a b s t r a c tA recursive generalized least squares algorithm and a filtering based least squares algorithm are developed for input nonlinear dynamical adjustment models with memoryless nonlinear blocks followed by linear dynamical blocks. The basic idea is to use the filtering technique and to replace the unknown terms in the information vectors with their estimates. The simulation results show the performance of the proposed algorithms.

Harmonic voltage of the concern bus in the power grid is the combinations of voltage contributions of each harmonic source. When the main harmonic sources are known to the system, how to distinguish the responsibility of these harmonic sources has important significance. This paper firstly established the multiple linear regression model and defined the harmonic responsibility. When the harmonic current is linearly related, using the least squares regression will get the wrong solution. Therefore, this paper proposed a method called improved least squares. Simulation results show that no matter the coefficient matrix is sick or not, this method can get a better result than the least squares method.

No abstract