倒立摆的PID神经元网络的间接控制系统的研究

Research of PID neuron network indirect control system on pendulum

摆系统是一个典型的强耦合、非线性、高阶次的不稳定系统。由于摆系统的数学模型是在忽略了次要因素的基础上得出来的,而实际上是一个非线性的系统,当系统受到外部的干扰时,这些次要因素的影响比较突出。实验采用PID神经元,设计一个神经网络间接自适应控制系统,首先用一个神经网络对摆系统模型进行辨识,辨识完成后,辨识模型的权值与隐层积分元的数值传递给具有同样结构的PID神经元的神经网络控制器,对倒立摆进行自适应控制。最后根据以上算法,采用6.0编写控制程序,实现对平面一级摆系统的实时控制。

Pendulum is typical non-stability system which is strong coupled,non-linear and higher order. Because mathematics model of the pendulum system is obtained by neglecting some secondary factors,that is nonlinear systems in fact,thus when pendulum is bear external disturbance,influence of secondary factors would be serious. PID neuron was adopted to design neural network indirect self-adaptive systems. First let a neural network to carry out model identification for pendulum systems ,when model is identified,weights and number of hidden layer integral neuron are delivered to another network structure of it is same as identification,and carry out self-adaptive control for pendulum Finally the real time control of the plane one rod inverted pendulum is achieved by control program written with VC＋＋6.0 according to control operator.