Inspection robot scheduling method and scheduling system

Abstract

The invention discloses an inspection robot scheduling method and scheduling system. The method comprises the following steps: S1, obtaining task information of each order; s2, generating an initial scheme: calculating a target income of each order, traversing all orders according to the target income, determining a feasible inspection robot trip, calculating time cost, reserving the first a feasible schemes from small to large, and optionally selecting one of the feasible schemes to obtain a target feasible scheme; s3, optimizing a target feasible scheme: presetting four destruction operators and a reconstruction operator, and combining to obtain a plurality of operator combinations; and selecting an operator combination to execute deletion and redistribution operations to obtain a solution, calculating an expected income of the solution, continuously performing iterative optimization until the expected income of the solution is greater than that of the target feasible scheme, and selecting the solution with the maximum expected income as an optimal scheduling scheme. The optimal scheme can be obtained through finite iteration calculation, calculation consumption is small, the response speed is high, and the obtained scheme is more reasonable and reliable.

Description

Technical Field

[0001] The present invention relates to the technical field of robot scheduling, and in particular to a patrol robot scheduling method and a scheduling system. Background Art

[0002] With the maturity of unmanned inspection technology and the increasing specialization of work scenarios, inspection robots are being applied to more and more fields, such as petrochemical, power inspection, agriculture and other industries. Compared with traditional manpower and ground vehicle services, inspection robots have the advantages of low cost, high efficiency and low carbon. Inspection robot scheduling is that inspection robot operators reasonably call the inspection robots in the system to provide customers with services such as package delivery, inspection, and monitoring based on customer needs, under the conditions that the load, working range, scene complexity and other requirements meet the requirements. In the planning process of the scheduling plan, the criteria are generally the more benefits, the higher efficiency, the less time spent, and the less energy consumed. In terms of benefits, efficiency and time consumption, it is necessary to comprehensively integrate the information of each task and the status of each robot for reasonable matching and optimization. As for energy consumption, at present, in order to ensure safety, most operators replace the inspection robot with a fully charged battery before each departure to prevent downtime caused by insufficient power during the journey. The energy consumption of inspection robots when performing tasks is mostly calculated by simulating a linear or nonlinear formula using a factor related to load, deadweight, speed, etc. This calculation method is extremely inaccurate and does not consider the impact of energy consumption uncertainty on scheduling. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and to provide a scheduling method for an inspection robot that can quickly and accurately obtain an optimization plan, as well as a scheduling system for an inspection robot that implements the method.

[0004] The present invention is achieved through the following technical solutions: A method for dispatching an inspection robot comprises the following steps: S1. Obtaining task information of each order, wherein the task information includes the location of the task point, the weight of the items to be carried, and the execution time; S2. Generate initial solution: S2-1. Calculate the target revenue of each order and sort them from large to small according to the target revenue to obtain an order set; S2-2, traverse all orders one by one in order of target benefits from large to small; for any order, the traversal method is: determine all inspection robot trips that can execute the order, and calculate the time cost of each inspection robot trip. If there is no inspection robot that can execute the order, delete the order from the order set; after sorting all inspection robot trips that can execute the order from small to large according to the time cost, retain the first a feasible solutions, and then randomly select one from the first a feasible solutions as the initial feasible solution for the order; wherein, the inspection robot completing an order and returning is a inspection robot trip, and the increase in time caused by the inspection robot executing a patrol robot trip is the time cost; S2-3, traverse all orders according to step S2-2 and output the target feasible solution; S3. Optimization of target feasible solutions: S3-1. Preset four destruction operators and one reconstruction operator, and combine the four destruction operators with the reconstruction operator to obtain four operator combinations; the four destruction operators are: randomly deleting orders of the allocated solutions, greedily deleting orders with the smallest target benefit in the allocated solutions, randomly deleting all orders of a certain inspection robot, greedily deleting all orders of the inspection robot with the smallest total target benefit; and a reconstruction operator is a solution with the smallest total time cost for rebuilding all orders; S3-2, based on the selected operator combination, perform deletion and reallocation operations on the target feasible solution obtained in step S2-3 to obtain a solution, and calculate the expected benefit of the solution; compare the expected benefit of the solution obtained with the target benefit of the target feasible solution obtained in step S2-3 (the sum of the target benefits of each order in the target feasible solution); S3-3. If the expected benefit of the solution is greater than the target benefit of the target feasible solution, the solution is taken as a new solution, and the number of consecutive times b that are not improved corresponding to the current operator combination is set to 0. Based on the current operator combination, the steps of performing the deletion and reallocation operations on the target feasible solution in step S3-2 to obtain the solution are recorded, and then step S3-2 is repeated based on the current operator combination; S3-4. If the expected benefit of the solution is less than or equal to the target benefit of the target feasible solution, then the number of unimproved consecutive times b corresponding to the current operator combination is increased by one, and step S3-2 is repeated based on the current operator combination until the number of unimproved consecutive times b corresponding to the current operator combination reaches the preset number threshold c; if the number of unimproved consecutive times b corresponding to the current operator combination reaches the preset number threshold c, step S3-5 is executed; S3-5, call the next operator combination, and repeat steps S3-2 to S3-4 until all operator combinations have completed iteration; S3-6. Select the solution with the largest expected benefit from all the new solutions obtained as the optimal solution. If the expected benefits of the new solutions obtained by all operator combinations are not greater than the target benefits of the target feasible solution obtained in step S2-3, then the target feasible solution is selected as the optimal solution to obtain the optimal scheduling plan.

