摘要
提出基于改进自适应粒子群算法(Improved Self-adaptation Particle Swarm Optimiza-tion,PSO)的T-S模糊模型辨识方法。首先,利用核函数的模糊聚类算法划分数据空间,尽可能少地提取模糊规则,并消除孤立点、噪声点数据等的不利影响;其次,基于ISPSO算法进行参数辨识,将待辨识的参数划分为若干粒子,自适应更新飞行速度,动态修改惯性权因子,惯性权因子呈非线性动态变化,不仅可以克服PSO算法陷入局部最优的早熟,失去多样性,而且可以提高粒子在全局最优位置绕行时的稳定性。提出的方法使得T-S模型辨识达到较高的辨识精度。仿真实例和比较分析证明了该算法的有效性。
An approach of improved adaptive particle swarm potimizaton is proposed for the identification of T-S fuzzy model. A fuzzy clustering based on kernel function is adopted to partition the data space and extract the fuzzy rules. The adverse effect of isolated point and noise point can be removed. The ISPSO proposed algonthm is applied to optimize the parameters. The parameters grouped as several particles are employed to search the space during the iteration. In each iteration, the velocity of particle is updated adaptively. The improved adaptive technique is utilized to modify the weight dynamically. The dynamical weight can overcome the premature and avoid falling into local optimum losting diversity. The stability of particles flying around the global optimum is improved. The identification accuracy is also improved. The simulation and comparative analysis show that the proposed method achieves very good results.
出处
《控制工程》
CSCD
北大核心
2011年第6期952-955,共4页
Control Engineering of China
基金
上海理工大学光电学院教师创新基金资助(GDCX-T-101)