Dynamic mutual-assistance intelligent scheduling system and method for vehicles

Abstract

The invention discloses a dynamic mutual-assistance intelligent scheduling system and method for vehicles. According to the invention, through three core modules of initial scheduling, fault monitoring and collaborative decision-making, and scheme execution closed loop, and in combination with a geographic information system and an optimization scheduling algorithm, efficient scheduling of engineering vehicles and equipment is realized. Initial scheduling takes the least enough use of equipment as a target, and meets task allocation, time window and equipment plan continuity constraints; when a fault occurs, constructing a global candidate set containing working and idle equipment, and generating a'borrowing-sharing-returning 'cooperative path scheme; and finally, the task is completed through closed-loop execution. The system stability and the resource utilization rate are remarkably improved, the emergency cost is reduced, the system can be seamlessly expanded to a large complex engineering scene, and the system is suitable for equipment scheduling in the fields of engineering construction, road maintenance and the like.

Description

Technical Field

[0001] This invention belongs to the field of intelligent scheduling and industrial automation technology, specifically relating to an intelligent scheduling system and method that combines geographic information system (GIS), equipment model and scheduling algorithm to realize dynamic mutual assistance between vehicles and equipment, applicable to fields such as engineering construction and road maintenance that require multi-equipment collaborative operation. Background Technology

[0002] In fields such as engineering construction and road maintenance, the efficiency of vehicle dispatching systems directly affects overall operating costs and project schedules. Currently existing vehicle dispatching solutions can be mainly categorized as follows: Scheduling system based on human experience These systems primarily provide map display and vehicle location functions, with scheduling decisions entirely reliant on manual intervention. They suffer from fundamental flaws such as high subjectivity, low efficiency, and difficulty in handling complex scenarios involving multiple vehicles, failing to meet the demands of modern large-scale engineering projects for precise and real-time scheduling.

[0003] Automated scheduling system based on static task allocation These systems achieve basic automation, typically using shortest path algorithms to allocate a fixed number of devices to each task. Their core flaw lies in the rigidity of their scheduling model: 1. Lack of fault tolerance and resilience: The system assumes that equipment is always running normally. Once a piece of equipment (such as the equipment at task point A) fails, the system can only call a new piece of equipment from the idle resource pool or garage for a "one-to-one replacement". This mode is slow to respond, has high empty running costs, and does not utilize the potential repair capabilities provided by other online equipment in the task network.

[0004] 2. Static binding of devices and tasks: The system treats devices as independent intelligent agents, lacking a mechanism for dynamic collaboration and mutual support between devices during task execution. Its algorithm model cannot support the advanced working mode of "one device interleaving and sharing multiple tasks".

[0005] Scheduling systems based on traditional algorithms Existing system algorithms (such as traditional genetic algorithms and ant colony algorithms) are based on the "one device, one task" paradigm for their optimization objectives and constraint models. They do not incorporate "dynamic cooperation between devices" as a feasible scheduling state into their solution space. Therefore, even if the scheduler can devise an efficient "borrowing-sharing-returning" collaborative scheme, the underlying system lacks the corresponding algorithmic support to automatically calculate and execute it.

[0006] In summary, existing technologies suffer from the following shortcomings: high emergency response costs and low efficiency in the event of equipment failure; rigid binding of equipment and tasks, resulting in low resource utilization and poor system resilience; and a lack of core scheduling algorithms that support dynamic mutual assistance and collaboration among multiple devices and tasks. Therefore, an intelligent scheduling technology capable of addressing these issues is urgently needed. Summary of the Invention

[0007] This invention provides an intelligent scheduling system and method. When one or more devices fail, other devices in the task network provide dynamic support and cooperation to execute a "borrow-share-return" logical scheduling scheme. In practical applications, this achieves the goal of completing all scheduling tasks with minimal equipment investment, maximum system stability, and minimum emergency costs.