[0005] Furthermore, the method for calculating the target revenue of each order in step S2-1 is as follows: constructing a first-stage model, wherein the first-stage model takes revenue maximization as the first objective function, and obtaining the target revenue by solving the maximum value of the first objective function; The first objective function is: , , The constraints are as follows: Each task point is executed at most once: , ; The inspection robot can only be called once per trip: , , ; The inspection robot’s energy consumption per trip cannot exceed its maximum energy capacity: , , ; The latest execution time constraint of the task: , , , ; Time continuity constraints of inspection robots: , , , , , ; u Representation and The index of another different inspection robot; The inspection robot continuously executes the following constraints: , , , , , ; Energy constraints of inspection robots: , , , as well as , ; Decision variable value constraints: , , , ; in, , represents the order set, m is a positive integer; , represents the maximum set of inspection robot trips; , represents the set of inspection robots; Indicates the energy consumption of the inspection robot; when other robot sets are introduced, Indicates robot subscript; Represents task completion The proceeds obtained afterwards; Represents task completion Time spent; Representative tasks The latest completion time required; Representing the inspection robot The number of trips to the mission point time; Indicates that the inspection robot is Complete the mission The accumulated energy consumption thereafter; represents the energy consumption of task j; Indicates the task energy consumption; E represents the full power of the inspection robot; Indicates uncertain variables; is a decision variable. If the inspection robot Complete the mission , then the decision variable The value is equal to 1, otherwise it is equal to 0; Used to determine whether the inspection robot's energy consumption exceeds the maximum energy capacity. ; Related to revenue, , , , , , ; Indicates completion of task The subsequent income; is a parameter in the model.

[0006] Furthermore, the method for calculating the expected benefit of the solution in step S3-2 is: S3-2-1. Based on the first-stage model, calculate the target revenue of all orders in the solution; S3-2-2. Construct the second-stage model. The second-stage model calculates the profit and loss with the second objective function, including judging whether the energy consumption of the inspection robot exceeds the maximum energy capacity, defining the profit and loss function, and clarifying the variable value range constraints; The second objective function is: ; in, , , , , , , ; Indicates completion of task The subsequent income; , represents the order set, m is a positive integer; , represents the maximum set of inspection robot trips; , represents the set of inspection robots; Indicates the energy consumption of the inspection robot; It represents the cumulative energy consumption of the inspection robot after completing the task for the rth time; E indicates full charge; Indicates uncertain variables; Used to determine whether the energy consumption of the inspection robot exceeds the maximum energy capacity; is a parameter in the model; S3-2-3. Using the second stage model, calculate the profit and loss of each order in the solution by solving the value of the second objective function. Based on the sum of the target profit and the sum of the profit and loss, calculate the expected profit of the solution.

