Tuning of PID controller based on improved particle-swarm-optimization
基于改进粒子群优化算法的PID控制器整定

Because the classical PID parameter settings obtained by Z-N(Ziegler-Nichols) method usually fail to achieve the best control performances, we propose an improved particle swarm optimization(IPSO) algorithm with fitness exponential scale and border buffer wall for tuning the PID parameters. First, by the selection-probability of the fitness exponential scale, we select the underbred particles for random mutations. Secondly, we employ the border buffer wall to block the slopping-over particles, making them to fall in the explored space of optima to enhance the diversity of the particle swarm. Meanwhile, by modifying the number of swarm particles as well as the social and cognitive factors, we improve the efficiency of optimum searching. In the simulation experiments, we apply the IPSO algorithm to the PID parameter tuning for 5 different industrial process models, the corresponding optimal PID parameters are obtained under the criterion of minimal integral-time-weighted-absolute error(ITAE). The effectiveness of the proposed improved particle swarm optimization(IPSO) algorithm is validated.