基于分类学习粒子群优化算法的液压矫直机控制

发布时间：2018-03-03 04:34

  本文选题：粒子群优化　切入点：分类学习　出处：《机械工程学报》2017年18期 　论文类型：期刊论文

【摘要】：在处理工程控制及设计中含有多参数,多约束的单目标优化问题时,为了获得更好的优化解,提出一种分类学习的粒子群优化算法。它根据每个粒子的函数适应值,将群体分为优势群体、中层群体和劣势群体三类,分别采取不同的学习方法和学习方向。优势群体继续保持自身的学习速度和学习方向;中层群体采取互相学习的策略;劣势群体采取加强向优势群体学习的策略。其优势在于不受函数连续、可导形式的制约。数值试验结果表明,相比于近年提出的一些改进粒子群算法,这种算法在处理含有单峰,多峰,离散,动态问题的函数时,具有良好的收敛性能。结合工程实例,在处理压力容器结构设计以及液压矫直机PID控制的参数优化问题时,此算法能够获得使系统性能更佳的参数组合。
[Abstract]:In order to obtain a better optimal solution, a particle swarm optimization algorithm is proposed for engineering control and design, which is based on the functional fitness of each particle, in order to obtain a better solution to the single-objective optimization problem with multiple parameters and constraints. The group is divided into three groups: the superior group, the middle group and the inferior group, which adopt different learning methods and learning directions, which continue to maintain their own learning speed and direction, and adopt the strategy of mutual learning. The disadvantaged group adopts the strategy of strengthening learning from superior group. The advantage of this strategy is that it is not restricted by continuous function and differentiable form. The numerical results show that compared with some improved particle swarm optimization algorithms proposed in recent years, This algorithm has good convergence performance when it deals with functions with single peak, multi-peak, discrete and dynamic problems. Combined with engineering examples, it deals with the optimization problems of pressure vessel structure design and PID control of hydraulic straightener. This algorithm can achieve better performance of the system parameters combination.
【作者单位】： 东北大学机械工程与自动化学院;
【分类号】：TH137.5;TP18
，

本文编号：1559582