CN-122022254-A - Multi-agent cooperative concrete building water spraying maintenance path scheduling algorithm

Abstract

The multi-agent cooperative concrete building water spraying maintenance path scheduling algorithm dynamically adjusts the maintenance priority scores of the grid units according to preset rules by constructing a dynamic priority map composed of regular hexagonal grid units and utilizing a temperature and humidity sensor to monitor environmental parameters and structural states of the building surfaces in real time. The system adopts a dual trigger mechanism to generate a maintenance task list, and performs centralized decision and distributed task allocation through a lightweight central scheduler, so that resource optimal allocation and efficient job path planning are realized. In addition, the monomer robot is provided with a fuzzy PID controller to accurately control the moving speed and the opening and closing of the water spraying valve, so that the accurate water quantity is sprayed at the correct position, and the accurate maintenance is realized. The system also has a closed-loop feedback function, and can update the global grid priority map in real time according to the execution condition, continuously optimize the maintenance strategy and improve the building maintenance efficiency and the intelligent level.

Inventors

• SHEN YUE
• CHENG LE
• WU YUFAN
• LI YUE
• WANG KAI
• WANG ZHENGDONG
• MENG XIANGYIN
• LIANG BIN
• JIANG ZHIJUN
• YAO KANG
• Zhai Haobo
• CHEN KE
• ZU YU

Assignees

• 中交建筑集团有限公司
• 中交建筑集团第一工程有限公司

Dates

Publication Date

20260512

Application Date

20251225

Claims (4)

