Some Improvements And Applications Of Particle Swarm Optimization

In 1980s,Swarm Intelligence Algorithms,a new technology of evolutionary computation, has become the focus of attention to more and more pursuer. The conception of swarm Intelligence originates from observation and investigation into behavior of gregarious colony such as bee, ant and wide goose. Usually, we call the intelligence computation or optimization method that simulates swarm intelligence behavior of gregarious colony as Swarm Intelligence Algorithms. Particle Swarm Optimization, a new Swarm Intelligence Algorithms, originates from the investigation into behavior of bird swarm prey. It is an optimization technology based on iteration as Genetic Algorithm. System is initialized by a group of random solution and search optimization value by iteration. Recently, Particle Swarm Optimization is applied into function optimization, Neural Networks, data mining, Fuzzy Control System and other application field.The paper begins from the three models of Particle Swarm Optimization and do many improvements. We apply these improvements into Function Optimization, Constrained Optimization, Integer Programming and Interactive Programming. The detailed jobs as follows:(1) The paper put forward an individual best position weight Particle Swarm Optimization based on shrinking gene from the angle of information exchange manner. The new algorithm make particle utilize more available information of other particle. Moreover, it balances the contradiction between efficiency and precision of algorithm search by weight of individual best value and change action mode of particle.(2) The paper put forward a Particle Swarm Optimization and anamorphosis with step-accelerating mutation operator which avoid for particle swarm to plunge into local optimization for multiple hump function, at the same time, we give the detailed analysis of mutation occasion and mutation probability.(3) The paper put forward a hybrid Particle Swarm Optimization which keeps particle swarm acting in feasible region for Constrained Optimization and three methods which construct initial particle swarm.(4)The paper put forward a Particle Swarm Optimization and improving algorithm which keeps particle swarm acting in feasible...