摘要
应用常规递归最小二乘法(RLS)同时估计2个或2个以上变量时容易产生较大误差,改进RLS根据情况选择遗忘因子的个数:利用前一次估计的结果,把误差分散到不同参数,然后各自选用合适的遗忘因子,使各参数以不同的比率变化,优化追踪结果。通过Matlab仿真,分别用常规RLS和改进RLS追踪电压闪变的包络线和初相角,用改进RLS追踪单频、多频闪变及其加噪后的包络线幅值,并对包络线进行了快速傅里叶变换分析。结果证实该方法提高了准确性。
Using the routine RLS(Recursive Least Square) to track two or more parameters may generate serious error. It is improved by selecting the quantity of forgetting factors in light of the situation. Based on the previous estimation,errors are distributed to different parameters and each is given an appropriate forgetting factor to change in different rate for optimizing tracking results. With Matlab simulation,the envelope of voltage flicker and the initial phase angle are tracked by the routine RLS and the improved RLS,and envelope amplitudes of single- and multi-frequency signals with and without noise are tracked by the improved RLS and further analyzed with the fast Fourier transform. Results show that the improved RLS enhances the precision.
出处
《电力自动化设备》
EI
CSCD
北大核心
2006年第6期30-32,共3页
Electric Power Automation Equipment
关键词
电压闪变
递归最小二乘法
遗忘因子
包络线追踪
voltage flicker
recursive least square
fbrgetting factor
envelope tracking