A class of heteroscedastic multiple change-points test method (H-SMUCE) is introduced to test the mean change-point of subway passenger flow data. And the Algorithm steps is proposed. In this paper, since the passenger flow data obeys the normal distribution, this test method has a high accuracy rate. Simulation shows that the accuracy is over 90%. This algorithm for change-points test is applied to the passenger flow of 28 stations in Shanghai Metro Line 1, the positions and times of the change-points are measured, and the law of the change in average passenger flow is analyzed.