摘要
针对疲劳试验机控制系统,设计了基于BP神经网络和PID的并行控制器。该控制器充分利用了经典PID控制算法简单的特点,又利用了神经网络良好的自适应能力,首先通过PID控制为神经网络的在线学习提供训练样本,然后神经网络逐渐学习被控对象的动态逆模型并取代PID控制器起主导作用。该方法降低了PID参数的调整难度,同时对控制对象的刚度变化表现出良好的鲁棒性,并通过仿真证明了所设计系统的有效性。
A collateral controller based on BP neural network (NN) and PID was designed for fatigue tester system. The controller exerts the simpleness of PID algorithm and the adaptability of NN. At first, PID controller offers training samples for online studying of NN, then NN studies the contrary model of controlled object step by step and makes the main function through replacing PID. This method reduces the adjusting difficulty of PID parameters. At the same time, it has good robustness for the control of the object's rigidity changes. Simulations prove this method's correctness.
出处
《机床与液压》
北大核心
2005年第8期187-189,共3页
Machine Tool & Hydraulics
关键词
神经网络
并行控制
BP
Neural network
Collateral control
BP