Real-time temperature measurement and optimal scheduling method for rail-mounted intelligent inspection robot

Abstract

The invention discloses a real-time temperature measurement and optimal scheduling method for a rail-mounted intelligent inspection robot, and the method comprises the steps: obtaining the temperature of equipment along a rail in real time through a temperature sensor on the inspection robot, analyzing the working state of the equipment along the rail, and recognizing a potential overheating risk region; establishing a multi-robot collaborative inspection task scheduling model according to the position information and the emergency degree of the potential overheating risk area; planning an inspection path for the multi-robot collaborative task scheduling data by taking the shortest time as an index, optimizing the inspection path by adopting a fusion improved ant colony algorithm and combining a TEB strategy, and obtaining a global planning path; a trajectory optimization strategy based on Minimum Snap is adopted for the global planning path, a Bezier curve is utilized to optimize the path trajectory, a multi-robot collaborative task scheduling model is combined with an auction algorithm to dynamically allocate tasks, efficient utilization of resources and priority processing of emergency tasks are ensured, the inspection efficiency and safety are improved, and the operation cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of rail robot intelligent inspection technology, and in particular to a real-time temperature measurement and optimized scheduling method for a rail-type intelligent inspection robot. Background Art

[0002] Track-mounted intelligent inspection robots play a vital role in modern rail transit systems. They conduct regular inspections and maintenance of track facilities through automated means, ensuring the safety and stability of the system. With the development of technology, these robots not only need to have high-precision detection capabilities, but also need to achieve efficient task scheduling and management. In particular, the addition of real-time temperature measurement functions enables inspection robots to detect potential safety hazards at an early stage, such as overheating components or abnormal temperature changes, so that timely measures can be taken to avoid accidents.

[0003] Current rail-mounted intelligent inspection robots rely primarily on the following key technologies: They utilize a fusion of multiple sensors, such as lidar, cameras, and infrared thermal imagers, to acquire environmental information, combined with deep learning algorithms for data analysis to achieve high-precision target detection and recognition. They also plan optimal inspection routes to ensure the robots can efficiently and safely perform tasks in complex environments. They also support centralized management and remote control of multiple inspection robots, enabling real-time data transmission, analysis, and early warning notifications. While existing technologies have made significant progress, some shortcomings remain in practical applications:

[0004] (1) Although existing real-time temperature measurement technologies can capture temperature changes to a certain extent, they may sacrifice some accuracy or real-time performance when processing large-scale data due to limitations in computing resources and response speed.

[0005] (2) Although the existing inspection robot scheduling algorithm can plan a relatively reasonable inspection route, it is still insufficient in terms of dynamic adjustment. When encountering emergencies, the existing scheduling system is difficult to respond quickly and re-plan the route, which may lead to low inspection efficiency and even delay the completion of key maintenance tasks, increasing the risk of safety hazards. Summary of the Invention

[0006] The purpose of the present invention is to provide a real-time temperature measurement and optimized scheduling method for a track-type intelligent inspection robot to solve the technical problems in the prior art of limited real-time and accuracy of temperature measurement and insufficient flexibility of the scheduling and dispatching mechanism of the inspection robot.

[0007] In order to solve the above technical problems, the present invention specifically provides the following technical solutions:

[0008] The present invention provides a real-time temperature measurement and optimization scheduling method for a track-type intelligent inspection robot, comprising the following steps:

[0009] The temperature of equipment along the track is obtained in real time through the temperature sensor on the inspection robot. The working status of each equipment along the track is analyzed based on the temperature of the equipment along the track, and potential overheating risk areas are identified;

[0010] Based on the location information and urgency of the potential overheating risk area and the inspection tasks of the current robots, a multi-robot collaborative inspection task scheduling model is established to obtain multi-robot collaborative task scheduling data;

[0011] The multi-robot collaborative task scheduling data is used to plan the inspection path with the shortest time as the indicator, and the inspection path is optimized by using the fusion improved ant colony algorithm combined with the TEB strategy to obtain the global planning path;

[0012] A Minimum Snap trajectory optimization strategy is adopted for the global planning path, and the speed and acceleration of the inspection robot are optimized using Bezier curves to obtain a smooth and safe trajectory.

[0013] As a preferred solution of the present invention, the temperature of the equipment along the track is obtained in real time by the temperature sensor on the inspection robot, and the working status of each equipment along the track is analyzed based on the temperature of the equipment along the track to identify potential overheating risk areas, including:

[0014] Using a high-precision temperature sensor on the inspection robot to perform real-time temperature detection on equipment along the track within a predetermined time interval to obtain temperature data of the equipment along the track;

[0015] The collected temperature data is transmitted to a remote monitoring center in real time through the communication module built into the inspection robot, and the temperature data is analyzed using a supervised learning algorithm to obtain the timestamp, location information, temperature value and environmental parameters contained in each temperature data point;

[0016] Annotate the collected temperature data based on historical temperature data of equipment along the track, and use the annotation results as labels for the supervised learning model to train the corresponding temperature data;

[0017] A feature selection algorithm based on feature correlation is used in the supervised learning model to select the feature with the greatest contribution value to the classification task and obtain temperature feature data;

[0018] Classifying the temperature characteristic data according to the annotations, dividing the classified temperature characteristic data into a training set and a data set, and using the training set to train the supervised learning model to optimize model parameters;

[0019] The trained supervised learning model is used to train the data set, and the supervised learning model is deployed to the remote monitoring center to directly process the temperature data, predict the temperature of equipment along the track in real time, mark the temperature, obtain the operating status of equipment along the track, and identify potential overheating risk areas.

[0020] As a preferred solution of the present invention, based on the location information and urgency of the potential overheating risk area and combined with the current robot's inspection task, a multi-robot collaborative inspection task scheduling model is established, including:

[0021] Based on the location information of the potential overheating risk area, an urgency assessment is performed on each identified risk area, and all identified risk areas are sorted according to their urgency to obtain a risk area priority set;

[0022] According to the risk area priority set, the current position, remaining power, and current task progress information of each inspection robot are collected in real time, and whether multiple robots are assigned to similar or identical tasks at the same time node are analyzed, and the inspection tasks of all inspection robots in the current state are obtained;

[0023] For all inspection robots in the current state, the Dijkstra path planning algorithm is used to calculate the optimal path from the current position of the inspection robot to each risk area;

[0024] Assign inspection tasks to the inspection robots according to the risk area priority set, and establish a multi-robot collaborative inspection task scheduling model using a load balancing strategy;

[0025] A dynamic adjustment mechanism is set for the multi-robot collaborative inspection task scheduling model to dynamically adjust the task allocation of the inspection robots according to the real-time risk area or the real-time status of the inspection robots.

