Abstract: The single channel signal separation is an important issue in signal processing field, and it is particularly difficult when lacking of observations. In this paper, we proposed an optimization method whose objective function is based on the temporal autocorrelation function of the underlying signal. The initial value of the optimization and the temporal autocorrelation were estimated by using the wavelet transform modulus maxima method. Then the optimal solution is the estimation of the underlying signal. Experimental results showed that the algorithm can extract the evoked potential signal well with single-trial from single channel.