[0007] Furthermore, in step S3-2-3, the second-stage model is used to calculate the profit and loss of each order in the solution by solving the value of the second objective function. Based on the sum of the target profit and the sum of the profit and loss, the method for calculating the expected profit of the solution is: S3-2-3-1. Randomly select N energy consumptions that conform to the normal distribution Inspection robot samples; among them, is the mean energy consumption, is the standard deviation; S3-2-3-2, traverse each of the N samples, use the second-stage model to calculate the profit loss of the solution under each sample, and superimpose and average all the profit losses under the N samples to calculate the total expected profit loss; S3-2-3-3. Subtract the target return of the solution obtained based on S3-2-1 and the expected return and total loss obtained based on S3-2-3-2 to obtain the expected return under sample size N, which is the expected return of the solution.

[0008] Furthermore, the energy consumption mean and standard deviation The calculation method is: Assume that the energy consumed by the inspection robot in performing task p is , then the mean energy consumption , standard deviation , , θ is the coefficient of dispersion, which can be obtained from experimental statistics; in, Indicates the weight of the inspection robot chassis; Indicates the weight of the equipment battery; represents the acceleration due to gravity; Indicates the weight of the current inspection robot equipment load, including the sensors, robotic arms, monitoring equipment, etc. It represents the one-way execution time of the inspection robot from the charging room to the mission target; Indicates the weight of the cargo to be transported for the task (can be 0).

[0009] Furthermore, the inspection robot consumes energy The calculation formula is: , in, Indicates the weight of the inspection robot chassis; Indicates the weight of the equipment battery; represents the acceleration due to gravity; Indicates the weight of the current inspection robot equipment load; It represents the one-way execution time of the inspection robot from the charging room to the mission target; Indicates the weight of the cargo that needs to be transported for the task.

[0010] A scheduling system for an inspection robot, used to implement the above-mentioned scheduling method for the inspection robot, comprising a task order module, a robot management module, a robot energy consumption calculation module, a target benefit calculation module, an expected benefit calculation module, an initial solution generation module and a target solution optimization module; The task order module is used for order generation and management; The robot management module is used to collect and update the status of the inspection robot in real time and assign tasks to the inspection robot; The robot energy consumption calculation module is used to calculate the energy consumption of the task according to the task information and the load of the robot, and to measure the energy consumption of the robot; The target benefit calculation module is used to calculate the target benefit of the initial plan; The expected benefit calculation module is used to obtain the expected benefits of solutions based on different operator combinations during the optimization process of the target feasible solution; The initial solution generation module is used to generate an initial target feasible solution according to the task order; The target solution optimization module is used to iteratively optimize the initial target feasible solution to obtain the optimal scheduling solution.

[0011] The present invention is based on task information and a random programming model, generates an initial plan, and then performs iterative optimization based on the initial plan based on an operator combination, and obtains an optimal scheduling plan by comparing the expected benefit of the solution obtained by the operator combination with the target benefit of the initial plan. This method can obtain the best plan through a finite number of iterative calculations, has low computational consumption, and fast response speed; at the same time, the uncertainty of energy consumption and the influence of the accuracy of energy consumption calculation are fully considered in the scheduling model and the solution algorithm, and the energy consumption of each part of the inspection robot is balanced one by one, and a number of reliable samples whose energy consumption conforms to the normal distribution are selected, and the benefits are balanced under these sample sizes, thereby eliminating the influence of energy consumption uncertainty; in addition, important factors such as operational safety, reliability and benefits are comprehensively considered in the scheduling model, so that the obtained scheduling plan is more reasonable and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The following is a flowchart of an embodiment of the present invention.

[0013] Figure 2 The figure is a flow chart of a method according to an embodiment of the present invention.

