基于差分扰动HBO算法的巡检机器人路径规划

Path Planning of Inspection Robot Based on Differential PerturbationHBO Algorithm

针对巡检机器人需在电站复杂环境下多任务点遍历巡检路径规划问题,提出一种多策略差分扰动机制改进的堆优化(Heap-Based Optimizer,HBO)算法。对不同个体实行差异化的扰动策略,增强新解的多样性。针对巡检任务起终点一致的特性,使用交错同步搜索策略,双种群各执行一半搜索任务,降低了搜索时间,提高了收敛速度。环境建模引入了无限邻域机制,可搜索邻域个数变为连续任意方向,减少了转折点数,使用四阶最小Snap平滑算法平滑整体路径,增大路径平顺度,满足机器人动力学特性。经仿真对比,改进后HBO算法在多任务点遍历巡检任务中性能提升明显,相对于改进前HBO算法和对比的混合粒子群灰狼(Hybrid Gray Wolf Optimizer with Particle Swarm Optimization,HGWOP)算法的步数最低,路径最平滑,最优路径分别缩短了11.1%和20.0%,平均迭代次数分别降低了25.78%和10.53%。

To solve the problem of multi-task point traversal and inspection path planning for inspection robots in the complex environment of power station,a Heap-Based Optimizer(HBO)with improved multi-strategy differential perturbation mechanism is proposed.Firstly,differentiated perturbation strategies are implemented for different individuals to enhance the diversity of new solutions.Secondly,for the characteristics of the inspection tasks with the same starting and ending points,an interleaved synchronous search strategy is used,where the dual populations each perform half of the search tasks to reduce the search time and improve the convergence speed.Finally,the infinite neighborhood mechanism is introduced into the environment modeling,and the number of searchable neighborhoods becomes continuous in any direction,which reduces the number of turning points,and then the fourth-order minimum Snap smoothing algorithm is used to smooth the overall path,increase the smoothness of the path,and satisfy the robot dynamics characteristics.After simulation and comparison,the performance of the improved HBO algorithm in the multi-task point traversal inspection task is significantly improved,and the improved HBO algorithm has the lowest number of steps and the smoothest paths,the optimal paths are shortened by 11.1%and 20.0%,and the average number of iterations is reduced by 25.78%and 10.53%,respectively,compared to the pre-improvement HBO algorithm and the comparative Hybrid Gray Wolf Optimizer with Particle Swarm Optimization(HGWOP).