摘要
为了解决利用粒子群算法对非线性和不确定系统进行PID控制参数整定时存在的种群多样性较低、控制参数在线学习能力差等问题,提出了增强学习的PID控制参数优化快速整定算法;首先,对进化学习算法进行改进;然后利用神经网络进行混合PID控制器设计,利用增强学习算法进行在线反馈学习;最后对每次种群进化后的多样性进行了自适应变异;通过对输入曲线的跟踪对比,验证了文中算法的参数整定效果,同时对种群的多样性进行了跟踪仿真;仿真实验表明,文中的算法具有较强的鲁棒性,算法收敛速度较快,整定效果较高。
In order to solve the nonlinear and uncertain systems using particle swarm algorithm for PID control parameters tuning, which exist the problem of the diversity of the population and the control parameters online learning ability is low, an enhanced learning of the PID controller parameters optimized for fast tuning algorithm was proposed. First, improve the evolutionary learning algorithm; then, neural net- work hybrid PID controller design based on the use of reinforcement learning algorithm for online feedback learning; Finally, after each time the population evolutionary diversity adaptive the mutation. By tracking the input curve compared verify the effect of the proposed algorithm parameter tuning, tracking simulation of the diversity of the population. Simulation results show that this algorithm is robust algorithm, and algorithm converges faster and has the higher the effect of tuning.
出处
《计算机测量与控制》
北大核心
2014年第2期467-470,479,共5页
Computer Measurement &Control