[0008] A dynamic, cooperative, and intelligent dispatching system for engineering vehicles and equipment includes: Initial scheduling module, fault monitoring and collaborative decision-making module, scheme execution and closed-loop module; The initial scheduling module is used to allocate devices from the device set to each task point with the primary optimization objective of having enough devices to use the minimum number of devices, thereby generating an initial scheduling scheme. The optimization objective is achieved through the objective function min∑yv, where v∈V, V is the set of all devices, yv=1 indicates that device v is enabled, and yv=0 indicates that device v is not enabled, and satisfies the following constraints: a. Task allocation constraints: Each task is assigned to only one device that meets the type and capability requirements; b. Time window constraint: The time when the device arrives at the task point is within the task time window, and the task execution time satisfies durj; c. Equipment schedule continuity constraint: The sequence of tasks executed by the equipment is feasible in time and space, that is, the end time of task j + the travel time to task k ≤ the start time of task k; The fault monitoring and collaborative decision-making module is used to monitor the operating status of the equipment. When a fault is detected in equipment v_f at task j_f, i.e. state_v=FAULT, a global dynamic candidate device set C is constructed, and a "borrowing-sharing-return" collaborative path scheme is generated based on the borrowing rules. The candidate device set C includes all devices with a status of WORKING or IDLE. The scheme execution and closed-loop module is used to output scheduling instructions to drive the equipment to execute the collaborative scheme until all tasks are completed.

[0009] A dynamic, cooperative, and intelligent vehicle scheduling method includes the following steps: S1: Initial scheduling, with the goal of having the minimum number of usable devices, is based on the objective function min∑yv (v∈V) and constraints on task allocation, time window, and device plan continuity. Devices are allocated to each task point to generate an initial scheduling scheme. S2: Fault monitoring and collaborative decision-making. When a device fault is detected, a global dynamic candidate device set C is constructed, and a collaborative path scheme of "borrowing-sharing-returning" is generated based on the borrowing rules. S3: Scheme execution and closed loop, outputting scheduling instructions to drive the equipment to execute the scheme until all tasks are completed.

[0010] This invention has the following significant beneficial effects: 1. Extremely high system stability and efficiency: Through the dynamic mutual assistance mechanism between devices, the system can quickly respond using nearby devices in the task network when a device fails, avoiding the high latency and high cost of dispatching from a remote location, ensuring the continuity of tasks, and significantly improving overall scheduling efficiency.

[0011] 2. High resource utilization: It breaks the rigid model of "one device, one task" and realizes cross-task sharing and flexible allocation of equipment resources, thereby enabling the same workload to be completed with fewer physical devices, directly reducing asset investment and operating costs.

[0012] 3. Advanced scheduling capabilities: This invention provides a novel scheduling paradigm that does not exist in the prior art. Its core algorithm can automatically generate and execute complex collaborative schemes, making it more capable of handling large-scale, dynamically changing complex engineering scenarios.

[0013] 4. Excellent scalability: The collaborative scheduling model of this invention is not limited to a specific number of tasks or devices, and can be seamlessly extended to large-scale scheduling networks, making it highly universal. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the dynamic mutual assistance intelligent scheduling method of the present invention. Detailed Implementation

[0015] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings: like Figure 1 As shown, a dynamic mutual-aid intelligent scheduling method for vehicles includes the following steps: S1: Initial Scheduling. The system prioritizes "minimum sufficient equipment" as its primary optimization objective, allocating equipment from the garage to each task point and generating an initial scheduling plan.

[0016] Algorithm details: min∑yv v∈V Here, V represents the set of all devices, and v is an individual device (i.e., an index variable) in the set V. If yv = 1, it means that device v is enabled (at least one task has been assigned); If yv = 0, it means that device v is not enabled.

[0017] min∑yv v∈V The meaning is: minimize the number of all enabled devices, that is, minimize the total number of enabled devices.

[0018] Constraints: a. Task assignment constraints: Each task must be assigned to one and only one device that meets its type and capability requirements.

[0019] b. Time window constraint: The time when the device arrives at each task point must be within the task time window, and the task execution time must meet the time constraint. j .

[0020] c. Equipment schedule continuity constraint: The sequence of tasks executed by the equipment is feasible in time and space, that is: the end time of task j + the travel time to task k <= the start time of task k.

[0021] Solution method: For small to medium-sized problems, mixed integer programming (MIP) solvers (such as CPLEX and Gurobi) can be used to find the optimal solution.

[0022] For large-scale problems, improve heuristic algorithms such as Clarke-Wright Savings or Adaptive Large-Scale Neighborhood Search (ALNS) are used, and the algorithm design prioritizes merging tasks into paths of existing devices to reduce the need to enable new devices.

