With the economic and social development, many kinds of advanced power electronic equipment and microprocessors have been used in industry. However the voltage sags may shut down the equipment and, as the result, cause enormous economic losses. The voltage sag characteristic value should be detected in time, and the fault branch should be removed from the power system as soon as possible. Therefore, it is of great practical significance to detect the characteristics values of voltage sags in a rapid and accurate manner.

In this study, the detection techniques for voltage sag were used as the research subject. First the characteristics, causes and hazards of voltage sag are introduced. Then, based on the Matlab/Simulink model, the characteristics of the voltage sag caused by three common causes were analyzed and evaluated. The traditional methods were analyzed and evaluated by detecting the starting and ending time of the voltage sag, depth of voltage sag and the phase jump.

The traditional αβ-dq coordinate transformation method which has the following defects: the anti-jamming ability is poor and the phase angle jump interval can not exceed [-π/2,π/2]; in view of this, a method of improving the αβ-dq coordinate transformation was proposed. The improvement method mainly solves the problem through two aspects. Firstly, the anti-jamming ability of the system is improved by placing FIR filter in the front; however this results in the disadvantage of phase delay at the same time, and the phase result is compensated by the linear delay relation of the filter. Secondly, after the transformation of dq coordinates, within the quadrant where the voltages in d axis and q axis,u_d、u_q is located, the detected hopping zone of the phase angle is extended to [-π,π] through combining the trigonometric function. The simulation results show that the amplitude of the voltage sag, the phase hopping, the duration and other characteristic quantities can be detected accurately in