[0014] Figure 3 It is a framework diagram of the scheduling system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The present invention proposes a method for scheduling an inspection robot, which divides the scheduling planning process of the inspection robot into two stages: an initial plan generation stage and an iterative optimization plan stage. Specifically, Figure 1 , Figure 2 As shown, the following steps are included: S1. Obtain the task information of each order, which includes the location of the task point, the weight of the items to be carried, and the execution time. The task information can be obtained from the map of the scheduling system, and the task point information can include the starting and ending position coordinates of the task on the map. The task order can be set and added by the user on the scheduling system.

[0016] S2. Generate initial solution: S2-1. For the task orders to be executed, calculate the target revenue of each order, and sort them from large to small according to the target revenue to obtain an order set; S2-2, traverse all orders one by one in order of target benefits from large to small; for any order, the traversal method is: determine all inspection robot trips that can execute the order, and calculate the time cost of each inspection robot trip. If there is no inspection robot that can execute the order, delete the order from the order set; after sorting all inspection robot trips that can execute the order from small to large according to the time cost, retain the first a feasible solutions, delete the subsequent solutions with large time costs, and then randomly select one from the first a feasible solutions as the initial feasible solution for the order; wherein, the inspection robot completing a certain order and returning is a inspection robot trip, and the increase in time caused by the inspection robot executing a patrol robot trip is the time cost; S2-3, traverse all orders according to step S2-2, and output the target feasible solution, which is the initial solution; S3. Optimization of target feasible solutions: S3-1. Preset four destruction operators and one reconstruction operator, and combine the four destruction operators with the reconstruction operator to obtain four operator combinations; the four destruction operators are: randomly deleting orders of the allocated solutions, greedily deleting orders with the smallest target benefit in the allocated solutions, randomly deleting all orders of a certain inspection robot, greedily deleting all orders of the inspection robot with the smallest total target benefit; and a reconstruction operator is a solution with the smallest total time cost for rebuilding all orders; S3-2, based on the selected operator combination, perform deletion and reallocation operations on the target feasible solution obtained in step S2-3 to obtain a solution, and calculate the expected benefit of the solution; compare the expected benefit of the solution obtained with the target benefit of the target feasible solution obtained in step S2-3 (the sum of the target benefits of each order in the target feasible solution); S3-3. If the expected benefit of the solution is greater than the target benefit of the target feasible solution, the solution is taken as a new solution, and the number of consecutive times b that are not improved corresponding to the current operator combination is set to 0. Based on the current operator combination, the steps of performing the deletion and reallocation operations on the target feasible solution in step S3-2 to obtain the solution are recorded, and then step S3-2 is repeated based on the current operator combination; S3-4. If the expected benefit of the solution is less than or equal to the target benefit of the target feasible solution, then the number of unimproved consecutive times b corresponding to the current operator combination is increased by one, and step S3-2 is repeated based on the current operator combination until the number of unimproved consecutive times b corresponding to the current operator combination reaches the preset number threshold c; if the number of unimproved consecutive times b corresponding to the current operator combination reaches the preset number threshold c, step S3-5 is executed; S3-5, call the next operator combination, and repeat steps S3-2 to S3-4 until all operator combinations have completed iteration; S3-6. Select the solution with the largest expected benefit from all the new solutions obtained as the optimal solution. If the expected benefits of the new solutions obtained by all operator combinations are not greater than the target benefits of the target feasible solution obtained in step S2-3, then the target feasible solution is selected as the optimal solution to obtain the optimal scheduling plan.

[0017] In the above iterative optimization of each operator combination, although the initial plan is randomly deleted and greedily deleted, during the iteration process, even if some orders may be deleted or adjusted, the overall idea of ​​sorting by profit helps to maintain the focus and priority on high profits under various operator operations, so that the system can develop in the direction of increasing profits under different operations.

