神经网络内模控制在非线性时滞系统中的应用研究

Applied research of neural network internal model control in nonlinear time-delay system

针对工业中存在的时滞、非线性问题,将内模控制方法和神经控制原理有效结合起来,并利用一种改进径向基神经网络(radial basis function neural networks,RBFNN)中心点学习方法分别对被控对象的模型和控制器进行自适应学习,通过对实验室电加热炉这种典型一阶滞后对象实验,仿真表明,所提出的方法具有快速跟踪输入、无超调等良好特性,并且能在系统受到干扰或对象参数发生变化时,仍然具有良好的自适应性和鲁棒稳定性。

As to the problem of the time-lag and non-linear in industry,combined an internal model control（IMC） method and neural control principle effectively,it recurred to the identification for controlled object model and controller using an improved RBF center-study neural network adaptively.Based on laboratory electric heating which is a typical first-order lagging object,simulation experiments show that the proposed method has good control characteristic on fast-track input and non-superscalar.The system still has good adaptability and robust stability when system interference or an object parameter changing because of circumstances.