| Chaos is a ubiquity of nonlinear phenomena, the behavior is complex and random-like,but it has a delicate inner regularity. Chaos optimization seek optimal points，using thefeatures of egrodicity of chaotic motion. At the same time, as nonlinear, strong constraints,random, large-scale features often exist in the practical system, and often there is a localminimum. The system real issue goes far beyond the scope of traditional optimizationalgorithm. Chaos Optimization (COA) algorithm overcome these shortcomings, with theunique characteristics of chaotic motion, which makes chaotic search to a certain extent itsown "laws", without traversing each state repeatedly, and more likely to jump out fromlocal minimum, compare traditional algorithm.Therefore, according to chaos with all the features, the main research achievement ofthis dissertation is as follows:1. In order to Provide basic theories of hybrid optimal algorithm，the basic content ofoptimal algorithm mare briefly introduced, and then put it into Particle SwarmOptimization and Neural Network training. Through validating the hybrid optimalalgorithm by numerical experiments，the good optimal impact is gained.2. Based on the analysis of path planning principles, the classical artificial potentialfield is introduced, and successfully applied to path planning through simulation.According to the special case in the simulation, Analysis the traditional artificial potentialfield method shortcomings, and propose an improved methods.3. For the shortcomings of artificial potential field method, the chaos optimizationstrategies are introduced into, and a new repulsive force field function is used, to ensurethat the target point is the global minimum, so that the robot can reach the targetsuccessfully. Simulation results show that the method can effectively achieve the robotpath planning, and solve the local minima problem, and cannot found path between closeobstructions， existing in the traditional artificial potential field. |
PDF Full Text Request