[0018] For each combination of destruction operator and reconstruction operator (four combinations in total), during the iterative optimization process, each time the target feasible solution is operated based on the selected operator combination result, a solution will be obtained. After multiple iterations of the current operator combination, among the multiple solutions, some solutions have expected benefits greater than the target benefit of the initial target feasible solution, and these solutions are retained. For some solutions with target benefits less than the initial target feasible solution, they are not retained. During the entire iteration process, different operator combination results will be continuously tried, and when the number of consecutive unimproved solutions does not reach the set number threshold, the current operator combination will continue to operate and update the solution until the solution is not improved for multiple consecutive times, so multiple new solutions that are better than the initial solution can be obtained. After completing the iteration of all operator combination results, the solutions that are better than the initial target feasible solution are retained. These new solutions are optimized, and finally an optimal solution is output, that is, the solution with the largest expected benefit during the iteration process. Since this optimal solution is compared with the initial solution and then screened out from many solutions, its benefit can be maximized compared with the initial solution and other solutions, and it is the best solution that can be found under finite calculations.

[0019] For the calculation of the above-mentioned target benefits and expected benefits, the present invention constructs a two-stage stochastic programming model including a first-stage model and a second-stage model, wherein the first-stage model mainly takes maximizing system benefits as the objective function and includes a variety of constraints, and the second-stage model is mainly used to calculate the benefit loss and define the state parameters and variables of the inspection robot during operation.

[0020] The process of establishing a stochastic programming model is as follows: 1. Consider energy consumption and determine the sample The existing calculations of robot energy consumption have the following problems: (1) Inaccurate calculation of energy consumption: The factors affecting the operation of the inspection robot are very complex, and the energy consumption is highly random. The simulation formulas based on factors such as load, deadweight, and speed are not accurate in calculating energy consumption. (2) The impact of energy consumption uncertainty on scheduling is not considered: The energy consumption uncertainty factor is not included in the inspection robot scheduling model and solution algorithm, and there is a lack of consideration for crucial aspects such as the inspection robot's operational safety and the benefits of the scheduling system.

[0021] In view of the above problems, the present invention introduces the symbol Q to represent the energy level of the fully charged battery equipped on the inspection robot, considering the limited endurance of the inspection robot. The energy consumed by the inspection robot in performing task p is , The calculation formula is: , in, Indicates the weight of the inspection robot chassis; Indicates the weight of the equipment battery; represents the acceleration due to gravity, Indicates the weight of the current inspection robot equipment load, including the sensors, robotic arms, monitoring equipment, etc. represents the one-way execution time of the inspection robot from the charging room to the mission target, Indicates the weight of the cargo to be transported for the task (can be 0). The above formula can be used to calculate the energy consumption of the inspection robot.

[0022] For the above formula, Assuming that the energy consumed by the inspection robot follows a normal distribution, the mean energy consumption is , standard deviation , θ is the coefficient of dispersion, which can be obtained through experimental statistics. According to the standard deviation principle (more than 95% of the true distribution value is within plus or minus 3 times the standard deviation Within), randomly select N energy consumptions that conform to the normal distribution The inspection robot samples, in the subsequent model construction and algorithm design, all the uncertain values ​​involved are expected values ​​calculated based on the N samples, that is, N energy consumption values ​​that conform to the normal distribution are randomly selected for calculation to obtain the expected values ​​of N energy consumption. For the key components of the inspection robot, such as batteries and motors, these N samples can be used to conduct accelerated life tests or actual operation monitoring, and their service life data under different working conditions can be recorded. Based on these sample data, the distribution law of component life is analyzed, and then the overall reliability index and life expectancy of the equipment are determined, providing a basis for the maintenance and replacement strategy of the equipment.

[0023] 2. Define parameters and variables The parameters and variables defined are: Task Order Collection , m is a positive integer; Inspection robot local status information collection; ; Maximum number of inspection robot trips ; Energy consumption of inspection robots ; When other robot sets are introduced, Indicates the robot subscript, ; Complete the task The benefits obtained later ; Complete the task Time spent ; Task Required latest completion time ; Inspection robots are in the The number of trips to the mission point Time ; Inspection robots are Complete the mission Cumulative energy consumption ; Task Energy consumption ; Energy consumption of task j ; The inspection robot is fully charged E; Uncertain variables ; Decision variables , if the inspection robot is in the Complete the mission , then the decision variable The value is equal to 1, otherwise it is equal to 0; Used to determine whether the energy consumption of the inspection robot exceeds the maximum energy capacity. ; Related to revenue, , , , , , ; is a parameter in the model.

