Topology Estimation of Uncertain General Complex Dynamical Networks from Noisy Time Series

This paper addresses the problem of simultaneous estimation of the topological structure and unknown parameters of uncertain general complex networks from noisy time series. Usually the complex networks consist of known node models with some unknown parameters and uncertain topological structure. At the same time, only partial states with heavy noise can be observed in real-world complex networks. By means of the unscented Kalman filter (UKF), we estimate the unknown states, parameters as well as topological structure with high accuracy only from partial heavily noise-corrupted states of the nodes. The simulation results verify the effectiveness of the proposed approach.