A blind identify algorithm for weak signals in chaotic noise

For a kind of weak harmonic wave signal in chaotic noise, this paper proposes a new blind identify algorithm which based on generalized regression neural network (GRNN) to estimate parameters of chaotic system, and then apply reiteration blind deconvolution method to remove remains noise in the system, and extract weak signal from chaotic background. From 2597 observation points, we obtain mean deviation as 0.0112, and peak value signal-noise ratio as -44.77 dB. Simulation result shows efficiency of the algorithm.