[0026] As a preferred solution of the present invention, obtaining multi-robot collaborative task scheduling data according to the multi-robot collaborative inspection task scheduling model includes:

[0027] Obtaining a task list for each inspection robot according to the real-time task status of the inspection robot, including the target location, estimated arrival time, and operations to be performed;

[0028] Classify the inspection robots' task scheduling according to the task list, establish a three-dimensional coordinate axis, divide the task scheduling area of multiple robots, and obtain the inspection type of the multi-robot collaborative inspection task;

[0029] Matching the inspection types of the multi-robot collaborative inspection tasks to inspection robots according to the inspection tasks. When the number of inspection robots is greater than the number of tasks, ensuring that the inspection robots can perform multiple tasks and generating corresponding task sequences;

[0030] The task scheduling is completed according to the task sequence, and the timeliness and rationality of the task scheduling are evaluated using an evaluation function. The evaluation function expression is:

[0031]

[0032] Where R represents the robot set consisting of n inspection robots R = {r1, r2, ..., r n}, T represents a task set consisting of m tasks T = {t1, t2, ..., t m}, i represents the i-th inspection robot, j represents the j-th task, Q i represents the task sequence of the i-th inspection robot, s(j,Q i ) is used to determine whether the jth task is in the task sequence Q i In the case of s(j,Q i )=1 means it is in the sequence, otherwise it is not, N T Represents the task scheduling evaluation index;

[0033] Real-time multi-robot collaborative task scheduling data is obtained according to the task scheduling evaluation index.

[0034] As a preferred solution of the present invention, the multi-robot collaborative task scheduling data is used to plan the inspection path with the shortest time as an indicator, including:

[0035] Summarize the location coordinates, urgency and inspection priority of all potential risk areas based on the multi-robot collaborative task scheduling data, and obtain the current position, remaining power and task completion status of each inspection robot in real time according to the task sequence;

[0036] Loading real-time rail transit map data, taking the current position of each inspection robot as the starting point, and determining the target destination of each task according to the priority ranking of the risk areas;

[0037] The Dijkstra algorithm is used to find the shortest path with the shortest time as the indicator. A conflict detection mechanism is introduced to identify the possible cross paths or time overlaps of different robots during the execution of tasks, and the inspection path is planned in real time.

[0038] As a preferred solution of the present invention, the task scheduling based on the shortest time is implemented based on an auction algorithm, wherein the inspection robot performing the task is regarded as the auctioneer, the task to be performed is regarded as the auctioned item, and the person who publishes the task information is regarded as the initiator of the auction;

[0039] Each inspection robot uses the constraints of highest mission benefit, lowest mission cost, and shortest mission time as bidding indicators. The robots exchange information, negotiate, and discuss with each other to determine who will take on a particular mission.

[0040] The task benefit index is set as the benefit value of each inspection task point given to the inspection robot before the inspection robot performs task scheduling according to the actual environment requirements;

[0041] The task cost indicator is set as the remaining power value of the inspection robot, the location of the inspection task point, and the distance of the inspection robot's current task, and its expression is:

[0042] W cost =d cost +E cost

[0043] Among them, W cost represents the cost of the inspection robot executing the task sequence, d cost Indicates the distance traveled by the inspection robot when performing the task, E cost Indicates the amount of power consumed by the inspection robot when performing tasks.

[0044] As a preferred solution of the present invention, the auction algorithm implements a task scheduling process, including:

[0045] In the task scheduling process, the number of tasks is set to z, the number of inspection robots that perform the tasks is n, and multiple tasks are auctioned simultaneously;

[0046] Calculate the maximum profit obtained when the inspection robot acts as an auctioneer, and schedule the tasks of multiple robots in real time based on the maximum profit value. The maximum profit expression is:

[0047]

[0048] Among them, i represents the i-th inspection robot, j represents the j-th task, and p ij represents the revenue value of the i-th auctioneer inspection robot performing the j-th task, v ij represents the bidding price of the i-th auctioneer inspection robot for the j-th task, x ij Indicates whether the jth task is assigned to the i-th auctioneer inspection robot.

[0049] As a preferred solution of the present invention, the inspection path is optimized by using a fusion improved ant colony algorithm combined with a TEB strategy to obtain a global planning path, including:

[0050] Modeling the rail transit environment based on the inspection route, initializing the coordinate information of each target node using a grid map, obtaining track, station and obstacle location information, and defining the key nodes of the inspection path;

[0051] The concentration level of pheromones on each inspection route is defined based on the location coordinates, urgency and inspection priority of all potential risk areas, and a possible path chain is gradually formed by randomly selecting destinations using ant colonies;

[0052] Calculate the efficiency index of the path taken by each ant, update the path attraction between adjacent target points, and iterate and record the optimal path. The probability of selecting the optimal path depends on the pheromone concentration and the predicted benefit of the path, and its expression is:

[0053]

[0054] Among them, τ xy (t) represents the pheromone concentration of the path (x, y) from node x to node y at time t, δ xy (t) represents the heuristic function of the ant from the target node x to the node y. The heuristic function is related to the Euclidean distance between the target node x and the node y. o (x) represents the set of target points that ant o is allowed to choose in the next step, α and β represent the relative importance weights of pheromone and heuristic factor, u represents any node on the path, τ xu (t) represents the pheromone concentration of the path (x,u) from node x to node u at time t, δ yu (t) represents the heuristic function of the ant from the target node x to the node u, which is related to the Euclidean distance between the target node x and the node u.

