Time-Domain Parameter Detection of Weak Signals in the Chaotic Background

Digital oscilloscope can not measure week singal in chaotic background. A method using Elman neural network is described to achieve signal parameter detection in chaotic background. With the phase space reconstruction theory on time series, the embedded dimension is calculated and used as the in-put dimension of a neural network considered. By adopting the single-step prediction method, the weak signals are detected directly and their waveforms can be gained as well in the chaotic state. Result shows that the method studied in this paper is superior to the existing detection principles. Its feasibility and practicability have been proved by the experiments.