Different denoising methods are used in parameter identification for the dynamic load modeling and the specific approach is proposed. The effects of different denoising methods including mean filtering, medial filtering and wavelet denoising are discussed. Mean filtering method is not helpful to contain the step changes of the measurement voltage, thus is unsuitable for the parameter identification process. Medial filtering method and wavelet denoising methods are suitable for the parameter identification in dynamic load modeling. Furthermore, experiment results based on the measurement data show that the wavelet denoising method is more efficient in some aspects such as the accuracy of identification and SSE.