DC／DC变换单神经元自适应PID控制器仿真研究

为解决传统PI控制对具有时变、非线性特性的DC/DC变换器动态控制性能不佳的问题,将人工神经网络与PID控制理论相结合,为DC/DC变换器设计了一种单神经元自适应PID控制器。该控制器算法简单,通过对加权系数的在线调整来实现自适应、自学习功能,从而满足DC/DC变换器的时变及非线性特性。以BUCK型变换器为例,建立了DC/DC变换器智能控制系统的仿真模型,在不同负载及参数变化的情况下与常规PI控制效果进行对比分析,结果表明,单神经元自适应PID控制器对DC/DC变换器具有很好的控制效果和鲁棒性。

Simulation Study of Single Neuron Adaptive PID Controller for DC/DC Convertor

To improve the dynamic performance of time-varying nonlinear DC/DC convertor,a single neuron adaptive PID controller is designed by means of the neural network theory combined with the PID control theory.The designed algorithm is easy.The self-adaptive and self-learning functions are realized by adjusting the weights online.Taking BUCK DC/DC convertor as an example,the simulation model with single neuron adaptive PID controller is built.The results show that the designed controller of DC/DC convertor has more excellent control performance and robustness than PI controller when load and parameters are different.