task priority ranking from Step S2, initialize the relevant parameters and population of the ID-TPCO algorithm, specifically including: Algorithm parameters: Population size (Can be adjusted according to planning complexity), maximum number of iterations Crossover probability Probability of mutation Initial value of dynamic weight coefficient (set according to the current inspection scenario); Population initialization: Real-number encoding is used, with each individual corresponding to a set of task assignment-path planning combinations. The individual encoding length is [length missing]. ,in This represents the total number of inspection tasks. This represents the total number of equipment to be inspected. Number of path nodes for each inspection device; the first part of the code. The bit represents the task allocation scheme, i.e., the th bit. The value is 0. , representing the The task is assigned to the first Each inspection device, after The bit represents the path planning scheme, that is, each group The bit represents the coordinates of the path node of one inspection device; Initial population screening: Based on the constraints, the randomly generated initial population is screened to remove individuals that do not meet the constraints (such as unreasonable task allocation or paths exceeding the passable area) to ensure the feasibility of the initial population. Step 3.3, Improved Fitness Function Calculation: To avoid premature convergence and improve optimization accuracy, an improved fitness function is designed, combining the collaborative optimization objective function with constraint penalty terms. The larger the fitness function value, the better the individual (planning scheme). The calculation formula is as follows: ,in : No. The fitness function value of each individual; : No. The collaborative optimization objective function value of each individual; : Penalty coefficient, with a value of 10 (which can be adjusted according to the strictness of the constraints), used to penalize individuals who do not meet the constraints; : No. The constraint violation degree of each individual is calculated using the following formula: , For the first Individuals violated Article The degree of class constraint (when the constraint is satisfied, When the constraints are violated, The value is a quantitative measure of the degree of violation; the more severe the violation, the larger the value.

[0025] For example, if in a certain entity (planning scheme), the inspection time of a certain inspection device exceeds its endurance by 1 hour (violating equipment performance constraints), then All other constraints are satisfied. Then the degree of constraint violation By reducing the fitness function value of an individual through a penalty term, the algorithm ensures that high-quality individuals that meet the constraints are selected first. Step S3.4, Improved Genetic Operations: Traditional genetic algorithms' crossover and mutation operations easily lead to a decrease in population diversity, premature convergence, and failure to find the global optimum. This invention improves genetic operations by designing an adaptive crossover and mutation mechanism, combining inspection task priority and equipment performance to improve the algorithm's optimization accuracy and convergence speed.

[0026] Adaptive crossover operation: A two-point crossover method is used, with the crossover probability dynamically adjusted based on the individual fitness function value. The higher the fitness function value (the better the individual), the lower the crossover probability, avoiding the destruction of high-quality individuals; the lower the fitness function value (the worse the individual), the higher the crossover probability, accelerating the elimination and optimization of inferior individuals. Subsequently, a single-point mutation method is used, with the mutation probability also dynamically adjusted based on the individual fitness function value. The higher the fitness function value, the lower the mutation probability; the lower the fitness function value, the higher the mutation probability. At the same time, combined with the performance constraints of the inspection equipment, the mutation range is limited to avoid individuals violating the constraints after mutation. Step S3.5, Population Iterative Update and Optimal Solution Selection: Population iteration: Based on improved genetic operations, the current population is crossovered and mutated to generate a new generation of population; at the same time, an elite retention strategy is adopted to directly retain the top 10% of high-quality individuals in the current population with the highest fitness function value into the new generation of population, so as to avoid the loss of high-quality individuals and accelerate the convergence of the algorithm.

