A piecewise polynomial truncated singular value decomposition (PPTSVD) algorithm for the identification of moving force under noise and various sensor is developed based on the truncated singular value decomposition (TSVD). An Euler beam is introduced to simulate the passing vehicle. The bending moment and acceleration responses are collected from the simplified system under various noise. Traditional time domain method (TDM),TSVD and PPTSVD are used to identify the load history, respectively. The results show that the precision and robust noise immunity of TSVD is better than TDM using singular value decomposition (SVD). PPTSVD is superior to TDM and TSVD with higher identification accuracy and stronger robust noise immunity. Besides, it is less sensitive to the variety of sensor and their ways of combination. These advantages are beneficial to the application of PPTSVD in the field identification of dynamic axle loads on bridge.