Power Quality Disturbance Location Method based on Cross-Feedback - Recursive Least Squares

Due to the randomness and complexity of power quality disturbances, there is lack of mature and reliable de- tection and analysing methods on power quality disturbance, especially in the construction site with the changeable opera- tion condition .In order to deal with the problems of non-stationary power quality signals and spectrum leakage, a new CF Recursive Least Squares (CF-RLS) based on blind sources separation method is proposed in this paper. Furthermore the way of converging on the proposed method is based on the maximum negative entropy gradient value. In this way, the verges can be detected and the CF-RLS method can meet the requirement of signal reconstruction condition. With the help of Matlab 7.0, the simulation cases with the power system harmonics with single disturbance and mixed disturbance are discussed. Moreover the simulation results show that the harmonics parameters, including amplitudes, phase angles and disturbance time, can be detected precisely. At last, the proposed method can completely meet the requirements of the power quality disturbance location.