[0023] When the number of devices is limited and the task requires relatively ample time, completing the task with the fewest devices within the specified time becomes the primary objective of device deployment decisions. Compared to cluster optimization configuration aimed at minimizing time, the cluster optimization configuration model aimed at minimizing devices only changes the objective function and time constraints.

[0024] "In this system, 'small to medium scale problems' refer to situations where the number of tasks does not exceed 50 and the number of devices does not exceed 20; 'large scale problems' refer to situations where the number of tasks or devices exceeds the above range, or where the problem structure is too complex for the precise algorithm to solve efficiently." S2: Fault Monitoring and Collaborative Decision Making. When the system detects a fault in any equipment at any task point (e.g., task point A), the engine starts: This step is triggered when the system detects that device v_f has failed at task (j_f point state_v = FAULT).

[0025] S2.1 Construct a global dynamic candidate device set: Unlike traditional methods that select only from the idle device library, this invention includes all devices in the WORKING or IDLE state into the candidate set C. For WORKING devices, the system treats their current plan sch_v as an "existing commitment" that must be followed.

[0026] S2.2 Generate a collaborative path scheme of "borrowing-sharing-returning": The core is to find one or more "support device chains" for the faulty task j_f. A support action is defined as a = (v_s, j_f, t_{borrow}, t_{return}).

[0027] Secondment rules: a. Parameter meaning: v_s: Candidate supported devices; type_v: Device type; type_jf: The type of device required for the faulty task; cap_v: Equipment capacity (such as load capacity, power, etc.); res_req_jf: The amount of resources required by the faulty task; dur_{j_f}: Duration of the faulty task execution; j_c: The current task of the candidate device; j_n: The next scheduled task for the candidate device.

[0028] b. Rules 1. Candidate selection: Device v_s must satisfy: type_v matches type_jf, and cap_v>= res_req_jf.

[0029] 2. Time feasibility: There exists a time gap that allows v_s to proceed from the completion of a task j_c in its current schedule to j_f, execute dur_{j_f}, and then return to the beginning of its next scheduled task j_n.

[0030] The system will calculate whether the following time window is valid: End time (j_c) + travel time (j_c → j_f) + dur_{j_f} + travel time (j_f → j_n) ≤ Start time (j_n) If the requirements are not met, the system assumes that the device is unable to provide support independently and will: a. Try multi-device relay (break the task into multiple sub-tasks); b. Consider task partitioning (some devices perform some tasks); c. Trigger global rescheduling (re-planning the task sequence of multiple devices).

[0031] 3. Shared nature: This rule itself embodies "shared" - device v_s is shared with the faulty task and its original task on the timeline.

[0032] Return rule: The time constraint mentioned above implicitly requires a return to the original plan. The target location for return is the next task point loc_{j_n} in the original plan.

[0033] Construct a global candidate set: Include all available devices in the system (including devices that are executing other tasks B, C, D, etc.) in the candidate scope to break task isolation.

[0034] Generate collaborative paths: Based on the global device state, calculate one or more "borrow-share-return" collaborative paths. This model supports complex collaborative networks, such as scheduling devices B, C, and D to support task point A in an alternating, relay, or joint manner, while ensuring that the supporting devices can return to complete their own tasks.

[0035] Alternating, relay: Rule: Multiple devices take turns performing different stages of the same task.

[0036] For example: Device B executes the first half of task A, and device C executes the second half.

[0037] Judgment criteria: The available time windows of each device are consecutive, and the transfer time is acceptable.

[0038] Path algorithm details: Separate the independent activities in the entire project and arrange them in a logical chronological order; Draw a network diagram for this project, using directional arrows to mark the relationships between the predecessor and successor activities of each node; The time points (latest start time, earliest finish time, and latest finish time) of each activity are calculated using forward and backward methods, and the time difference of each activity is also calculated. Identify the activities whose earliest start time and latest start time have zero time difference, and then arrange them in chronological order to form the critical path of the project.

[0039] Solution evaluation: Evaluate the generated collaborative solutions and select the optimal solution.

[0040] Solution generation algorithm: 1. Single device support: Traverse the candidate set C, check all possible planning gaps for each device v_s, and find an insertion point that can satisfy the above borrowing rules.

[0041] 2. Multi-device relay support: If a single device cannot fully cover `dur_{j_f}`, consider splitting `j_f` into multiple subtasks. The system searches for two or more devices so that their available time windows can be consecutively covered [t_{j_f}^{start}, t_{j_f}^{end}], and the transition time between devices is acceptable. This involves task splitting optimization.

