基于BP神经网络算法的孤岛微网控制策略研究

Research on Control Strategy for Islanded Microgrid Based on BP Neural Network Algorithm

摘要: 微网在孤岛运行时，须依靠自身的调节能力来维持电压和频率的稳定。本文针对微网孤岛运行中采用的U/f控制方法的不足，提出了一种基于BP神经网络算法的电压电流双闭环控制策略，设计了可自适应调整参数的PID控制器。采用具有自学习能力的BP神经网络算法，在线调整U/f控制中电压电流双闭环控制器的参数，减小微网孤岛运行时由分布式电源输出和负荷波动引起的电压及频率波动。仿真研究表明，基于BP神经网络算法的电压电流双闭环PID控制器具有良好的适应性和抗干扰性，有利于孤岛微网的稳定运行。

Abstract: When microgrid operates in islanded mode, it must maintain voltage and frequency by its own regulating ability. To overcome shortcomings of U/f control for microgrid in island operation, an improved voltage and current double-loop controller with a proportional-integral-derivative (PID) structure is proposed in this paper. The PID parameters are adaptively adjusted using BP neural network earning algorithm. The PID control with adaptive adjusted parameters contributes to reducing variations of voltage and frequency in isolation due to the fluctuation of DGs and loads. The simulation demonstrates that the PID controller based on BP neural network with voltage and current double-loop has good adaptability and anti-disturbance, which help to maintain stable operation of microgrid.