[0055] As a preferred solution of the present invention, a Minimum Snap trajectory optimization strategy is adopted for the global planning path, including:

[0056] The global path planning is divided into several segments according to the predicted benefits of the path, each segment consists of a starting point, an end point and intermediate key control points;

[0057] Assign a time variable t to each trajectory segment, define the initial and final states of each trajectory segment, including position, velocity, and acceleration, and use a polynomial function to represent the change of the path position node over time. The expression is:

[0058] q(t)=a0+a1t+a2t 2+a3t 3 +a4t 4

[0059] Among them, a0, a1, a2, a3 and a4 all represent polynomial coefficients;

[0060] The Minimum Snap objective function is used to optimize the trajectory equation. The expression of the Minimum Snap objective function is:

[0061]

[0062] in, represents the fourth-order derivative of the trajectory, and T represents the time constant of the entire trajectory optimization;

[0063] Constraints are established on the speed and acceleration of the inspection robot to ensure the continuity between adjacent trajectories.

[0064] As a preferred solution of the present invention, the speed and acceleration of the inspection robot are optimized using Bezier curves to construct constraint conditions and obtain a smooth and safe trajectory, including:

[0065] The starting point, end point and intermediate key control points on the path trajectory are connected in sequence using Bezier curves, and control points that affect the predicted benefits are randomly selected on the path trajectory. The position and number of the control points can be changed. The Bezier curve expression is constructed as follows:

[0066]

[0067] Where n represents the order of the curve, θ represents the Bernstein coefficient of the Bezier curve, and P b represents the control point of the control path trajectory, b represents the control node on any trajectory path, and t represents the time variable;

[0068] The first-order derivative and multi-order derivative of the Bezier curve are respectively performed to obtain the recursive relationship between the control points on the trajectory path, and the speed and acceleration of the inspection robot are calculated in real time.

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] The present invention uses high-precision temperature sensors to monitor equipment temperature in real time and uses supervised learning algorithms to analyze and identify potential overheating risk areas. It adopts a multi-robot collaborative task scheduling model combined with an auction algorithm to dynamically allocate tasks, ensuring efficient resource utilization and prioritizing urgent tasks. It uses an improved ant colony algorithm and a time dilation and time-dilation (TEB) strategy to optimize global path planning, and achieves smooth and safe motion trajectories through Minimum Snap trajectory optimization and Bezier curves. This not only improves inspection efficiency and safety, but also reduces operating costs, providing an intelligent, automated and efficient solution for the maintenance and management of rail transit facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other implementation drawings based on the provided drawings without inventive effort.

[0072] Figure 1 Flowchart of the real-time temperature measurement and optimized scheduling method for a track-mounted intelligent inspection robot provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] like Figure 1 As shown, the present invention provides a real-time temperature measurement and optimization scheduling method for a track-type intelligent inspection robot, comprising the following steps:

[0075] The temperature of equipment along the track is obtained in real time through the temperature sensor on the inspection robot. The working status of each equipment along the track is analyzed based on the temperature of the equipment along the track, and potential overheating risk areas are identified;

[0076] In this embodiment, by real-time monitoring of temperature changes of equipment along the track, an early warning can be issued before the equipment overheats, and timely measures can be taken to prevent failures from occurring.

[0077] In this embodiment, the temperature sensor is integrated with the sensor group on the inspection robot to perform multi-dimensional data analysis, which can comprehensively evaluate the health status of the equipment from multiple angles and provide more accurate diagnosis results.

[0078] Based on the location information and urgency of the potential overheating risk area and the inspection tasks of the current robots, a multi-robot collaborative inspection task scheduling model is established to obtain multi-robot collaborative task scheduling data;

[0079] In this embodiment, by analyzing temperature data in real time and identifying potential overheating risk areas, the system can quickly dispatch nearby inspection robots to the area for detailed inspection. This greatly shortens the time from problem discovery to action, improves emergency response efficiency, and sorts tasks according to the urgency of the risk area to ensure that the most urgent problems can be handled as quickly as possible, thereby reducing the occurrence of safety hazards.

[0080] In this embodiment, based on the current position, task status and distribution of risk areas of each robot, the system can intelligently allocate tasks to avoid waste of resources or excessive concentration in a certain area, thereby achieving a more balanced task distribution. The collaborative work between multiple robots can cover a larger range, complete more inspection tasks at the same time, and speed up problem detection.

[0081] The multi-robot collaborative task scheduling data is used to plan the inspection path with the shortest time as the indicator, and the inspection path is optimized by using the fusion improved ant colony algorithm combined with the TEB strategy to obtain the global planning path;

[0082] In this embodiment, by planning the path with the shortest time as the goal, it can be ensured that the inspection robot completes all scheduled tasks in the shortest possible time. This is especially important for emergency situations that require a quick response. For example, when a potential overheating risk area is discovered, a robot can be quickly dispatched to deal with it. The optimized path reduces unnecessary driving distance and time, allowing the robot to cover more checkpoints in a limited time, thereby improving overall inspection efficiency.

[0083] In this embodiment, the fusion of the improved ant colony algorithm and the TEB strategy can not only consider the single goal of shortest time, but also comprehensively consider other factors such as minimizing energy consumption and maximizing path smoothness, to achieve multi-objective optimization, thereby better adapting to the needs of different application scenarios.

[0084] In this embodiment, the improved ant colony algorithm combined with the TEB strategy can effectively process large-scale data sets and maintain efficient computing performance when facing complex rail transit environments. This is especially important for large-scale inspection tasks involving multiple inspection robots, a large number of checkpoints, and complex terrain conditions. The entire scheduling system has good flexibility and scalability, and can dynamically adjust algorithm parameters and strategy settings according to actual needs, making it suitable for different types of tasks and environmental changes.

[0085] A Minimum Snap trajectory optimization strategy is adopted for the global planning path, and the speed and acceleration of the inspection robot are optimized using Bezier curves to obtain a smooth and safe trajectory.

[0086] In this embodiment, by optimizing the high-order derivatives of the trajectory, the continuity of the position, velocity, acceleration, and the rate of change of acceleration along the entire path is ensured, thereby avoiding mechanical stress concentration caused by sharp changes and extending the service life of the equipment.