[0042] Task partitioning optimization mathematical model Task j_f is split into m subtasks, each subtask j_f^k having an execution time of dur_k, and: Σ_{k=1}^{m} dur_k = dur_{j_f} Each subtask is assigned to a device `v_s^k`, whose available time window is `[start_k, end_k]`, and must satisfy the following: start_k + dur_k + travel_time(v_s^k → v_s^{k+1}) ≤ start_{k+1} and all subtask time windows cover the full task time: start_1 = t_{j_f}^{start}, end_m = t_{j_f}^{end} Optimization objective: Minimize total transfer time or maximize time window utilization.

[0043] 3. Global Rescheduling: As an alternative strategy, when plug-in support is not feasible, the engine can take the current moment as the starting point and perform local rescheduling on all devices that were originally scheduled to perform subsequent tasks involving the faulty device, replanning the path and task allocation, and forming a brand-new collaborative solution that includes mutual assistance.

[0044] Decision algorithm details: Based on experience, decisions are made to dispatch equipment to perform tasks.

[0045] Optimization of swarm configuration with the goal of minimizing time: Particle swarm encoding and solving the model using Matlab programming. Set the population size and number of iterations, and solve for the optimal result in the shortest time.

[0046] Particle encoding method: Each particle represents a scheduling scheme, encoded as a device-task sequence matrix, for example: Particle = [Device 1: Task A → Task C → Task B; Device 2: Task D → Task E; …] Objective function: Minimize the completion time of all tasks: min max_{j ∈ J} (completion_time(j)) Iterative process: Initialization: Randomly generate `N` particles (scheduling scheme) and initialize the velocity.

[0047] Evaluation: Calculate the fitness of each particle.

[0048] Update the individual optimal and global optimal.

[0049] Update particle velocity and position: v_{id} = w·v_{id} + c1·rand()·(pbest_{id} - x_{id}) + c2·rand()·(gbest_d - x_{id}) x_{id} = x_{id} + v_{id} Iterate until the maximum number of iterations is reached or convergence occurs.

[0050] Optimization of the configuration of a fleet of machinery with the goal of minimizing the amount of construction equipment.

[0051] The decision-making option is determined after comparing the three options.

[0052] S3: Solution Execution and Closed Loop. The engine outputs the final instruction, driving the equipment to execute the above dynamic cooperative scheduling until all tasks are completed.

[0053] Overcoming the limitation of existing systems that can only resort to costly "equipment replacement" when equipment fails, this enables rapid and low-cost emergency response; Break the rigid binding between devices and tasks, and design a mechanism that supports the dynamic allocation of devices as "mobile shared resources" among different tasks to achieve the optimization goal of "minimum sufficient devices to use"; It provides specific and computable collaborative scheduling methods to achieve "mutual assistance in repair" among multiple devices and tasks, ensuring that all tasks can still be completed efficiently when some devices in the system fail.

[0054] This invention provides a dynamic mutual assistance intelligent scheduling system and method for engineering vehicles and equipment. Through three core links—initial scheduling, fault monitoring and collaborative decision-making, and closed-loop execution of the plan—dynamic mutual assistance scheduling of equipment is achieved.

[0055] Those skilled in the art can adjust the solution method and algorithm parameters according to the number of tasks, equipment parameters, time requirements, etc. in the actual engineering scenario, and all of these adjustments fall within the protection scope of this invention.

Claims

1. A dynamic, cooperative, intelligent dispatching system for vehicles, characterized in that, include: Initial scheduling module, fault monitoring and collaborative decision-making module, scheme execution and closed-loop module; The initial scheduling module is used to allocate devices from the device set to each task point with the primary optimization objective of having enough devices to use, thereby generating an initial scheduling scheme. The optimization objective is expressed by the objective function min∑yv, where v∈V, V is the set of all devices, yv=1 indicates that device v is enabled, and yv=0 indicates that device v is not enabled; and it satisfies the following constraints: a. Task allocation constraints: Each task is assigned to only one device that meets the type and capability requirements; b. Time window constraint: The time when the device arrives at the task point is within the task time window, and the task execution time satisfies durj; c. Equipment schedule continuity constraint: The sequence of tasks executed by the equipment is feasible in time and space, that is, the end time of task j + the travel time to task k ≤ the start time of task k; The fault monitoring and collaborative decision-making module is used to monitor the operating status of the equipment. When a fault is detected in equipment v_f at task j_f, i.e. state_v=FAULT, a global dynamic candidate device set C is constructed, and a "borrowing-sharing-return" collaborative path scheme is generated based on the borrowing rules. The candidate device set C includes all devices with a status of WORKING or IDLE. The scheme execution and closed-loop module is used to output scheduling instructions to drive the equipment to execute the collaborative scheme until all tasks are completed.

