基于蒙特卡洛树搜索算法实现轨道交通车辆多功能车辆总线周期调度表优化

耿力1耿强2

Optimization of Multi\|Functional Vehicle Bus Scheduling Table Based on Monte Carlo Tree Search Algorithm

GENG LiGENG Qiang
摘要:
目的:为改善轨道车辆MVB(多功能车辆总线)周期调度表的负载均衡度,提高MVB带宽利用率和车辆的通信系统实时性,需要对MVB周期调度表进行优化。方法:简述了MVB周期信息通信原理,明确了MVB调度表的2个约束条件,建立将MCTS(蒙特卡洛树搜索)算法应用于MVB调度表优化的理论方法,并通过Python软件编程实现该算法。明确了MCTS算法的优化目标,结合MVB调度表的约束条件设计了优化MCTS算法的预剪枝策略。选取了RMS(单调速率调度)算法、MCTS算法和GA(遗传算法)三种算法进行仿真试验,并对各算法的结果进行对比分析。结果及结论:优化后的MCTS算法有效避免了大量无效的搜索,搜索效率非常高。与RMS算法、GA相比,优化后的MCTS算法能在相同的搜索时间内获得更均衡的解。如果车辆通信设备增加,需要生成包含更多变量的调度表时,MCTS算法更能凸显其搜索优势。
Abstracts:
Objective: In order to improve the load balance of the MVB (multi\|functional vehicle bus) periodic scheduling table for rail vehicles, increase the utilization of MVB bandwidth, and achieve the real\|time performance of the vehicle's communication system, it is necessary to optimize the MVB periodic scheduling table. Method: The principle of MVB periodic information communication is briefly introduced, the two constraints of MVB scheduling table are clarified, a theoretical method for applying MCTS (Monte Carlo Tree Search) algorithm to MVB scheduling table optimization is established, and the algorithm through Python software programming is implemented. The optimization objectives of the MCTS algorithm are clarified, and a pruning strategy for optimizing the MCTS algorithm is designed based on the constraints of the MVB scheduling table. Three algorithms, RMS (Monotonic Rate Scheduling) algorithm, MCTS algorithm, and GA (Genetic Algorithm), are selected for simulation experiments, and the results of each algorithm are compared and analyzed. Result &Conclusion: The optimized MCTS algorithm effectively avoids a large number of invalid searches with a very high search efficiency. Compared with RMS algorithm and GA, the optimized MCTS algorithm can obtain more balanced solutions within the same search time. If the number of vehicle communication devices increases and a scheduling table containing more variables needs to be generated, the advantages of the MCTS algorithm in searching can be further highlighted.
论文检索