[0087] In this embodiment, a smooth scheduling path is generated by using Bezier curves, which can naturally connect different path segments, making the robot smoother when turning or changing direction, reducing unnecessary pauses and sudden acceleration / deceleration operations. Combined with the TEB strategy, the path can be quickly adjusted when encountering temporary obstacles or other emergencies, ensuring that the robot can safely bypass obstacles and continue to move forward, thereby improving the dynamic response capability and safety of the system.

[0088] The temperature sensors on the inspection robots are used to obtain the temperature of equipment along the track in real time. The working status of each piece of equipment along the track is analyzed based on the temperature of the equipment along the track, and potential overheating risk areas are identified, including:

[0089] Using a high-precision temperature sensor on the inspection robot to perform real-time temperature detection on equipment along the track within a predetermined time interval to obtain temperature data of the equipment along the track;

[0090] The collected temperature data is transmitted to a remote monitoring center in real time through the communication module built into the inspection robot, and the temperature data is analyzed using a supervised learning algorithm to obtain the timestamp, location information, temperature value and environmental parameters contained in each temperature data point;

[0091] In this embodiment, by real-time monitoring and analysis of temperature changes in equipment along the track, an early warning can be issued before the equipment overheats, which helps prevent safety accidents such as fire and equipment damage. The inspection robot can monitor the equipment along the track 24 / 7 to ensure that any abnormal situation can be discovered and handled in a timely manner, enhancing the overall safety of the system.

[0092] Annotate the collected temperature data based on historical temperature data of equipment along the track, and use the annotation results as labels for the supervised learning model to train the corresponding temperature data;

[0093] A feature selection algorithm based on feature correlation is used in the supervised learning model to select the feature with the greatest contribution value to the classification task and obtain temperature feature data;

[0094] In this embodiment, a supervised learning model is used to analyze temperature data, and the features with the greatest contribution to the classification task are screened out based on a feature selection algorithm, thereby improving the prediction accuracy and reliability of the model and making the analysis of temperature data more accurate.

[0095] Classifying the temperature characteristic data according to the annotations, dividing the classified temperature characteristic data into a training set and a data set, and using the training set to train the supervised learning model to optimize model parameters;

[0096] The trained supervised learning model is used to train the data set, and the supervised learning model is deployed to the remote monitoring center to directly process the temperature data, predict the temperature of equipment along the track in real time, mark the temperature, obtain the operating status of equipment along the track, and identify potential overheating risk areas.

[0097] In this embodiment, through real-time analysis of temperature data, the system can dynamically adjust the inspection frequency and path according to the current operating status of the equipment and the temperature change trend, ensuring that resources are concentrated in the areas that need the most attention, reducing unnecessary inspection work, and automated and intelligent task scheduling reduces the need for manual intervention, reduces labor costs, and improves work efficiency and service response speed.

[0098] Based on the location information and urgency of the potential overheating risk areas and the current robot inspection tasks, a multi-robot collaborative inspection task scheduling model is established, including:

[0099] Based on the location information of the potential overheating risk area, an urgency assessment is performed on each identified risk area, and all identified risk areas are sorted according to their urgency to obtain a risk area priority set;

[0100] According to the risk area priority set, the current position, remaining power, and current task progress information of each inspection robot are collected in real time, and whether multiple robots are assigned to similar or identical tasks at the same time node are analyzed, and the inspection tasks of all inspection robots in the current state are obtained;

[0101] In this embodiment, by analyzing temperature data in real time and identifying potential overheating risk areas, the system can quickly dispatch nearby inspection robots to the area for detailed inspection. This greatly shortens the time from problem discovery to action, improves emergency response efficiency, and sorts tasks according to the urgency of the risk area to ensure that the most urgent problems can be handled as quickly as possible, thereby reducing the occurrence of safety hazards.

[0102] For all inspection robots in the current state, the Dijkstra path planning algorithm is used to calculate the optimal path from the current position of the inspection robot to each risk area;

[0103] In this embodiment, the Dijkstra path planning algorithm is used to calculate the optimal path from the robot's current position to each risk area, ensuring that each robot can reach the target location in the shortest time, thereby improving the task execution efficiency. The load balancing strategy is used to reasonably allocate tasks to avoid situations where some robots are overloaded while other robots are idle, thereby improving the stability and resource utilization of the overall system.

[0104] Assign inspection tasks to the inspection robots according to the risk area priority set, and establish a multi-robot collaborative inspection task scheduling model using a load balancing strategy;

[0105] A dynamic adjustment mechanism is set for the multi-robot collaborative inspection task scheduling model to dynamically adjust the task allocation of the inspection robots according to the real-time risk area or the real-time status of the inspection robots.

[0106] In this embodiment, based on the current position, remaining power and current task progress information of each robot, the system can intelligently allocate tasks to avoid wasting resources or over-concentration in a certain area, thereby achieving a more balanced task distribution. The collaborative work between multiple robots can cover a larger range and complete more inspection tasks at the same time. For example, when a large area of overheating is discovered, multiple robots can be dispatched to work simultaneously to speed up the problem investigation.

[0107] Acquiring multi-robot collaborative inspection task scheduling data according to the multi-robot collaborative inspection task scheduling model includes:

[0108] Obtaining a task list for each inspection robot according to the real-time task status of the inspection robot, including the target location, estimated arrival time, and operations to be performed;

[0109] In this embodiment, by monitoring the task status of each inspection robot in real time and obtaining its task list, efficient classification and task scheduling of the inspection robots can be achieved, which helps to ensure that each robot is assigned to the most suitable task and improves overall work efficiency.

[0110] Classify the inspection robots' task scheduling according to the task list, establish a three-dimensional coordinate axis, divide the task scheduling area of multiple robots, and obtain the inspection type of the multi-robot collaborative inspection task;

[0111] In this embodiment, establishing a three-dimensional coordinate axis and dividing the task scheduling area of multiple robots can more accurately locate the task range, avoid task overlap or omission, further optimize task allocation, and improve task allocation efficiency.