1. 1. The concrete building water spray maintenance path scheduling algorithm with the cooperation of multiple intelligent agents comprises the following specific steps of: step 1, generating a data base and a task; based on BIM and structural analysis data, constructing a building surface dynamic maintenance priority grid map; step 2, centralized decision and distributed allocation; establishing a lightweight central scheduler, grasping a global state by the scheduler, and periodically executing task allocation work and conflict resolution work; step3, controlling a single robot; A fuzzy PID controller is adopted to accurately control the moving speed of the robot and the opening and closing of the water spraying valve, so that the correct water quantity is sprayed at the correct position; Step 4, closed loop feedback and self-adaptive optimization; The scheduler updates the global grid priority map according to the actual water spraying quantity and actual humidity, and starts task allocation and planning of the next round to form a closed loop of sensing, decision making, executing and feedback.
2. 2. The multi-agent collaborative concrete building water spray curing path scheduling algorithm according to claim 1, wherein the data base and task generation in step 1 can be expressed as: the method comprises the steps of taking regular hexagonal grids as basic units, setting the side length of each grid to be 1m, and ensuring that the operation ranges of adjacent grids are not overlapped and omitted, collecting the environment and structure state data of the building surface in real time through temperature and humidity by using a four-level coding system of region numbers, floors, grid line numbers and grid column numbers, dynamically adjusting the grid priority according to priority scores, and ensuring that maintenance requirements are synchronous with the actual state of a component; the maintenance task generation adopts a threshold trigger and period trigger double mechanism: Triggering a threshold value, namely automatically generating a maintenance task by the system when the grid unit priority score is more than or equal to 70; And (3) periodically triggering, namely executing 1 full map grid scanning every 1 hour, and generating an early warning maintenance task for grid units with priority scores between 60-69 partitions if scores have no descending trend in 3 continuous updating periods, so as to avoid potential maintenance risk accumulation.
3. 3. The multi-agent cooperative concrete building water spray maintenance path scheduling algorithm of claim 1, wherein the centralized decision and the distributed allocation in the step 2 can be represented as follows: the task allocation work is defined as adopting bidding mechanism based on marginal cost, broadcasting new task or overload area task to idle or low-load robot, receiving cost quotation calculated based on self state and allocating to bidder with lowest bidding; the conflict resolution is defined as planning paths of all robots in advance, detecting potential collision points and coordinating; step 2.1, designing a lightweight central scheduler architecture; The central dispatcher is a decision core and is divided into a task management module, a global state database, a robot state monitoring module, a bidding management module, a path planning module and a conflict coordination module according to functional modules; the task management module uniquely receives the maintenance task list generated in the step 1 and synchronizes the list to the global state database; the global state database of the central dispatcher stores two types of core data, namely a building component maintenance basic database generated in the step 1 and a grid priority map updated in real time; The robot state monitoring module synchronizes the monitored robot state to a global state database for being called by the bidding management module, and provides a basis for screening basic conditions of task allocation; the bidding management module screens tasks to be allocated according to priority from the list received by the task management module, so that the optimal matching of the tasks and the robot is realized; The path planning module adopts an A-algorithm to plan a safe and efficient global path for the robot with the assigned tasks, so that the robot can accurately reach the task grid; the collision coordination module quantitatively judges potential collision based on a time window and a space distance, detects and counteracts robot path collision in advance, and ensures the safety of cooperative operation of multiple robots; step 2.2, a task allocation mechanism based on marginal cost; Task allocation aims at minimizing global operation cost, scheduler broadcasting, robot quotation and scheduler optimization flow are adopted, and the core is quantitative calculation of marginal cost, and task and grid information in the step 1 are combined; Step 2.2.1, defining a marginal cost model and a formula; Marginal cost of When the robot i receives the task j, the additional operation cost is added, and the space distance between the robot and the task, the current load and the energy consumption state are calculated according to the following formula: ; In the formula, The marginal cost of the task j is quoted for the robot i, and the unit is an element; The value of the distance weight is 0.6; The value of the load weight is 0.3; the energy consumption weight is 0.1; For the linear distance from the current position of the robot i to the center of the grid to which the task j belongs, the unit is m, and after resolving the coordinates by the grid codes in the step 1, the linear distance is calculated by an Euclidean distance formula: ; In the formula, The three-dimensional coordinates of the current position of the robot i; The three-dimensional coordinates of the grid center to which the task j belongs; the current load rate of the robot i is calculated by the following formula: ; In the formula, The number of tasks assigned to the robot i is, The maximum load task number of the robot i; The calculation formula is as follows, wherein the unit of the normalized value of the residual electric quantity of the robot i is: ; In the formula, For the remaining power of the robot i, Rated power for the robot i; step 2.2.2 task allocation execution step: Step 2.2.2.1 task screening and broadcasting; The dispatcher screens out tasks to be distributed from the task list generated in the step 1, preferentially screens out tasks with priority P1, screens out P2 and P3, and obtains task priorities according to priority grading and sequencing, and broadcasts task information to robots meeting basic conditions, wherein the task information comprises target water spraying amount, task ID, associated grid codes and corresponding three-dimensional coordinates Taking the grid map from the step 1; basic condition is robot load factor <50% Of the residual quantity More than or equal to 30 percent of distance task grid ≤50m; Step 2.2.2.2 robot quotation calculation; After receiving the broadcast, the robot meeting the basic conditions is based on the current coordinates of the robot 、 、 Calculation of And will offer results The self ID and the current state are fed back to the dispatcher; step 2.2.2.3, selecting and distributing tasks by a scheduler; After the scheduler collects all valid offers, it selects The smallest robot is used as a task receiver, and an allocation instruction is sent in a unicast mode, wherein the instruction comprises complete information of a task; If present The