2. The intelligent scheduling system according to claim 1, characterized in that, The solution method for the initial scheduling module is as follows: For small- to medium-sized problems with ≤50 tasks and ≤20 devices, a mixed-integer programming solver is used to find the optimal solution. For large-scale problems with more than the above-mentioned number of tasks or devices, or complex problem structures, improved heuristic algorithms are adopted, including saving algorithms or adaptive large-scale neighborhood search algorithms. In addition, the algorithm design prioritizes merging the paths of tasks to existing devices to reduce the activation of new devices.

3. The intelligent scheduling system according to claim 1, characterized in that, The secondment rules include: a. Candidate selection rules: The type_v of the supporting device v_s matches the type_jf of the faulty task j_f, and the device capability cap_v ≥ the amount of resources required by the faulty task res_req_jf; b. Time feasibility rule: The time for support device v_s to go to j_f to execute dur_{j_f} after the current task j_c ends, and then return to the originally planned next task j_n, satisfies: End time (j_c) + travel time (j_c→j_f) + dur_{j_f} + travel time (j_f→j_n) ≤ Start time (j_n).

4. The intelligent scheduling system according to claim 1, characterized in that, The generation method of the "borrowing-sharing-return" collaborative path scheme includes: a. Single device support: Traverse the candidate set C, find the plan gap that satisfies the borrowing rules for each device v_s, and determine the insertion point; b. Multi-device relay support: When a single device cannot fully cover dur_{j_f}, j_f is split into m subtasks. The execution time of each subtask satisfies Σ_{k=1}^{m}dur_k=dur_{j_f}. Each subtask is assigned to device v_s^k, and the time window of each subtask satisfies start_k+dur_k+travel_time(v_s^k→v_s^{k+1})≤start_{k+1}, and start_1=t_{j_f}^{start} and end_m=t_{j_f}^{end}. c. Global rescheduling: When plug-in support is not feasible, starting from the current moment, all devices involved in the original planned subsequent tasks of the faulty device are locally rescheduled, and the paths and task allocations are replanned.

5. The intelligent scheduling system according to claim 4, characterized in that, The optimization objective of the multi-device relay support is to minimize the total transfer time or maximize the utilization of the time window.

6. The intelligent scheduling system according to claim 1, characterized in that, The global dynamic candidate device set C in the fault monitoring and collaborative decision-making module regards the current plan sch_v of the WORKING state device as an existing commitment that must be followed.

7. The intelligent scheduling system according to claim 1, characterized in that, The global rescheduling is solved using the particle swarm optimization algorithm. The particles are encoded as a device-task sequence matrix, and the objective function is min max_{j∈J}(completion_time(j)). The iterative process includes initialization, fitness evaluation, updating individual optimal and global optimal, updating particle velocity and position, until the maximum number of iterations is reached or convergence is achieved.

8. A dynamic, cooperative, and intelligent scheduling method for engineering vehicles and equipment, characterized in that, Includes the following steps: S1: Initial scheduling, with the goal of having the minimum number of usable devices, is based on the objective function min∑yv (v∈V) and constraints on task allocation, time window, and device plan continuity. Devices are allocated to each task point to generate an initial scheduling scheme. S2: Fault monitoring and collaborative decision-making. When a device fault is detected, a global dynamic candidate device set C is constructed, and a collaborative path scheme of "borrowing-sharing-returning" is generated based on the borrowing rules. S3: Scheme execution and closed loop, outputting scheduling instructions to drive the equipment to execute the scheme until all tasks are completed.

9. The intelligent scheduling method according to claim 8, characterized in that, In step S2, if the time feasibility rule is not met, first try multi-device relay support; if it is still not feasible, then trigger global rescheduling.

10. The intelligent scheduling method according to claim 8, characterized in that, The method is applicable to multi-device, multi-task scheduling scenarios in engineering construction and road maintenance.

Images