DC/DC变换器的神经网络鲁棒控制

Robust Control for DC/DC Converters Using Neural Network

针对有非最小相位特性的二阶DC/DC(直流/直流)变换器平均值模型的特性,以Buck-Boost变换器为典型例子,提出了一个非线性反馈做内环控制器,控制其电感电流;用RBF(径向基函数)神经网络作为自适应机构,提出了一个神经网络鲁棒控制器作为电压外环控制器,控制其输出电压.证明了系统跟踪误差和神经网络权值的有界性.仿真结果表明,提出的控制器对于系统参数的不确定性具有很强的鲁棒性,并具有很好的动态性能.

As a typical example of second-order DC/DC converters, which has the nonminimum phase property, the model of Buck-Boost converter was analyzed. The average model of Buck-Boost converter was transformed into a first order nonlinear uncertain system using a nonlinear feedback inner loop controller. Employing a RBF（Radical Basis Function） neural network compensator, a neural network robust controller has been proposed as the outer loop controller to control the output voltage of Buck-Boost converter. The tracking error and the weights of the neural network have been proved to be bounded. Simulation results have indicated that the controller is robust to the uncertainties of the converter and has high dynamic performances.