论文部分内容阅读
首先提出一种基于混沌映射的差分进化算法,通过引入混沌映射的概念,在群体初始化和子代重构两个方面对经典差分进化算法进行改进,提高其寻优精度及稳定性,并通过对几个典型的Benchmark函数进行对比测试,验证该算法的全局收敛能力与稳定性.然后将该改进算法应用于在线轨迹优化,利用其快速寻优、不依赖梯度信息等特点,结合滚动窗口的思想,提出局部极值逃逸方法,实现了轨迹的在线优化.最后在板球系统上通过仿真实验,验证了所提出方法的有效性.
Firstly, a differential evolution algorithm based on chaos mapping is proposed. By introducing the concept of chaos mapping, the classical differential evolution algorithm is improved in both group initialization and descendant reconstruction to improve its accuracy and stability, A typical Benchmark function is used to test and verify the global convergence ability and stability of the algorithm.And then, the improved algorithm is applied to online trajectory optimization.Using its characteristics of rapid optimization and independent of gradient information, combined with the idea of rolling window, The local extremum escape method is proposed to realize the on-line optimization of the trajectory.Finally, the simulation experiment is carried out on the cricket system to verify the effectiveness of the proposed method.