[0027] Convergence Criterion: Calculate the average fitness function value and the maximum fitness function value of the new generation population. If one of the following two conditions is met, the iteration terminates, and the individual with the largest fitness function value in the current population is output as the optimal task assignment-path planning combination scheme: The maximum number of iterations has been reached. ; The change in the maximum fitness function value over 10 consecutive iterations is less than 10⁻ 4 ,Right now ,in For the first The maximum fitness function value in the next iteration; Optimal solution verification: Verify the constraints of the selected optimal planning scheme to ensure that the scheme meets all constraints; if not, return to step S3.4, adjust the genetic operation parameters, and iterate again until the optimal scheme that meets the constraints is obtained. Step 3.6, Dynamic Adjustment Mechanism: To adapt to the dynamic changes in gas-fired power plant inspection scenarios (such as the temporary appearance of obstacles, sudden changes in equipment operating status, and the addition of new inspection tasks), a dynamic adjustment mechanism is designed to update the planning scheme in real time, specifically including: Dynamic scene adjustment: By monitoring the changes in obstacles and equipment status in the inspection area through real-time sensing data of the inspection equipment, if a temporary obstacle or sudden change in equipment status is found (such as a new fault point), the path replanning is immediately triggered to adjust the path nodes of the affected inspection equipment to avoid obstacles or add fault inspection tasks. Dynamic task adjustment: If a new inspection task (such as a special inspection) is added or an existing task is canceled, the priority dynamic value of all tasks is recalculated, triggering a re-optimization of task allocation and path planning to ensure the reasonable allocation of inspection resources. Dynamic equipment adjustment: If a certain inspection device malfunctions (such as insufficient battery life or sensor abnormality), the unfinished tasks assigned to that device will be reassigned to other suitable inspection devices, and the route will be replanned to ensure that the inspection tasks are completed on time. Furthermore, the formula for calculating the crossover probability in step S3.4 is as follows: The formula for calculating the mutation probability is: in , These are the maximum and minimum crossover probabilities, respectively. , These are the maximum and minimum mutation probabilities, respectively. This represents the current maximum fitness function value of the population. The average fitness function value of the current population Furthermore, an unmanned inspection task planning system for a gas-fired power plant includes a basic perception module, a data storage module, a task priority adjustment module, a collaborative planning module, a task execution module, a monitoring feedback module, and an iterative optimization module. The collaborative planning module incorporates an improved dynamic task allocation and path planning collaborative optimization algorithm, namely the ID-TPCO algorithm, to achieve collaborative optimization and dynamic adjustment of inspection task allocation and path planning. The basic perception module consists of LiDAR, a visual camera, various sensors, and a wireless communication module, used for inspection scene modeling, basic information collection, and real-time status perception. The data storage module uses a distributed database to store all basic data and inspection-related data. The task priority adjustment module calculates and sorts the dynamic values ​​of inspection task priorities. The task execution module consists of various inspection devices to perform inspection operations and provide data feedback. The monitoring feedback module monitors the inspection process in real time, analyzes data, and handles faults. The iterative optimization module adjusts the ID-TPCO algorithm parameters to achieve iterative upgrades of the planning scheme.

[0028] This invention discloses a method and system for unmanned inspection task planning in gas-fired power plants, relating to the field of gas-fired power plant operation and maintenance technology. It aims to solve the technical problems of poor adaptability, low resource utilization, slow fault response, and disconnect between planning and actual needs in existing unmanned inspection task planning methods. This invention uses "dynamic perception-collaborative planning-real-time adjustment" as its core process, breaking the traditional separate "path planning-task allocation" model. The core lies in an improved dynamic task allocation and path planning collaborative optimization algorithm (ID-TPCO algorithm), which, combined with dynamic adjustment of inspection task priorities and dynamic scene perception, achieves collaborative linkage between inspection task allocation and path planning. By constructing a multi-objective collaborative optimization function, an improved fitness function, and adaptive genetic operations, premature convergence of the algorithm is avoided, balancing inspection efficiency, resource load balancing, inspection cost, and task completion timeliness. Simultaneously, a dynamic adjustment mechanism is designed to adapt to the complex dynamic scenarios of gas-fired power plants. The system of this invention includes seven modules, including basic perception and collaborative planning, with the collaborative planning module being the core, incorporating the ID-TPCO algorithm. This invention significantly improves the intelligence level, efficiency, and quality of unmanned inspections in gas-fired power plants, reduces inspection costs, and ensures the safe and stable operation of power plants. It has strong practicality and inventiveness. Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for planning unmanned inspection tasks in a gas-fired power plant, characterized in that, Includes the following steps: Step S1, Inspection Scene Modeling and Basic Information Collection: A three-dimensional spatial topology map of the inspection area is constructed using LiDAR and visual fusion perception technology. At the same time, basic information of the equipment to be inspected, performance parameters of the equipment to be inspected, and basic information of the inspection task are collected to form a basic information database. Step S2, Dynamic Adjustment of Inspection Task Priority: Based on the basic information collected in Step S1 and the real-time operating status parameters of the gas-fired power plant equipment, the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation method are used to calculate and sort the dynamic priority value of each inspection task. The formula for calculating the dynamic priority value is as follows: ,in For the first The dynamic priority value of each inspection task. For the first The weight of each primary evaluation indicator and , For the first The first inspection task was in the Scores under each primary evaluation indicator; Step S3: Improved dynamic task allocation and path planning collaborative optimization: Based on the basic information in step S1 and the dynamic value of task priority in step S2, the ID-TPCO algorithm is used to realize the collaborative linkage between inspection task allocation and path planning. Step S4: Inspection task execution and data feedback: The optimal planning scheme obtained in step S3 is sent to each inspection equipment to execute the inspection operation and provide real-time feedback of inspection data. The monitoring center analyzes the data and handles potential faults. Step S5: Planning Scheme Optimization and Iteration: Evaluate the optimization results of the ID-TPCO algorithm based on the inspection data, adjust the algorithm parameters, and realize the continuous iterative upgrading of the planning scheme.

