Time delay estimation algorithms play an important role in the localization techniques for local discharge sources. In this paper, we first briefly analyze the currently existing time delay estimation algorithms such as traditional mutual correlation, generalized mutual correlation and cubic correlation, and then propose an improved generalized cubic correlation time delay estimation algorithm. And by constructing partial discharge ultrasonic signals, the delay estimation performance of GCC-PHAT-β algorithm, quadratic correlation GCC-PHAT-β algorithm, cubic correlation GCC-PHAT-β algorithm and the improved algorithm in this paper are verified in different signal-to-noise ratio environment, the delay estimation errors of the four algorithms are compared. The simulation results show that this improved algorithm in this thesis has better delay estimation ability and better noise immunity at low signal-to-noise ratio.