same robots are secondarily ordered according to the task priority and the robot load rate, namely, the tasks are preferentially allocated to the P1 task, and the tasks are selected under the same priority A dispatcher updates a global state database after distribution is completed, and a robot Adding 1, wherein the task ID is newly added to the allocated task list; step 2.2.2.4, calculating target water spraying amount; After the task is constructed, the reference value of the target water spray amount of the single grid is defined as follows: The standard water spraying amount of the concrete column in the curing period is 5L/m <2 >, the standard water spraying amount of the non-bearing brick wall is 3L/m <2 >, and the target water spraying amount correction formula of the single grid is calculated as follows: ; In the formula, The final target water spraying amount; Maintaining a target humidity for the component; the current humidity of the grid is acquired in real time for the sensor; The water spray amount is the reference water spray amount; step 2.3, path planning and conflict resolution; The core of conflict resolution is that potential collisions are detected and coordinated by planning a robot path in advance, and quantitative management and control are realized by combining the motion characteristics of the robot based on grid map coordinates in the step 1; Step 2.3.1 global path planning algorithm; The scheduler plans a global path for each robot with assigned tasks, the path needs to cover the current position of the robot, a task grid and a subsequent potential task grid, an A-algorithm is adopted, an objective function is used for minimizing the path length and minimizing the turning times, and a path planning formula is as follows: ; In the formula, The total cost at the center of the node n grid cell; for the actual path length from the current position of the robot to the node n, the calculation mode is the sum of Euclidean distances of adjacent nodes in the path; Taking Euclidean distance as a heuristic function of a slave node n to a task grid center; The constraint condition is that the ① path needs to avoid building walls and equipment barriers, and the coordinates of the equipment barriers are extracted from BIM geometric data in the step 1 and are set as non-passable nodes, the turning angle of the ② adjacent path section is less than or equal to 90 degrees, the robot is prevented from turning suddenly, and the passing width of each node in the ③ path is more than or equal to the width of the robot; Step 2.3.2 potential collision detection models; the collision is defined as that two robots are positioned on the same grid or adjacent grids in the same time window, the space distance is smaller than a safety threshold, and the two dimensions of time and space to be quantized are detected: Time window calculation, travel time of robot k from node n to node m The calculation formula is as follows: ; In the formula, Distance from node n to m; is the rated moving speed of the robot k; as a speed decay factor, associated with the robot load factor, , Based on the formula, a time window for robot k to reach each node n in the path can be calculated Wherein For the arrival time of the robot k node n, For the robot k node n departure time, Next is the next node; collision determination condition calculation for any two robots 、 If the node n satisfies the following conditions, the potential collision is determined: ; In the formula, Is a robot The straight line distance to the node n, Is a robot The straight line distance to the node n, The safety distance is 1.5 times of the width of the robot, so that scratch is avoided; Representing a time window intersection; And Respectively represent robots And The arrival time at node n, And Respectively represent robots And Departure time from node n; Step 2.3.3 conflict coordination policies; for the detected potential collision, the scheduler adopts a priority and time offset coordination strategy, and a priority quantification formula is as follows: ; In the formula, A coordination priority for robot k; a priority score p1=100, p2=60, p3=30 for the current task of robot k; =0.6、 =0.2、 =0.2 is a weight; 、 the load rate and the residual electric quantity of the robot k are respectively; The coordination executing step: ① Collision calculating robot 、 A kind of electronic device 、 Robot with high priority Maintaining an original path time window; ② For robots with low priority Adjust its time window to reach collision node n, offset The calculation formula is as follows: ; ③ Based on Recalculation of Is updated to the robot And the global state database ensures that the path has no collision.
4. 4. The multi-agent cooperative concrete building water spray maintenance path scheduling algorithm of claim 1, wherein the single robot control in the step 3 can be represented as follows: A fuzzy PID controller is adopted to accurately control the moving speed of the robot and the opening and closing of the water spraying valve, so that the correct water quantity is sprayed at the correct position; The maintenance robot motion control needs to realize accurate adjustment of the moving speed and accurate opening and closing of the water spraying valve, a fuzzy PID controller is adopted in the core, and closed-loop control is constructed by combining the positioning deviation fed back by the sensor in the step 3 and water flow data, so that the task requirements in the step 1 are met; Step 3.1, fuzzy PID control of the moving speed; step 3.1.1 PID basic control formula; The PID output quantity of the robot moving speed control is a rotating speed instruction u (t) of a motor, and the calculation formula is as follows: ; Wherein e (t) is the positional deviation at time t, , For the x-coordinate of the path-target point, Positioning an x coordinate for the robot in real time; is a proportionality coefficient; is an integral coefficient; Is a differential coefficient; Step 3.1.2, optimizing fuzzy control parameters; the traditional PID parameters are difficult to adapt to uneven ground and load change of building environment, and are dynamically adjusted through fuzzy control , , : The fuzzy input quantity is the position deviation e (t), the quantization range is [ -10cm,10cm ], the fuzzy subset is { Negative Big (NB), negative Medium (NM), negative Small (NS), zero (Z), positive Small (PS), medium (PM), positive Big (PB) }, the deviation change rate Quantization range [ -5cm/s,5cm/s ], fuzzy subset with e (t); fuzzy output is , , The quantization ranges are respectively [ -2,2], [ -0.5,0.5], [ -0.1,0.1], and the fuzzy subset is input together; Fuzzy rule table when e (t) =pb, ec (t) =pb, the deviation needs to be reduced rapidly, so =PB、 =NB、 =PS; Outputting corrected PID parameters through fuzzy reasoning and definition: ; step3.2, fuzzy PID control of the water spraying valve; the valve control takes actual water spray amount as a target, and water flow sensor data construct a closed loop: step 3.2.1, calculating the deviation of the water spraying amount; target water jet quantity The actual water spraying quantity is obtained from a task list in the step 1 The water flow sensor collects the water flow in real time, and the water spray quantity is deviated ; Step 3.2.2 valve opening control formula; Valve opening command The range is 0-100%, and the fuzzy PID control formula from full closing to full opening corresponds to the valve: ; Fuzzy optimization logic And For input, dynamically adjusting , 、 Realizing the accurate control of the water spraying quantity.

