摘要
为了克服单相有源电力滤波器的非线性特性,在基于数据的基础上,由极限学习机建立有源滤波器非线性系统的内模和逆模,实现基于极限学习机的内模控制.根据内模控制系统稳态误差,评估内模控制系统的性能,并将极限学习机的内模控制系统与神经网络、核岭回归、支持向量机等方法进行比较分析.试验仿真表明,基于极限学习机的内模控制系统具有系统稳态误差小、鲁棒性强等特点.
To overcome the nonlinear characteristics of single-phase active power filter（APF）for better dynamic performance,a new control strategy for internal model control is proposed by setting up APF internal model and inverse model respectively on data by extreme learning machine（ELM）.Moreover,the relative stable error is presented to evaluate the system performance and the features of the ELM internal model control system is compared with neural network,kernel ridge regress and vector machine support.The experimental results show that the ELM internal model control system has small stable error and strong robustness.
出处
《厦门理工学院学报》
2012年第4期38-41,46,共5页
Journal of Xiamen University of Technology
基金
厦门市科技计划项目(3502Z20111008
3502Z20123043)
厦门理工学院高层次人才引进项目(YKJ07008R)
关键词
有源电力滤波器
单相
内模控制系统
非线性控制
极限学习机
稳态误差
active power filter（APF）
single phase
internal model control
nonlinear control
extreme learning machine（ELM）
stable error
