Adaptive Particle Swarm Optimization Algorithm And The Application Research

The problem of parameters optimization is involved in many scientific, engineering and economic fields. Numerous optimization algorithms, such as Genetic Algorithm (GA), Immune Algorithm (IA) and Particle Swarm Optimization (PSO), have been proposed to solve this problem. As a relatively new evolutionary computation, PSO algorithm applies the information communication within particles to update itself, instead of any genetic operator such as crossover and mutation. All the particles are able to adjust their velocity and remember the best position in the past. This optimization algorithm has been applied to many engineering problem successfully.Base on the study of PSO basic principal, a new adaptive PSO (APSO) with catastrophe operator and linear increment is presented. APSO solve the local best solution during the optimization better than standard PSO. By applying APSO to FIR digital filter design, multi-criterion satisfactory optimization and PID controller design, the simulations have approved the validity and superiority of APSO.The main contributions of this thesis are listed as following:1. Synthesis the history and the current research progress of Computational Intelligence, PSO, multi-criterion satisfactory optimization and robust PID controller.2. Summarize the principles of PSO and frequency sampling filter (FIR). A new method of designing FIR based on PSO is presented. The efficiency and feasibility of the proposed method are proved by the simulation results of low-pass FIR.3. Focus on the lack of theory in setting parameters and the disadvantage of getting into local best solution easily, based on analyzing particle trajectory and convergence, Adaptive Particle Swarm Optimization (APSO) is proposed, with an increasing inertia weight and the mechanism for selecting the superior and eliminating the inferior. It is clear that APSO outperforms the standard PSO, both in convergence speed and searching ability.4. Introducing satisfactory optimization function and traditional multi-criterion optimization method, a multi-criterion satisfactory optimization strategy (MAPSO) is proposed based on APSO, substituting "satisfactory solution" for "optimal solution". The satisfactory strategy can give attention to the incompatible criterions under the condition of valid solution or minor valid solution.5. This thesis analyzes performance criteria of Linear Quadratic Regulator (LQR) and minmax fundamental. APSO is used to PID parameter tuning due to its global search ability, and simulation examples show that the PID controller designed by this new method has good tradeoff between disturbance attenuation and robustness.