[0024] 3. Build the first stage model The first-stage model takes profit maximization as the first objective function, and obtains the target profit by solving the maximum value of the first objective function.

[0025] The first objective function is: , , The constraints are as follows: Each task point is executed at most once: , ; The inspection robot can only be called once per trip: , ; The inspection robot’s energy consumption per trip cannot exceed its maximum energy capacity: , ; The latest execution time constraint of the task: , , ; Time continuity constraints of inspection robots: , , , , , ; u refers to the index of another inspection robot different from i. If multiple inspection robots are used, this constraint ensures the temporal continuity of inspection robot i. The inspection robot continuously executes the following constraints: , , , , , ; Energy constraints of inspection robots: , , ,

[0026] as well as , ; Decision variable value constraints: , , , ; The definitions of various parameters and variables refer to the previous section.

[0027] 4. Build the second stage model The second stage model calculates the profit and loss with the second objective function, including judging whether the energy consumption of the inspection robot exceeds the maximum energy capacity, defining the profit and loss function, and clarifying the variable value range constraints; The second objective function is: ; The definitions of various parameters and variables refer to the previous section.

[0028] Based on the stochastic programming model constructed above, the target return in step S2-1 and the expected return in step S3-2 are calculated. Specifically: In step S2-1, the target profit of each order is obtained by solving the maximum value of the first objective function. The sum of the target profits of each order in the target feasible solution is the target profit of the target feasible solution, which is compared with the expected profit of the solution in step S3-2.

[0029] In step S3-2, the method for calculating the expected return of the solution is: S3-2-1. Based on the first-stage model, the target revenue of all orders in the solution is calculated by solving the maximum value of the first objective function. The sum of the target revenues of all orders is the target revenue of the solution. S3-2-2. Using the second stage model, calculate the profit and loss of each order in the solution by solving the value of the second objective function, and calculate the expected profit of the solution based on the sum of the target profit and the sum of the profit and loss. The specific method is: (1) Randomly select N energy consumptions that conform to the normal distribution Inspection robot sample.

[0030] (2) Traverse each of the N samples, use the second-stage model to calculate the profit loss of the solution under each sample, and superimpose and average all the profit losses under the N samples to calculate the total expected profit loss; the energy consumed is , please refer to the calculation in Section 1 above.

[0031] (3) The target return of the solution obtained based on S3-2-1 and the expected return total loss obtained in (2) above are subtracted from each other to obtain the expected return under sample size N, which is the expected return of the solution.

[0032] A dispatching system for an inspection robot is used to implement a dispatching method for an inspection robot, such as Figure 3 As shown, it includes a task order module, a robot management module, a robot energy consumption calculation module, a target profit calculation module, an expected profit calculation module, an initial solution generation module and a target solution optimization module.

[0033] The task order module is used for order generation and management; The robot management module is used to collect and update the status of the inspection robot in real time and assign tasks to the inspection robot; The robot energy consumption calculation module is used to calculate the energy consumption of the task according to the task information and the load of the robot, and to measure the energy consumption of the robot; The target benefit calculation module is used to calculate the target benefit of the initial plan; The expected benefit calculation module is used to obtain the expected benefits of solutions based on different operator combinations during the optimization process of the target feasible solution; The initial solution generation module is used to generate an initial target feasible solution according to the task order; The target solution optimization module is used to iteratively optimize the initial target feasible solution to obtain the optimal scheduling solution.

[0034] The above detailed description is a specific description of a feasible embodiment of the present invention. The embodiment is not intended to limit the patent scope of the present invention. Any equivalent implementation or modification that does not deviate from the present invention should be included in the patent scope of this case.

