It is generally known that the arti

It is generally known that the artificial neural networks were proposed on account of the study achievements of modern neuroscience and they consist of a host of processing units which are closely interconnectedcite{ref11078}. Even as they reflect the basic characteristics of the human brain, the natural neural networks are still difficult to be truly depicted. Artificial neural networks use resistors as biological synapses in circuit implementations. One of main reason is that resistor does not possess memory function just as neuronal synapsecite{ref11079}. Memristor have memory characteristics as well as nanoscale, and can simulate biological synapses when information is transmitted between different neurons, in terms of these merits different types of memristive artificial neural networks have been proposed in recent yearscite{ref11074,ref11075,ref6,ref22,ref57}, which are nonlinear dynamical systems with state-dependent switching and have been extensively studied, and applied in different fields cite{ref11074,ref11075,ref28,ref50}.