Classic Search Strategy-based Particle Swarm Algoritm And Its Application

Optimization technique is a kind of applied technology, based on mathemcatics,which can solve various optimal solution or satisfactory solution in engineering. Anycontrol and decision making problems can be attributed to the optimization problem.Classical optimization methods, include gradient algorithm and Direct Searchalgorithm, are used to solve optimization problem. However, these algorithms arelocal algorithms, cannot solve global optimization problem of non-convex function.For solving globally optimization problem, scholars have proposed a series ofbio-heuristic method. Particle Swarm Optimization algorithm is a global optimizationheuristic evolutionary algorithm for swarm intelligence，which is the simulation ofbirds’ foraging behavior. With the advantages such as its simple concept, it is easy tobe implemented and it has fast convergence rate. Particle Swarm Optimizationalgorithm has been successfully applied in many areas. But, Particle SwarmOptimization algorithm has some inherent defects, such as following: prematureconvergence, easily turning into local optimization and poor search result. So, toimprove its defects is needed.First of all，In the iteration process of the Particle Swarm Optimization algorithm,conjugate gradient method was introduced to enhance development capabilities, andchaos sequence was introduced to enhance explore ability. Futher, based on thenonlinear conjugate gradient algorithm, a chaos-PSO algorithm was proposed. Toverify the performance of the hybrid optimization algorithm, the algorithm wasapplied to search the all local optimization of the multi-modal function.Theexperiments include the reliability of finding all extremums and the accuracy of thefound extremums.The experiments manifest that it not only has powerful globalexplore capability, but also can effectively avoid premature convergence and improvesearch result accuracy.Then, in the particle swarm optimization algorithm, the introduction of patternsearch enhanced its local minute search ability, while the introduction of Cauchymutation improved its diversity as well. Therefore, a Pattern Search-based ParticleSwarm Optimization Algorithm was proposed, and this improved algorithm can beconveniently applied to Parameters Tuning of the ADRC. The two-order ADRC thatoptimized by Pattern search-particle swarm optimization algorithm, are used in time-delay systems. The experiments include the unit step responsive and theperformance of anti-interference and the robustness.The experiments manifest thatoptimized ADRC has strong control ability and produces good disturbances rejectionand robustness.