Abstract: Removing the noise in the experimental data signals and aligning the angle and force signal sequences of the initial data and the test data are the two key issues that remain to be re-solved in the large-amplitude-oscillation test.The basic principle of all-phase digital filtering and the method of processing unsteady experimental data are introduced.The theoretical formula of all-phase data processing is derived in detail.Spectrum analysis of each channel of the balance is used to choose the cutoff frequency in the digital filtering process.The three methods of moving phase,interpolating fitting and Moving Least Square(MLS)fitting are studied in detail for alig-ning the initial data with the test data.The experimental results obtained by the three methods are compared.The results show that all-phase digital filtering can remove the noise in the experi-mental signal data successfully and reduce the signal phase and amplitude distortion.The experi-mental results obtained by the three different methods are consistent and reasonable.