Abstract: This paper analyzes the data of Hebei digital seismographic stations Chengde, Laiyuan, Luanxian, Jingtanggang and Qianxi before the Tangshan M5.1 earthquake in 2020. The data are processed by wavelet transform, power spectrum analysis, fast Fourier transform and so on to study the abnormal variation characteristics of the microtremor before earthquake. The results show that: ① Before the Tangshan earthquake, there are low-frequency microtremor precursory anomalies at 5 stations, and the anomalies of low-frequency at 4 stations are consistent. ② The amplitudes of the four stations increased before the earthquake and decreased gradually after the earthquake. ③ The PSD varieties of the four stations before earthquake are consistent. They all increased before the earthquake and gradually decreased after the earthquake, and the trend is the same as the amplitude curve.