[0112] Matching the inspection types of the multi-robot collaborative inspection tasks to inspection robots according to the inspection tasks. When the number of inspection robots is greater than the number of tasks, ensuring that the inspection robots can perform multiple tasks and generating corresponding task sequences;

[0113] In this embodiment, when the number of inspection robots is greater than the number of tasks, it is ensured that each robot can perform multiple tasks and generate corresponding task sequences to maximize the use of existing resources. This not only improves resource utilization, but also enhances the flexibility of the system to adapt to different task requirements. By reasonably arranging task sequences, robots can complete more tasks in the shortest time, reducing idle time and unnecessary waiting, and improving inspection efficiency.

[0114] The task scheduling is completed according to the task sequence, and the timeliness and rationality of the task scheduling are evaluated using an evaluation function. The evaluation function expression is:

[0115]

[0116] Where R represents the robot set consisting of n inspection robots R = {r1, r2, ..., r n}, T represents a task set consisting of m tasks T = {t1, t2, ..., t m}, i represents the i-th inspection robot, j represents the j-th task, Q i represents the task sequence of the i-th inspection robot, s(j,Q i ) is used to determine whether the jth task is in the task sequence Q i In the case of s(j,Q i )=1 means it is in the sequence, otherwise it is not, N T Represents the task scheduling evaluation index;

[0117] In this embodiment, an evaluation function is used to evaluate the timeliness and rationality of task scheduling, providing a quantitative method to measure the effectiveness of task allocation. The evaluation method based on mathematical models is more scientific and accurate than simple empirical judgment, which helps to discover potential problems and adjust strategies in a timely manner.

[0118] In this embodiment, by calculating the task scheduling evaluation index, the quality of the current task allocation plan can be intuitively understood, providing data support for subsequent optimization. For example, if a task is not reasonably allocated, that is, it is not in the task sequence of any robot, the overall scheduling effect can be improved by adjusting the task sequence.

[0119] Real-time multi-robot collaborative task scheduling data is obtained according to the task scheduling evaluation index.

[0120] In this embodiment, a dynamic adjustment mechanism is established based on real-time changes in risk areas or the status of the inspection robot, such as insufficient power or task changes, to dynamically adjust task allocation and ensure the continuity and efficiency of tasks, so that the system can flexibly respond to emergencies and maintain high-efficiency operation.

[0121] The multi-robot collaborative task scheduling data is used to plan the inspection path with the shortest time as the indicator, including:

[0122] Summarize the location coordinates, urgency and inspection priority of all potential risk areas based on the multi-robot collaborative task scheduling data, and obtain the current position, remaining power and task completion status of each inspection robot in real time according to the task sequence;

[0123] In this embodiment, by summarizing the location coordinates, urgency and inspection priority of all potential risk areas in real time, and obtaining the current location and status information of each inspection robot according to the task sequence, the system can respond quickly and use the Dijkstra algorithm to find the shortest path to ensure that the robot can reach the target location as quickly as possible, thereby improving the efficiency of emergency handling, sorting tasks according to the urgency of the risk area, ensuring that the most urgent problems can be handled as quickly as possible, and reducing the occurrence of safety hazards.

[0124] Loading real-time rail transit map data, taking the current position of each inspection robot as the starting point, and determining the target destination of each task according to the priority ranking of the risk areas;

[0125] In this embodiment, based on the current position, remaining power, and current task progress information of each robot, the system can intelligently allocate tasks to avoid wasting resources or over-concentration in a certain area, thereby achieving more balanced task distribution.

[0126] The Dijkstra algorithm is used to find the shortest path with the shortest time as the indicator. A conflict detection mechanism is introduced to identify the possible cross paths or time overlaps of different robots during the execution of tasks, and the inspection path is planned in real time.

[0127] In this embodiment, a conflict detection mechanism is introduced to identify possible cross paths or time overlaps between different robots during task execution, thereby ensuring efficiency and safety during task execution and reducing delays caused by task conflicts.

[0128] The task scheduling based on the shortest time is a task scheduling process implemented based on an auction algorithm, wherein the inspection robot performing the task is regarded as the auctioneer, the task to be performed is regarded as the auctioned item, and the person who publishes the task information is regarded as the initiator of the auction;

[0129] Each inspection robot uses the constraints of highest mission benefit, lowest mission cost, and shortest mission time as bidding indicators. The robots exchange information, negotiate, and discuss with each other to determine who will take on a particular mission.

[0130] The task benefit index is set as the benefit value of each inspection task point given to the inspection robot before the inspection robot performs task scheduling according to the actual environment requirements;

[0131] The task cost indicator is set as the remaining power value of the inspection robot, the location of the inspection task point, and the distance of the inspection robot's current task, and its expression is:

[0132] W cost =d cost +E cost

[0133] Among them, W cost represents the cost of the inspection robot executing the task sequence, d cost Indicates the distance traveled by the inspection robot when performing the task, E cost Indicates the amount of power consumed by the inspection robot when performing tasks.

[0134] In this embodiment, task scheduling is performed through an auction algorithm. Each inspection robot bids based on its own status, such as remaining power and distance, and selects the task that is most suitable for itself to perform, thereby achieving load balancing and avoiding situations where some robots are overloaded while others are idle.

[0135] In this embodiment, the method of planning the inspection route based on the shortest time as the indicator and implementing the task scheduling process based on the auction algorithm not only greatly improves the safety and reliability of rail transit facilities, but also optimizes resource utilization and maintenance processes, and reduces operating costs.

[0136] The auction algorithm implements a task scheduling process, including:

[0137] In the task scheduling process, the number of tasks is set to z, the number of inspection robots that perform the tasks is n, and multiple tasks are auctioned simultaneously;

[0138] Calculate the maximum profit obtained when the inspection robot acts as an auctioneer, and schedule the tasks of multiple robots in real time based on the maximum profit value. The maximum profit expression is:

[0139]

[0140] Among them, i represents the i-th inspection robot, j represents the j-th task, and p ij represents the revenue value of the i-th auctioneer inspection robot performing the j-th task, v ijrepresents the bidding price of the i-th auctioneer inspection robot for the j-th task, x ij Indicates whether the jth task is assigned to the i-th auctioneer inspection robot.

