Voltage Sag Detection Based On Strong Tracking Kalman Filtering

In recent years, there has been a major change in the structure of the load and electrical characteristics of the grid with a large of distributed generation and new types of power load connected into the grid, and the problems of power quality are more serious. Among them, voltage sag is one of the issues which concerned most because of its huge extent of harm and high frequency. It is of great significance to detect the characteristic quantity of voltage sag accurately and effectively, which is the premise of dynamic compensation for sagged voltage.Firstly, this thesis introduces the basic conception of voltage sag and the current research situation of voltage sag at home and abroad. Then, the basic theory of strong Kalman filtering is introduced and we use it to detect the sagged voltage. The main research results are as follows:(1) This thesis proposes an idea that applying strong tracking extended Kalman filtering to the detection of voltage sag, aiming at the disadvantage that extended Kalman filtering cannot track the state of the system accurately in the case of non-determinacy of system model and sudden change of system state. Further optimizations are made on the basis of the normal detection method:the voltage signal model with harmonic components is established considering the inevitable harmonics in the power system; the multiple fading factors are introduced, so different filtering channels are faded with different fading factors. The results of simulation show that the method of voltage sag detection based on optimized strong tracking extended Kalman filtering can detect the sagged voltage under the environment of harmonics, and it also enhances the precision of detection.(2) An idea that applying strong tracking unscented Kalman filtering to the detection of voltage sag is proposed. Strong tracking unscented Kalman filtering, which combines the advantages of strong tracking filtering and unscented Kalman filtering, is applied to the detection of voltage sag. The detection method is optimized in order to further improve the estimation accuracy and reduce the complexity of the detection method. The mode of action of fading factor is improved. In addition, unscented transform is used only once per cycle of state estimation. Experimental data suggest that the method based on optimized strong tracking unscented Kalman filtering advances the performance of accuracy and robustness.