Claims

1. A method for dispatching an inspection robot, characterized in that: The steps include: S1. Obtaining task information of each order, wherein the task information includes the location of the task point, the weight of the items to be carried, and the execution time; S2. Generate initial solution: S2-1. Calculate the target revenue of each order and sort them from large to small according to the target revenue to obtain an order set; S2-2, traverse all orders one by one in order of target benefits from large to small; for any order, the traversal method is: determine all inspection robot trips that can execute the order, and calculate the time cost of each inspection robot trip. If there is no inspection robot that can execute the order, delete the order from the order set; after sorting all inspection robot trips that can execute the order from small to large according to the time cost, retain the first a feasible solutions, and then randomly select one from the first a feasible solutions as the initial feasible solution for the order; wherein, the inspection robot completing an order and returning is a inspection robot trip, and the increase in time caused by the inspection robot executing a patrol robot trip is the time cost; S2-3, traverse all orders according to step S2-2 and output the target feasible solution; S3. Optimization of target feasible solutions: S3-1. Preset four destruction operators and one reconstruction operator, and combine the four destruction operators with the reconstruction operator to obtain four operator combinations; the four destruction operators are: randomly deleting orders of the allocated solutions, greedily deleting orders with the smallest target benefit in the allocated solutions, randomly deleting all orders of a certain inspection robot, greedily deleting all orders of the inspection robot with the smallest total target benefit; and a reconstruction operator is a solution with the smallest total time cost for rebuilding all orders; S3-2, based on the selected operator combination, perform deletion and reallocation operations on the target feasible solution obtained in step S2-3 to obtain a solution, and calculate the expected benefit of the solution; compare the expected benefit of the solution obtained with the target benefit of the target feasible solution obtained in step S2-3; S3-3. If the expected benefit of the solution is greater than the target benefit of the target feasible solution, the solution is taken as a new solution, and the number of consecutive times b that are not improved corresponding to the current operator combination is set to 0. Based on the current operator combination, the steps of performing the deletion and reallocation operations on the target feasible solution in step S3-2 to obtain the solution are recorded, and then step S3-2 is repeated based on the current operator combination; S3-4. If the expected benefit of the solution is less than or equal to the target benefit of the target feasible solution, then the number of unimproved consecutive times b corresponding to the current operator combination is increased by one, and step S3-2 is repeated based on the current operator combination until the number of unimproved consecutive times b corresponding to the current operator combination reaches the preset number threshold c; if the number of unimproved consecutive times b corresponding to the current operator combination reaches the preset number threshold c, step S3-5 is executed; S3-5, call the next operator combination, and repeat steps S3-2 to S3-4 until all operator combinations have completed iteration; S3-6. Select the solution with the largest expected benefit from all the new solutions obtained as the optimal solution. If the expected benefits of the new solutions obtained by all operator combinations are not greater than the target benefits of the target feasible solution obtained in step S2-3, then the target feasible solution is selected as the optimal solution to obtain the optimal scheduling plan.

2. The method for dispatching an inspection robot according to claim 1, characterized in that: The method for calculating the target profit of each order in step S2-1 is: Construct the first-stage model. The first-stage model takes profit maximization as the first objective function, and obtains the target profit by solving the maximum value of the first objective function; The first objective function is: , ， The constraints are as follows: Each task point is executed at most once: , ; The inspection robot can only be called once per trip: , , ; The inspection robot’s energy consumption per trip cannot exceed its maximum energy capacity: ， ； The latest execution time constraint of the task: , , , ; Time continuity constraints of inspection robots: , ， , ， , ； u Representation and The index of another different inspection robot; The inspection robot continuously executes the following constraints: , ， , ， , ； Energy constraints of inspection robots: , ， , ； as well as , ； Decision variable value constraints: , , , ; in, , represents the order set; , represents the maximum set of inspection robot trips; , represents the set of inspection robots; Indicates the energy consumption of the inspection robot; Represents task completion The proceeds obtained afterwards; Represents task completion Time spent; Representative tasks The latest completion time required; Representing the inspection robot The number of trips to the mission point time; Indicates that the inspection robot is Complete the mission The accumulated energy consumption afterwards; Indicates the task energy consumption; represents the energy consumption of task j; E represents the full power of the inspection robot; Indicates uncertain variables; is a decision variable. If the inspection robot Complete the mission , then the decision variable The value is equal to 1, otherwise it is equal to 0; Used to determine whether the inspection robot's energy consumption exceeds the maximum energy capacity. ； ， ， , ， , ； is a parameter in the model.