[0141] In this embodiment, the task ownership is determined by the auction price, ensuring that each task is assigned to the inspection robot that can bring the greatest benefit. This not only improves the efficiency of the overall system, but also ensures the effective use of resources. All inspection robots have the opportunity to participate in the auction and make bids based on their own conditions such as remaining power and current location, ensuring fairness and transparency in the task allocation process and improving the efficiency and fairness of task allocation.

[0142] The inspection path is optimized by using the fusion improved ant colony algorithm combined with the TEB strategy to obtain the global planning path, including:

[0143] Modeling the rail transit environment based on the inspection route, initializing the coordinate information of each target node using a grid map, obtaining track, station and obstacle location information, and defining the key nodes of the inspection path;

[0144] The concentration level of pheromones on each inspection route is defined based on the location coordinates, urgency and inspection priority of all potential risk areas, and a possible path chain is gradually formed by randomly selecting destinations using ant colonies;

[0145] In this embodiment, by integrating the improved ant colony algorithm, destinations are randomly selected to gradually form a possible path chain, and the efficiency index of the path taken by each ant is calculated to update the path attraction between adjacent target points. It is possible to quickly find an efficient inspection path in a complex environment, and define the pheromone concentration level on each inspection route according to the location coordinates, urgency and inspection priority of all potential risk areas, so that the system can give priority to exploring areas that need more attention, thereby improving the pertinence and effectiveness of path planning.

[0146] Calculate the efficiency index of the path taken by each ant, update the path attraction between adjacent target points, and iterate and record the optimal path. The probability of selecting the optimal path depends on the pheromone concentration and the predicted benefit of the path, and its expression is:

[0147]

[0148] Among them, τ xy (t) represents the pheromone concentration of the path (x, y) from node x to node y at time t, δ xy (t) represents the heuristic function of the ant from the target node x to the node y. The heuristic function is related to the Euclidean distance between the target node x and the node y. o(x) represents the set of target points that ant o is allowed to choose in the next step, α and β represent the relative importance weights of pheromone and heuristic factor, u represents any node on the path, τ xu (t) represents the pheromone concentration of the path (x,u) from node x to node u at time t, δ yu (t) represents the heuristic function of the ant from the target node x to the node u, which is related to the Euclidean distance between the target node x and the node u.

[0149] In this embodiment, the predicted benefit is achieved by predicting the benefit between the ant's target node x and any node y, taking into account factors such as pheromone concentration, Euclidean distance between nodes, and path smoothness, and dynamically adjusting the path linking mode.

[0150] In this embodiment, the TEB strategy is mainly used to solve the problem of path smoothing. By adjusting the connection method between path points, a smoother and continuous path is generated. This not only reduces sharp turns and unnecessary pauses, but also reduces the mechanical stress concentration caused by sudden turns or acceleration and deceleration during the robot's mission, thereby extending the service life of the equipment. The TEB strategy can adjust the path in real time according to environmental changes, avoid the impact of temporary obstacles or other emergencies on the inspection task, and ensure that the robot completes the inspection task safely and efficiently.

[0151] The global planning path is optimized based on the Minimum Snap trajectory strategy, including:

[0152] The global path planning is divided into several segments according to the predicted benefits of the path, each segment consists of a starting point, an end point and intermediate key control points;

[0153] Assign a time variable t to each trajectory segment, define the initial and final states of each trajectory segment, including position, velocity, and acceleration, and use a polynomial function to represent the change of the path position node over time. The expression is:

[0154] q(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4

[0155] Among them, a0, a1, a2, a3 and a4 all represent polynomial coefficients;

[0156] In this embodiment, a polynomial function is used to represent the changes in path position nodes over time, so that the trajectory connects different path segments more naturally, reducing sharp turns or sudden acceleration / deceleration operations, thereby improving the dynamic performance and operational safety of the robot. Combined with the TEB strategy or other dynamic adjustment mechanisms, the path can be quickly adjusted when encountering temporary obstacles or other emergencies, ensuring that the robot can safely bypass the obstacles and continue to move forward, thereby improving the dynamic response capability and safety of the system.

[0157] The Minimum Snap objective function is used to optimize the trajectory equation. The expression of the Minimum Snap objective function is:

[0158]

[0159] in, represents the fourth-order derivative of the trajectory, and T represents the time constant of the entire trajectory optimization;

[0160] Constraints are established on the speed and acceleration of the inspection robot to ensure the continuity between adjacent trajectories.

[0161] In this embodiment, the Minimum Snap objective function is used to generate a trajectory that is as smooth as possible. Compared with traditional methods that only consider position, velocity or acceleration, this strategy can significantly reduce the vibration and energy loss of the robot during movement, enhance the global search capability, avoid the local optimal solution problem that may occur in traditional methods, improve the reliability and adaptability of path planning, and realize multi-objective optimization, thereby better adapting to the needs of different application scenarios.

[0162] The speed and acceleration of the inspection robot are optimized using Bezier curves to construct constraint conditions and obtain a smooth and safe trajectory, including:

[0163] The starting point, end point and intermediate key control points on the path trajectory are connected in sequence using Bezier curves, and control points that affect the predicted benefits are randomly selected on the path trajectory. The position and number of the control points can be changed. The Bezier curve expression is constructed as follows:

[0164]

[0165] Where n represents the order of the curve, θ represents the Bernstein coefficient of the Bezier curve, and P b represents the control point of the control path trajectory, b represents the control node on any trajectory path, and t represents the time variable;

[0166] In this embodiment, by randomly selecting key control points that affect the predicted benefits and adjusting their positions and numbers, the system can optimize local path details according to actual conditions while maintaining the overall path efficiency, thereby enhancing the flexibility and adaptability of path planning.

[0167] The first-order derivative and multi-order derivative of the Bezier curve are respectively performed to obtain the recursive relationship between the control points on the trajectory path, and the speed and acceleration of the inspection robot are calculated in real time.

[0168] In this embodiment, by performing first-order derivative of the Bezier curve to obtain speed, second-order derivative to obtain acceleration, and even higher-order derivatives, it is possible to ensure that the dynamic characteristics of the trajectory such as speed and acceleration at each control point remain continuous, thereby avoiding mechanical stress concentration or control system instability caused by discontinuous changes.

