基于单神经元PID控制的DC/DC变换器研究

Research of DC/DC Converter Based on Single Neural Network PID

DC/DC变换器是一种强非线性电路。电路的电气参数存在不确定性,负载性质也是多变的。主电路的性能必须满足负载大范围的变化,同时它还具有离散和变结构的特点,所有这些使DC/DC变换器控制器的设计较为复杂。由于传统的控制方法是基于线性系统理论,所以应用于DC/DC变换器中并不能获得理想的动态性能。文中针对DC/DC变换器的特点以及单神经元网络具有出色的知识抽取与学习能力及较强的控制鲁棒性。在现有的控制和建模方法的基础上,将神经网络控制策略引入DC/DC变换器,构造了一种基于单神经元的DC/DC变换器的新型控制方法。

The DC/DC converter is a strongly nonlinear circuit. The electrical parameters are uncertain. Loads characters are also changeful. The main electrical circuit must satisfy load＇s large- scale change. It also has discrete and variable structure characteristics. These characters make DC/DC converter controller design more complex. Due to the tradition control method is based on linear system theory, so apply to DC/DC can not gain the ideal dynamic performance. This paper according to DC/DC converter＇s features and the single neural network has excellent knowledge extraction, learning ability, strong robust control. Based on the existed control strategy and modeling method introduce neural control strategy to DC/DC converters.