Abstract: This paper proves the proposition that complex Gaussian wavelet can satisfy the Mercy conditions and exhibits reproduction features in Hilbert space.A special kernel function named complex Gaussian wavelet kernel function is built,which is combined with principal component analysis(PCA), to identify parameters and forecast future information of the non-linear dynamic signals.A multi-parameter synchronous optimization method is proposed to shorten the optimizatoin time in multi-parameter models.Simulation experiment results validate the feasibility and(effectiveness) of the method.