Removing noise from the output of the measurement system is important for accurate acquisition of transient voltage waveforms. Consequently

a de-noising method based on wavelet decomposition is proposed. By calculating the energy and energy increment of wavelet coefficients in each layer of the original noisy signal

the wavelet decomposition layer with the best de-noising effect is determined. Moreover

the peak-to-effect ratio of the wavelet coefficients of each layer is calculated to determine the optimal wavelet basis. De-noising simulation experiments on transient voltage signals and sinusoidal signals show that the proposed scheme can correctly select the optimal wavelet decomposition layer and the optimal wavelet base. The de-noising scheme is applied to the transient voltage signals in the laboratory and the substation site. The results show that the scheme can preserve the high frequency signal of the signal while removing noise.