摘要
为有效提取多变量系统中对系统解释性最强的综合变量,克服变量多重相关性在系统建模中的不良影响,提出了基于偏最小二乘回归进行数据建模的一种新方法。介绍了该方法的基本原理和建模基本思想,并以实例分析了该方法对多变量信息的综合与筛选作用。将该方法应用于谐波源的定量分析中,利用测试的谐波电压和电流信号,通过偏最小二乘回归算法求解回归系数,进而可计算系统谐波阻抗与用户的谐波发射水平。仿真示例说明了该方法的有效性和优越性。
A novel modeling method is proposed in the paper based on PLSR（partial least-squares regression） in order to extract the comprehensive variables which can explain the system performance most effectively in multivariable system and overcome the system-modeling difficulty caused by multiple correlation of the variables. The fundamental principle of the method as well as its modeling technique is introduced in the paper,and its functions as synthesizing and filtering the multivariable information are revealed by an example. The proposed method is meant to be applied in the quantitative analysis of harmonics sources. After obtaining the measured harmonic sgnals of voltage and current, the regression coefficients are solved by PLSR. Furthermore, the system harmonic impedance and the harmonic emission level of the customers are calculated. The validity and feasibility of the proposed method are proved by the simulation results.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2007年第6期57-61,71,共6页
Proceedings of the CSU-EPSA
关键词
偏最小二乘回归
多重相关性
信息综合与筛选
回归系数
谐波发射水平
partial least-squares regression
multiple-correlation characteristic
information synthesizing andfiltering
regression coefficient
harmonic emission level
