摘要
将多项式基函数神经网络作为模型算法控制（ＭＡＣ）中的内部模型，以此逼近被控对象的非线性特性，推导出相应的最优控制策略算法，克服了ＭＡＣ不能用于非线性预测控制的缺陷．该方法在被控对象未知或建模困难的情况下能很好地实现对系统的预测控制，具有很强的鲁棒性和自适应性．
The nonlinear inner model of model algorithmic control adopts the neural network based on polynomial basis functions,and produces optimal control algorithm.It overcomes the disadvantage that MAC doesn't be applied for nonlinear predication control.In the case of controlled object unknown or model idenfication,this method can obtain satisfied results of the system predication,and which is very robust and selfadaptive.
