Multiple attractors and chaos synchronization of memristor-based Hopfield neural networks

Abstract This study investigates the dynamic characteristics and synchronisation abilities of Hopfield neural networks (HNNs) that are augmented by memristor technology. Memristors are novel electronic devices that can retain information concerning their previous resistance states, enabling them to function as synapses within neural networks. Substituting traditional synapses in an HNN with memristors yields an innovative chaotic model within the memristor-based HNN framework that can support single- and double-scroll attractors. The model’s intricate dynamic behaviour, encompassing aspects such as multistability and the coexistence of attractors, was analysed. The practical validation of this concept was carried out via circuit construction. On the basis of this model, we designed a sliding mode control system, achieving fixed-time synchronisation and laying the groundwork for future applications having this feature, such as image encryption.