2. The method for planning unmanned inspection tasks in a gas-fired power plant according to claim 1, characterized in that: In step S1, the basic information of the equipment to be inspected includes equipment number, equipment type, installation location coordinates, equipment importance level, regular inspection cycle, inspection items and inspection standards; the performance parameters of the equipment to be inspected include equipment number, equipment type, maximum inspection speed, battery life, sensing range, load capacity and passable area restrictions; the basic information of the inspection task includes task number, task type, task initiation time, associated equipment, task completion time requirement and initial task priority value.

3. The method for planning unmanned inspection tasks in a gas-fired power plant according to claim 1, characterized in that: In step S2, the primary evaluation indicators include equipment importance weight, equipment operating status abnormality degree, and task urgency degree, with corresponding weights preset as follows: , , ; and The range of values ​​is .

4. The method for planning unmanned inspection tasks in a gas-fired power plant according to claim 1, characterized in that: The specific implementation steps of step S3 are as follows: Step S3.1: Determine the collaborative optimization objective and constraints: a. Collaborative Optimization Objective: Construct a multi-objective optimization function that takes into account four core objectives. Dynamic weighting coefficients are used to adjust the emphasis of the objectives in different scenarios. The optimization function is as follows: , The smaller the value of the overall objective function, the better the planning effect. , , , Dynamic weighting coefficients Dynamically adjust and set up routine inspection scenarios based on the inspection scenarios: , , , Fault inspection scenario: , , , ; The total inspection time for all inspection equipment reflects the inspection efficiency. , For the first Total inspection time for each inspection device; Inspection of equipment load balance reflects resource utilization. , The average inspection time for all inspection equipment. The smaller the value, the more balanced the load. The total length of the inspection route reflects the inspection cost. , For the first Total length of the inspection path for each inspection device; Inspection task delay rate reflects the timeliness of task completion. Delayed completion refers to a task being completed in a time that exceeds the required completion time. b. Constraints: Based on the inspection scenario and equipment performance of the gas-fired power plant, four types of constraints are set to ensure the feasibility of the planning results: Task allocation constraints: Each inspection task is assigned to only one compatible inspection device, and high-priority tasks... Prioritize allocation and ensure no one is overlooked; Equipment performance constraints: The total inspection time of the inspection equipment must not exceed its operating time, the inspection path must not exceed its passable area, and the inspection speed must not exceed its maximum inspection speed. , , ,in , , The first The maximum battery life, maximum inspectable path length, and maximum inspection speed of each inspection device; Scenario constraints: The inspection path must avoid fixed and temporary obstacles within the inspection area, and the spacing between path nodes must meet the sensing range requirements of the inspection equipment to ensure no blind spots in the inspection. Time constraint: The actual completion time of all inspection tasks must not exceed the required completion time; Step S3.2, Initialize Parameters and Population: Based on the basic information from Step S1 and the task priority ranking from Step S2, initialize the relevant parameters and population of the ID-TPCO algorithm, specifically including: Algorithm parameters: Population size Maximum number of iterations Crossover probability Probability of mutation Initial values ​​of dynamic weighting coefficients; Population initialization: Real-number encoding is used, with each individual corresponding to a set of task assignment-path planning combinations. The individual encoding length is [length missing]. ,in This represents the total number of inspection tasks. This represents the total number of equipment to be inspected. Number of path nodes for each inspection device; the first part of the code. The bit represents the task allocation scheme, i.e., the th bit. The value is 0. , representing the The task is assigned to the first Each inspection device, after The bit represents the path planning scheme, that is, each group The bit represents the coordinates of the path node of one inspection device; Initial population screening: Based on the constraints, the randomly generated initial population is screened to remove individuals that do not meet the constraints, thus ensuring the feasibility of the initial population. Step 3.3, Improved Fitness Function Calculation: To avoid premature convergence and improve optimization accuracy, an improved fitness function is designed, combining the collaborative optimization objective function with constraint penalty terms. The larger the fitness function value, the better the individual. The calculation formula is as follows: ,in : No. The fitness function value of each individual; : No. The collaborative optimization objective function value of each individual; : Penalty coefficient, with a value of 10, used to penalize individuals who do not meet the constraints; : No. The constraint violation degree of each individual is calculated using the following formula: , For the first Individuals violated Article The degree of class constraints; Step S3.4, Improved genetic operation: Adaptive crossover operation: A two-point crossover method is adopted, and the crossover probability is dynamically adjusted according to the individual fitness function value. The higher the fitness function value, the lower the crossover probability, avoiding the destruction of high-quality individuals; the lower the fitness function value, the higher the crossover probability, accelerating the elimination and optimization of low-quality individuals. Subsequently, a single-point mutation method is adopted, and the mutation probability is also dynamically adjusted according to the individual fitness function value. The higher the fitness function value, the lower the mutation probability; the lower the fitness function value, the higher the mutation probability. At the same time, combined with the performance constraints of the inspection equipment, the mutation range is limited to avoid individuals violating the constraints after mutation. Step S3.5, Population Iterative Update and Optimal Solution Selection: Population iteration: Based on improved genetic operations, the current population is crossovered and mutated to generate a new generation of population; at the same time, an elite retention strategy is adopted to directly retain the top 10% of high-quality individuals in the current population with the highest fitness function value into the new generation of population, so as to avoid the loss of high-quality individuals and accelerate the convergence of the algorithm. Convergence Criterion: Calculate the average fitness function value and the maximum fitness function value of the new generation population. If one of the following two conditions is met, the iteration terminates, and the individual with the largest fitness function value in the current population is output as the optimal task assignment-path planning combination scheme: The maximum number of iterations has been reached. ; The change in the maximum fitness function value over 10 consecutive iterations is less than 10⁻ 4 ,Right now ,in For the first The maximum fitness function value in the next iteration; Optimal solution verification: Verify the constraints of the selected optimal planning scheme to ensure that the scheme meets all constraints; if not, return to step S3.4, adjust the genetic operation parameters, and iterate again until the optimal scheme that meets the constraints is obtained. Step 3.6, Dynamic Adjustment Mechanism: To adapt to the dynamic changes in gas-fired power plant inspection scenarios, a dynamic adjustment mechanism is designed to update the planning scheme in real time, specifically including: Dynamic scene adjustment: By monitoring the changes in obstacles and equipment status in the inspection area through real-time sensing data of the inspection equipment, if a temporary obstacle or sudden change in equipment status is found, the path replanning is immediately triggered to adjust the path nodes of the affected inspection equipment to avoid obstacles or add fault inspection tasks. Dynamic task adjustment: If a new inspection task is added or an existing task is canceled, the priority dynamic value of all tasks is recalculated, triggering a re-optimization of task allocation and path planning to ensure the reasonable allocation of inspection resources. Dynamic equipment adjustment: If a certain inspection equipment malfunctions, the unfinished tasks assigned to that equipment will be reassigned to other suitable inspection equipment, and the route will be replanned to ensure that the inspection tasks are completed on time.

