摘要
针对传统系统辨识存在的缺点,提出了基于预报误差法的神经网络辨识方法,将神经网络的预报误差法应用于系统辨识中,通过调节神经网络连接权值可使网络输出逼近系统输出。神经网络作为实际系统的辨识模型,可以用于在线控制。仿真实例表明其收敛速度快于BP算法。
This paper presents neural network identification method based on predict error aimed at the disadvantage of traditional system identification. Predict error method of neural network is applied to system identification. Neural network output can approach to the system output by adjusting the weight of the neural network. Neural network can he used to control the plant online as identification model of the plant. Simulation example shows its convergence speed is faster than BP algorithm.
出处
《现代电子技术》
2006年第23期90-91,共2页
Modern Electronics Technique
基金
河北师范大学校立基金资助项目(L2005Q06)
关键词
神经网络
系统辨识
误差预报法
遗忘因子
neural network
system identification
error predict
forget factor
