基于改进神经网络算法控制开关电源的研究  

Research on controlling switching power supply based on improved neural network algorithm

作　　者：张淼 陈颀[1] 谢佳明 ZHANG Miao;CHEN Qi;XIE Jia-ming(Information Engineering and Automation,Kunming University of Science and Technology,Kunming,650500,Yun-nan,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650500

出　　处：《光电子．激光》2021年第4期349-355,共7页Journal of Optoelectronics·Laser

基　　金：云南省教育厅科学研究基金项目(2019Y0035)(超高频低延时功率管关键技术研究)资助项目。

摘　　要：针对数字开关电源的控制策略问题,提出一种改进共轭梯度算法的BP神经网络PID控制系统。以BUCK变换器为研究对象,在BP神经网络PID算法的基础上,通过改进共轭梯度算法优化控制系统的调节时间和恢复速度,以提高数字开关电源系统的控制性能和输出性能。基于MATLAB软件完成系统建模进行仿真研究。结果表明:改进后的控制系统比改进前的调节时间提高0.03 s,超调量降低10%,恢复时间缩短0.03 s。从理论上验证了改进共轭梯度算法的BP神经网络PID控制系统在响应时间上更短,超调量更低,抗干扰能力更强。Aiming at the control strategy of digital switching power supply,a BP neural network PID control system with improved conjugate gradient algorithm is proposed.Taking BUCK converter as the research object,on the basis of BP neural network,the regulation time and recovery speed of the control system are optimized by improving conjugate gradient algorithm and combined with PID algorithm,in order to achieve the optimization of the control performance and output performance of the digital switching power supply system.At the same time,based on the MATLAB platform to complete the system modeling,The BP neural network PID control system and the BP neural network PID control system with improved conjugate gradient algorithm were simulated by MATLAB software.The results show that the adjustment time of the improved control system is 0.03 s higher,the overshoot is 10%lower,The rise time is 0.015 s lower than before improvement,and the recovery time is 0.03 s shorter than that before the improvement.It is verified theoretically that the BP neural network PID control system with improved conjugate gradient algorithm has shorter response time,lower overshot and stronger anti-interference ability than BP neural network control system.

关 键 词：BUCK开关电源 BP神经网络控制 共轭梯度 PID控制 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]