Dynamical System Identification of Complex Nonlinear System Based on Phase Space Topological Features

Dynamical system identification of complex nonlinear system can be regarded as parameter estimation task of uncovering the nature of the system, which is of much importance for many engineering applications. Sea clutter is one kind of complex nonlinear system, which has intrinsic complexity and high nonstationarity. Therefore, it is difficult to detect marine targets due to the low Signal-Noise-Ratio(SNR) and low Signal-Clutter-Ratio(SCR). To improve the performance of marine target detection in, a method to discover fundamental geometric structures from high-dimensional and multivariable data sets is proposed. The method can improve the target detection performance and its main novelty lies in phase space reconstruction and topological features extraction, based on the dynamical system identification scheme and topological data analysis (TDA) techniques. Experiments with measured sea-clutter data demonstrate the performance of the method proposed in this work.