3. The method for dispatching an inspection robot according to claim 2, characterized in that: The method for calculating the expected return of the solution in step S3-2 is: S3-2-1. Based on the first-stage model, calculate the target revenue of all orders in the solution; S3-2-2. Construct the second-stage model. The second-stage model calculates the profit and loss with the second objective function, including judging whether the energy consumption of the inspection robot exceeds the maximum energy capacity, defining the profit and loss function, and clarifying the variable value range constraints; The second objective function is: ； in, , ， ， , ， , ； Indicates completion of task The subsequent income; , represents the order set; , represents the maximum set of inspection robot trips; , represents the set of inspection robots; Indicates the energy consumption of the inspection robot; It represents the cumulative energy consumption of the inspection robot after completing the task for the rth time; E indicates full charge; Indicates uncertain variables; Used to determine whether the energy consumption of the inspection robot exceeds the maximum energy capacity; is a parameter in the model; S3-2-3. Using the second stage model, calculate the profit and loss of each order in the solution by solving the value of the second objective function. Based on the sum of the target profit and the sum of the profit and loss, calculate the expected profit of the solution.

4. The method for dispatching an inspection robot according to claim 3, characterized in that: In step S3-2-3, the second-stage model is used to calculate the profit and loss of each order in the solution by solving the value of the second objective function. Based on the sum of the target profit and the sum of the profit and loss, the method for calculating the expected profit of the solution is: S3-2-3-1. Randomly select N energy consumptions that conform to the normal distribution Inspection robot samples; among them, is the average energy consumption of the inspection robot, is the standard deviation of energy consumption; S3-2-3-2, traverse each of the N samples, use the second-stage model to calculate the profit loss of the solution under each sample, and superimpose and average all the profit losses under the N samples to calculate the total expected profit loss; S3-2-3-3. Subtract the target return of the solution obtained based on S3-2-1 and the expected return and total loss obtained based on S3-2-3-2 to obtain the expected return under sample size N, which is the expected return of the solution.

5. The method for dispatching an inspection robot according to claim 4, characterized in that: The energy consumption mean and standard deviation The calculation method is: Assume that the energy consumed by the inspection robot in the round trip to perform task p is , then the mean energy consumption , standard deviation , , is the coefficient of dispersion, in, Indicates the weight of the inspection robot chassis; Indicates the weight of the equipment battery; represents the acceleration due to gravity; Indicates the weight of the current inspection robot equipment load; It represents the one-way execution time of the inspection robot from the charging room to the mission target; Indicates the weight of the cargo that needs to be transported for the task.

6. A method for dispatching an inspection robot according to claim 2 or 3, characterized in that: The energy consumption of the inspection robot The calculation formula is: ， in, Indicates the weight of the inspection robot chassis; Indicates the weight of the equipment battery; represents the acceleration due to gravity; Indicates the weight of the current inspection robot equipment load; It represents the one-way execution time of the inspection robot from the charging room to the mission target; Indicates the weight of the cargo that needs to be transported for the task.

7. A dispatching system for an inspection robot, used to implement the dispatching method for an inspection robot as claimed in any one of claims 1 to 6, characterized in that: It includes a task order module, a robot management module, a robot energy consumption calculation module, a target benefit calculation module, an expected benefit calculation module, an initial solution generation module and a target solution optimization module; The task order module is used for order generation and management; The robot management module is used to collect and update the status of the inspection robot in real time and assign tasks to the inspection robot; The robot energy consumption calculation module is used to calculate the energy consumption of the task according to the task information and the load of the robot, and to measure the energy consumption of the robot; The target benefit calculation module is used to calculate the target benefit of the initial plan; The expected benefit calculation module is used to obtain the expected benefits of solutions based on different operator combinations during the optimization process of the target feasible solution; The initial solution generation module is used to generate an initial target feasible solution according to the task order; The target solution optimization module is used to iteratively optimize the initial target feasible solution to obtain the optimal scheduling solution.

Images