基于BP神经网络的SPWM逆变器控制仿真研究  

Simulation and Research on PI Control Based on BP Neural Network of SPWM Inverter

作　　者：代冀阳[1] 朱丽芳[1] 易宏[1] 

机构地区：[1]南昌航空大学无损检测技术教育部重点实验室,江西南昌330063

出　　处：《计算机仿真》2010年第3期161-165,共5页Computer Simulation

摘　　要：采用控制和实现方法是决定单相SPWM逆变器输出波形质量和动态性能的主要因素。在分析逆变器常规PID控制方法优缺点的基础上,针对带非线性负载和负载跳变的单相SPWM逆变器,输出波形畸变较大,动态性能差和THD值较高的缺点,提出了一种基于BP神经网络自整定PI双闭环控制方案,并用MATLAB软件工具进行了仿真验证。仿真实验结果表明,方法能同时实现逆变器的高精度稳态输出波形、低的总谐波畸变率和快速动态响应性能,适用于感应电源、UPS不间断电源等需要高性能输出电压波形的场合。Generally, the key factor which determines the output waveform quality and dynamic characteristics of single - phase SPWM inverter are what kind of control method is selected and how the control law is implemented. On the basis of analysis of the conventional PID control for the inverter, to deal with the defects of larger output waveform distortion, poor dynamic performance and higher value of THD for single - phase SPWM inverter with non - linear load and load - hopping, a novel control method is put forward, which is based on BP neural network adaptive PID and Double - loop. The simulation on an inverter based on MATLAB were implemented. The result show that the system can achieve higher output voltage waveform, lower total harmonic distortion and fast transient responses. This control scheme was best suited for inductive power supply and UPS applications where high performance voltage waveforms are needed.

关 键 词：逆变器 误差反传神经网络 双闭环 仿真 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TM464[自动化与计算机技术—控制科学与工程]