改进模糊神经网络控制在伺服系统中的应用

稳定跟踪系统中,被控对象呈现出强耦合、非线性现象多和系统参数多且难以辨识的特点。针对传统的PID难以对其进行精确控制的实际情况,提出了一种新的控制方法。通过在神经网络结构中加入递归层而构造的改进的模糊人工神经网络,可以很好地描述动态系统。采用高斯基函数可以使网络具有很好的泛化能力。仿真试验表明,这种模糊人工神经网络和PID联合起来对一类伺服系统进行多模态控制,可以提高系统的自适应性和鲁棒性,有效提高控制精度,具有很高的应用价值。

Application of a Kind of Improved Fuzzy Neural-network in Servo-systems

A new kind of neural-network is proposed. There are much nonlinear phenomena in the three-axe tracing system. The pure PID can not control the complex system precisely. In order to eliminate the tracing error,this kind of neural-network is presented in the simulation model. The delay layer included in the neural-network and the Gause member function entitle the neural-network to learn from the unknown system better. Compared with traditional control method,the changed neural network is more affective. The simulation result proves that multi-mode control method can deal with nonlinear phenomena in servo systems and has remarkable application value.