摘要
文中在分析RBF神经网络整定PID算法优缺点的基础上,给出了一种采用遗传模拟退火算法来优化网络结构和权值参数的RBF神经网络,将改进的RBF神经网络用于整定PID控制,并给出了相应的仿真测试例子。仿真实验结果表明,与采用梯度法优化网络权值等参数的RBF神经网络相比,给出的优化算法能更好地辨识控制系统,具有通用性好、调节精度高、在抑制超调量能力强等优点。
Based on the analysis of the advantage and disadvantage of RBF neural networks tuning PID algorithm, a RBF neural network with optimized structure and weights by using genetic and simulated annealing algorithm is described. The novel RBF neural network is used in tuning PID control, and the relevant simulated example is given. The simulation results shows that, as compared with the RBF neural network with parameters such as network weights optimized by gradient descent algorithm, the optimized algorithm described in this paper could provide better control system distinction, and has the advantages of excellent general purpose capability, high adjustment precision and strong ability in suppression overshoot.
出处
《通信技术》
2009年第11期219-221,共3页
Communications Technology
