Improved Particle Swarm Optimization Algorithm And Its Application Research

Particle Swarm Optimization (Jane said PSO), because of its fast, few parameters and easy to implement, since it was proposed became the focus of many scholars attention and research. The principle of PSO algorithm is inspired by simulating a bird flock in search of food and, but the PSO algorithm is a global optimization algorithm, As with other global optimization algorithm, it is easy to fall into the local extreme value point and poor precision, and slow convergence velocity in late evolution algorithm. How to accelerate the convergence speed of the particle swarm algorithm, improve the convergence precision of the algorithm, has been the focus of scholars. At present, accelerate the convergence speed of the algorithm strategies including how to choose the parameter, and by combining other optimization algorithm of particle swarm optimization algorithm to modify. In order to improve the convergence precision and prevent premature algorithm of mature, experts and scholars have conducted a lot of research, including keep the diversity of particle swarm, and can make particles jump out of local optimal point (for example mutation operators of genetic algorithm, etc.). At present some of the improved particle swarm algorithm are mainly fuzzy adaptive algorithm, the hybrid PSO algorithm, the discrete binary algorithm and PSO algorithm and the immune particle swarm optimization algorithm etc.In real life, We deal with are usually the multi-objective problem, design some effective optimization algorithm to deal with the multi-objective problem is an urgent and realistic problem, so the multi-objective optimization problem is a very important research subject.Based on particle swarm optimization algorithm and differential evolution algorithm, I analysised the existing literature, and make full use of particle swarm optimization algorithm parameters less, easy to implement, and the advantage of fast convergence rate, less controlled parameters for differential evolution algorithm, using random, direct and parallel global search, and has the characteristics of easy to understand and implement, Crossover operation variation of the differential evolution algorithm is introduced into the particle swarm algorithm, puts forward an improved adaptive differential particle swarm optimization algorithm (Jane said EPH),In the evolution process removes the influence of variable speed, simplify the algorithm complexity, and the local extremum and inert particles of the population variation and adaptive crossover operation, The individual based on iterative steps adjustment of adaptive crossover probability, and the algorithm is applied to the single objective and multi-objective function optimization.