5. The method for planning unmanned inspection tasks in a gas-fired power plant according to claim 4, characterized in that: The formula for calculating the crossover probability in step S3.4 is as follows: The formula for calculating the mutation probability is: in , These are the maximum and minimum crossover probabilities, respectively. , These are the maximum and minimum mutation probabilities, respectively. This represents the current maximum fitness function value of the population. This represents the current average fitness function value of the population.

6. A task planning system for unmanned inspection of a gas-fired power plant according to any one of claims 1-5, characterized in that: The system comprises a basic perception module, a data storage module, a task priority adjustment module, a collaborative planning module, a task execution module, a monitoring and feedback module, and an iterative optimization module. The collaborative planning module incorporates an improved dynamic task allocation and path planning collaborative optimization algorithm, namely the ID-TPCO algorithm, used to achieve collaborative optimization and dynamic adjustment of inspection task allocation and path planning. The basic perception module consists of LiDAR, a visual camera, various sensors, and a wireless communication module, used for inspection scene modeling, basic information collection, and real-time status perception. The data storage module uses a distributed database to store all basic data and inspection-related data. The task priority adjustment module calculates and sorts the dynamic values ​​of inspection task priorities. The task execution module consists of various inspection equipment used to execute inspection operations and provide data feedback. The monitoring and feedback module monitors the inspection process in real time, analyzes data, and handles faults. The iterative optimization module adjusts the ID-TPCO algorithm parameters to achieve iterative upgrades of the planning scheme.

Images