Study Of Immune Particle Swarm Optimization

The characteristics of the practical engineering problems, such as complexity, constraint, nonlinearity and difficulty of modeling etc, asking for the higher effectiveness of optimization and computation technology, therefore, it is attractive for more and more researchers to find a new type of intelligent optimization method. Swarm intelligence, as a kind of intelligence computation, which exhibits a number of interesting properties such as cooperation, distribution, robustness and rapidness, provides reliable and convenient resolutions to complicated problems under no global information. Particle swarm optimization (PSO), which stems from the simulation of birds flock's looking for food, has been paid attention and researched widely. Relative to genetic algorithm, PSO can be implemented easily because it hasn't crossover and mutation operators and many parameters to be adjusted. Meanwhile, PSO's convergence rate is generally faster than GA. Due to these advantages above mentioned, PSO has been widely applied to the objective functions optimization, dynamic environments optimization, neural network training, and so on.In this paper we review PSO algorithm and its application. PSO algorithm is combined with immune algorithm. Convergence rate and precision are improved by using better diversity in immune system and other immune operators. The disadvantages that PSO algorithm is easy to fall into local best and search ability is weak for multi-peak functions are validly avoided. The main works of this dissertation are as follows:(1)PSO algorithm combines with immune algorithm, and operators that immune memory and clone selection and so on are introduced into PSO to form a mixed algorithm: Immune Particle Swarm Optimization.(2) Diversity criteria is introduced into immune particle swarm optimization. Diversity function is used to monitor population diversity, in order that the particles are updated real time.(3) Using sharing mechanism, fitness values of particles are updated for sharing fitness values. So population diversity is validly ensured. The operators that clone suppression, immune memory and recruiting new number are introduced into PSO algorithm. It not only speeds up search rate but also avoids falling into local best.We test the modified PSO model on the standard test functions, and compare result to standard PSO. The experimental results show that the modified algorithm has better convergence performance for both single-mode function and multi-peak function, and has better convergence rate and precision than standard particle swarm optimization algorithm.