A novel locally active memristive autapse-coupled Hopfield neural network under electromagnetic radiation

In this paper, a memristive Hopfield neural network (HNN) model in complex electromagnetic environment is established by introducing a novel locally active memristor as a connected autapse in neural network, where the effect of electromagnetic radiation is described by a quadratic nonlinear memristor. The non-volatile and locally active characteristics of the proposed locally active memristor are demonstrated by the power-off curve and DC V-I plot respectively. With the autapse connection weight and electromagnetic radiation considered,the memristive HNN model is investigated in detail by theoretical and numerical analysis. Compared with the other neural networks, the abundant dynamics can be observed in the proposed memristive HNN model when the appropriate system parameters are chosen, including stable points, periodic attractors, chaotic attractors, period doubling bifurcation, chaos crisis, coexistence multiple attractors, and so on. Besides, an analog circuit on PSIM and a digital hardware platform on FPGA are implemented to confirm the feasibility of the memristive HNN model and the experimental results are highly consistent with the numerical ones.