[0169] In this embodiment, the path trajectory constructed using the Bezier curve can ensure that the robot has smooth speed and acceleration changes throughout the entire movement process, reduce vibration and impact, and improve the safety and comfort of operation. The position and number of control points can be adjusted according to actual needs, so that the path trajectory can be flexibly modified according to environmental changes or task requirements to adapt to different application scenarios.

[0170] In this embodiment, based on the real-time calculated speed and acceleration information, combined with the TEB strategy or other dynamic adjustment mechanisms, the path can be quickly adjusted when encountering temporary obstacles or other emergencies, ensuring that the robot can safely bypass the obstacles and continue to move forward, thereby improving the dynamic response capability and safety of the system.

[0171] The present invention uses high-precision temperature sensors to monitor equipment temperature in real time and uses supervised learning algorithms to analyze and identify potential overheating risk areas. It adopts a multi-robot collaborative task scheduling model combined with an auction algorithm to dynamically allocate tasks, ensuring efficient resource utilization and prioritizing urgent tasks. It uses an improved ant colony algorithm and a time dilation and time-dilation (TEB) strategy to optimize global path planning, and achieves smooth and safe motion trajectories through Minimum Snap trajectory optimization and Bezier curves. This not only improves inspection efficiency and safety, but also reduces operating costs, providing an intelligent, automated and efficient solution for the maintenance and management of rail transit facilities.

[0172] The above embodiments are merely exemplary embodiments of the present application and are not intended to limit the scope of the present application. The scope of protection of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and scope of protection of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the scope of protection of the present application.

Claims

1. A real-time temperature measurement and optimization scheduling method for a track-type intelligent inspection robot, characterized in that: The following steps are involved: The temperature of equipment along the track is obtained in real time through the temperature sensor on the inspection robot. The working status of each equipment along the track is analyzed based on the temperature of the equipment along the track, and potential overheating risk areas are identified; Based on the location information and urgency of the potential overheating risk area and the inspection tasks of the current robots, a multi-robot collaborative inspection task scheduling model is established to obtain multi-robot collaborative task scheduling data; The multi-robot collaborative task scheduling data is used to plan the inspection path with the shortest time as the indicator, and the inspection path is optimized by using the fusion improved ant colony algorithm combined with the TEB strategy to obtain the global planning path; A Minimum Snap trajectory optimization strategy is adopted for the global planning path, and the speed and acceleration of the inspection robot are optimized using Bezier curves to obtain a smooth and safe trajectory.

2. A method for real-time temperature measurement and optimized scheduling of a track-type intelligent inspection robot according to claim 1, characterized in that: The temperature sensors on the inspection robots are used to obtain the temperature of equipment along the track in real time. The working status of each piece of equipment along the track is analyzed based on the temperature of the equipment along the track, and potential overheating risk areas are identified, including: Using a high-precision temperature sensor on the inspection robot to perform real-time temperature detection on equipment along the track within a predetermined time interval to obtain temperature data of the equipment along the track; The collected temperature data is transmitted to a remote monitoring center in real time through the communication module built into the inspection robot, and the temperature data is analyzed using a supervised learning algorithm to obtain the timestamp, location information, temperature value and environmental parameters contained in each temperature data point; Annotate the collected temperature data based on historical temperature data of equipment along the track, and use the annotation results as labels for the supervised learning model to train the corresponding temperature data; A feature selection algorithm based on feature correlation is used in the supervised learning model to select the feature with the greatest contribution value to the classification task and obtain temperature feature data; Classifying the temperature characteristic data according to the annotations, dividing the classified temperature characteristic data into a training set and a data set, and using the training set to train the supervised learning model to optimize model parameters; The trained supervised learning model is used to train the data set, and the supervised learning model is deployed to the remote monitoring center to directly process the temperature data, predict the temperature of equipment along the track in real time, mark the temperature, obtain the operating status of equipment along the track, and identify potential overheating risk areas.

3. A method for real-time temperature measurement and optimized scheduling of a track-type intelligent inspection robot according to claim 2, characterized in that: Based on the location information and urgency of the potential overheating risk areas and the current robot inspection tasks, a multi-robot collaborative inspection task scheduling model is established, including: Based on the location information of the potential overheating risk area, an urgency assessment is performed on each identified risk area, and all identified risk areas are sorted according to their urgency to obtain a risk area priority set; According to the risk area priority set, the current position, remaining power, and current task progress information of each inspection robot are collected in real time, and whether multiple robots are assigned to similar or identical tasks at the same time node are analyzed, and the inspection tasks of all inspection robots in the current state are obtained; For all inspection robots in the current state, the Dijkstra path planning algorithm is used to calculate the optimal path from the current position of the inspection robot to each risk area; Assign inspection tasks to the inspection robots according to the risk area priority set, and establish a multi-robot collaborative inspection task scheduling model using a load balancing strategy; A dynamic adjustment mechanism is set for the multi-robot collaborative inspection task scheduling model to dynamically adjust the task allocation of the inspection robots according to the real-time risk area or the real-time status of the inspection robots.

4. A method for real-time temperature measurement and optimized scheduling of a track-type intelligent inspection robot according to claim 3, characterized in that: Acquiring multi-robot collaborative inspection task scheduling data according to the multi-robot collaborative inspection task scheduling model includes: Obtaining a task list for each inspection robot according to the real-time task status of the inspection robot, including the target location, estimated arrival time, and operations to be performed; Classify the inspection robots' task scheduling according to the task list, establish a three-dimensional coordinate axis, divide the task scheduling area of multiple robots, and obtain the inspection type of the multi-robot collaborative inspection task; Matching the inspection types of the multi-robot collaborative inspection tasks to inspection robots according to the inspection tasks. When the number of inspection robots is greater than the number of tasks, ensuring that the inspection robots can perform multiple tasks and generating corresponding task sequences; The task scheduling is completed according to the task sequence, and the timeliness and rationality of the task scheduling are evaluated using an evaluation function. The evaluation function expression is: Where R represents the robot set consisting of n inspection robots R = {r1, r2, ..., r n }, T represents a task set consisting of m tasks T = {t1, t2, ..., t m }, i represents the i-th inspection robot, j represents the j-th task, Q i represents the task sequence of the i-th inspection robot, s(j,Q i ) is used to determine whether the jth task is in the task sequence Q i In the case of s(j,Q i )=1 means it is in the sequence, otherwise it is not, N T Represents the task scheduling evaluation index; Real-time multi-robot collaborative task scheduling data is obtained according to the task scheduling evaluation index.

