Predictive algorithm for duffing demodulation system by Kalman gain

Generally, traditional identification algorithm of Duffing oscillator bases on the transitions of phase diagram, which often needs a large number of cumulative inputs to transform the system from chaos to the large-scale periodic state, which limits the applications of Duffing oscillator. A new algorithm based on Kalman gain is proposed in the paper to predict state transitions of the Duffing oscillator before the transitions in the phase diagram. By the establishment of Duffing state equations and setting the control condition of measurement equations, Kalman gain of Extend Kalman Filter (EKF) which are used to estimate the Duffing system can effectively predict the state changes of the Duffing oscillator, experiments of demodulations of communication signals show that the predictive algorithm can not only reduce at least 50% input points to identify the phase transitions of oscillator, but also the demodulation accuracy is obviously improved.