In order to accurately detect power quality of transient signal, a high-efficient power algorithm is presented with a dual-tree complex wavelet denoising method based maximum posteriorim( MAP) estimation is used to preprocess power quality disturbance signal． Firstly, using the maximum posteriorim estimation of improved wavelet threshold to disturbance signals were denoised, then for different decomposition layer detail coefficient noise variance and signal variance estimation, and calculate the layer threshold of different decomposition, thus out noise threshold. Simulation experimental results show that the proposed algorithm has simple steps, fast calculation speed and easy to realize, the denoising effect is ideal, has the very good application prospect.