5. A method for real-time temperature measurement and optimized scheduling of a track-mounted intelligent inspection robot according to claim 4, characterized in that: The multi-robot collaborative task scheduling data is used to plan the inspection path with the shortest time as the indicator, including: Summarize the location coordinates, urgency and inspection priority of all potential risk areas based on the multi-robot collaborative task scheduling data, and obtain the current position, remaining power and task completion status of each inspection robot in real time according to the task sequence; Loading real-time rail transit map data, taking the current position of each inspection robot as the starting point, and determining the target destination of each task according to the priority ranking of the risk areas; The Dijkstra algorithm is used to find the shortest path with the shortest time as the indicator. A conflict detection mechanism is introduced to identify the possible cross paths or time overlaps of different robots during the execution of tasks, and the inspection path is planned in real time.

6. A method for real-time temperature measurement and optimized scheduling of a track-mounted intelligent inspection robot according to claim 5, characterized in that: include: The task scheduling based on the shortest time is a task scheduling process implemented based on an auction algorithm, wherein the inspection robot performing the task is regarded as the auctioneer, the task to be performed is regarded as the auctioned item, and the person who publishes the task information is regarded as the initiator of the auction; Each inspection robot uses the constraints of highest mission benefit, lowest mission cost, and shortest mission time as bidding indicators. The robots exchange information, negotiate, and discuss with each other to determine who will take on a particular mission. The task benefit index is set as the benefit value of each inspection task point given to the inspection robot before the inspection robot performs task scheduling according to the actual environment requirements; The task cost indicator is set as the remaining power value of the inspection robot, the location of the inspection task point, and the distance of the inspection robot's current task, and its expression is: W cost =d cost +E cost Among them, W cost represents the cost of the inspection robot executing the task sequence, d cost Indicates the distance traveled by the inspection robot when performing the task, E cost Indicates the amount of power consumed by the inspection robot when performing tasks.

7. A method for real-time temperature measurement and optimized scheduling of a track-mounted intelligent inspection robot according to claim 6, characterized in that: The auction algorithm implements a task scheduling process, including: In the task scheduling process, the number of tasks is set to z, the number of inspection robots that perform the tasks is n, and multiple tasks are auctioned simultaneously; Calculate the maximum profit obtained when the inspection robot acts as an auctioneer, and schedule the tasks of multiple robots in real time based on the maximum profit value. The maximum profit expression is: Among them, i represents the i-th inspection robot, j represents the j-th task, and p ij represents the revenue value of the i-th auctioneer inspection robot performing the j-th task, v ij represents the bidding price of the i-th auctioneer inspection robot for the j-th task, x ij Indicates whether the jth task is assigned to the i-th auctioneer inspection robot.

8. The method for real-time temperature measurement and optimized scheduling of a track-mounted intelligent inspection robot according to claim 6, characterized in that: The inspection path is optimized by using the fusion improved ant colony algorithm combined with the TEB strategy to obtain the global planning path, including: Modeling the rail transit environment based on the inspection route, initializing the coordinate information of each target node using a grid map, obtaining track, station and obstacle location information, and defining the key nodes of the inspection path; The concentration level of pheromones on each inspection route is defined based on the location coordinates, urgency and inspection priority of all potential risk areas, and a possible path chain is gradually formed by randomly selecting destinations using ant colonies; Calculate the efficiency index of the path taken by each ant, update the path attraction between adjacent target points, and iterate and record the optimal path. The probability of selecting the optimal path depends on the pheromone concentration and the predicted benefit of the path, and its expression is: Among them, τ xy (t) represents the pheromone concentration of the path (x, y) from node x to node y at time t, δ xy (t) represents the heuristic function of the ant from the target node x to the node y. The heuristic function is related to the Euclidean distance between the target node x and the node y. o (x) represents the set of target points that ant o is allowed to choose in the next step, α and β represent the relative importance weights of pheromone and heuristic factor, u represents any node on the path, τ xu (t) represents the pheromone concentration of the path (x,u) from node x to node u at time t, δ yu (t) represents the heuristic function of the ant from the target node x to the node u, which is related to the Euclidean distance between the target node x and the node u.

9. A method for real-time temperature measurement and optimized scheduling of a track-mounted intelligent inspection robot according to claim 8, characterized in that: The global planning path is optimized based on the Minimum Snap trajectory strategy, including: The global path planning is divided into several segments according to the predicted benefits of the path, each segment consists of a starting point, an end point and intermediate key control points; Assign a time variable t to each trajectory segment, define the initial and final states of each trajectory segment, including position, velocity, and acceleration, and use a polynomial function to represent the change of the path position node over time. The expression is: q(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 Among them, a0, a1, a2, a3 and a4 all represent polynomial coefficients; The Minimum Snap objective function is used to optimize the trajectory equation. The expression of the Minimum Snap objective function is: in, represents the fourth-order derivative of the trajectory, and T represents the time constant of the entire trajectory optimization; Constraints are established on the speed and acceleration of the inspection robot to ensure the continuity between adjacent trajectories.

10. A method for real-time temperature measurement and optimized scheduling of a track-mounted intelligent inspection robot according to claim 9, characterized in that: The speed and acceleration of the inspection robot are optimized using Bezier curves to construct constraint conditions and obtain a smooth and safe trajectory, including: The starting point, end point and intermediate key control points on the path trajectory are connected in sequence using Bezier curves, and control points that affect the predicted benefits are randomly selected on the path trajectory. The position and number of the control points can be changed. The Bezier curve expression is constructed as follows: Where n represents the order of the curve, θ represents the Bernstein coefficient of the Bezier curve, and P b represents the control point of the control path trajectory, b represents the control node on any trajectory path, and t represents the time variable; The first-order derivative and multi-order derivative of the Bezier curve are respectively performed to obtain the recursive relationship between the control points on the trajectory path, and the speed and acceleration of the inspection robot are calculated in real time.

Images