Description

Multi-agent cooperative concrete building water spraying maintenance path scheduling algorithm Technical Field The invention relates to the technical field of maintenance robot control, in particular to a multi-agent cooperative concrete building water spraying maintenance path scheduling algorithm. Background In the current construction industry, building surface maintenance work is often dependent on manual inspection and repair. This approach is time consuming and laborious, and due to human factors, may lead to missed or inaccurate diagnosis of problems, thereby affecting the safety and aesthetics of the building structure. In addition, with the acceleration of the urban process and the continuous increase of the building height, the maintenance difficulty of high-rise buildings and complex structures is gradually increased, and the requirements of modern buildings are difficult to meet by the traditional manual maintenance mode. In recent years, with the development of sensor technology, robotics and information technology, the concept of intelligent building maintenance has been attracting attention. For example, some studies have attempted to apply Unmanned Aerial Vehicles (UAVs) to the inspection of building surfaces by capturing high resolution images to identify damage conditions such as cracks, spalling, etc. However, these methods are mostly limited to the data acquisition stage, lack of effective data analysis and automatic repair strategies, and fail to implement integrated solutions from monitoring to repair. In addition, the application of Building Information Model (BIM) technology brings revolutionary changes to building engineering management. BIM not only can provide three-dimensional digital model, but also can integrate multidimensional information such as time dimension (4D), cost dimension (5D) and the like, thereby greatly improving the working efficiency of each stage of project planning, design, construction and operation and maintenance. However, existing BIM applications are mainly focused on early stages of projects, and have shortcomings in dynamic monitoring and maintenance during building operations. In view of the above background, the present invention aims to provide a dynamic maintenance system combining a BIM model, a real-time sensor network, an automated robot technology and an intelligent algorithm, so as to achieve efficient and accurate maintenance of a building surface. The system not only can automatically identify the damage condition of the building surface, but also can automatically generate and execute a maintenance scheme according to the analysis result, thereby greatly improving the quality and efficiency of maintenance work, reducing the labor cost and ensuring the safety and durability of the building structure. Disclosure of Invention In order to solve the problems, the invention provides a concrete building water spraying maintenance path scheduling algorithm with multi-agent cooperation, which comprises the following specific steps: Step 1, data base and task generation Based on BIM and structural analysis data, constructing a building surface dynamic maintenance priority grid map; step 2-centralized decision and distributed distribution Establishing a lightweight central scheduler, grasping a global state by the scheduler, and periodically executing task allocation work and conflict resolution work; Step 3, monomer robot control A fuzzy PID controller is adopted to accurately control the moving speed of the robot and the opening and closing of the water spraying valve, so that the correct water quantity is sprayed at the correct position; step 4, closed loop feedback and adaptive optimization The scheduler updates the global grid priority map according to the actual water spraying quantity and actual humidity, and starts task allocation and planning of the next round to form a closed loop of sensing, decision making, executing and feedback. The multi-agent cooperative concrete building water spraying maintenance path scheduling algorithm has the beneficial effects that: 1. The invention can automatically identify the damage condition of the building surface and quickly respond and execute the repair task by integrating the real-time sensor network and the automatic robot technology. Compared with the traditional manual inspection and maintenance mode, the time period from problem discovery to solution is greatly shortened. 2. The method and the device analyze BIM model and sensor data based on intelligent algorithm, and can more accurately locate the damage position and evaluate the damage degree. Meanwhile, the automatically generated maintenance scheme ensures the pertinence and the effectiveness of maintenance work, and reduces errors caused by artificial factors. 3. The invention can be integrated with other intelligent management systems to realize information sharing and resource integration, further optimize project management flow and improv