.png){kind=icon}
.png){kind=icon}
Return
Heterogeneous memristive Hopfield neural network under external stimulation: dynamic analysis and hardware implementation
.png){kind=icon}
PRE
AI
W
Z
DOI：10.1007/s11071-026-12952-4.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Biological neural networks consist of heterogeneous neurons operating under time-varying internal and external stimuli. However, the dynamical effects of continuous external stimulation on heterogeneous memristive Hopfield neural networks remain largely unexplored. This paper proposes a heterogeneous memristive Hopfield neural network under sinusoidal electromagnetic stimulation, in which neurons are modeled with hyperbolic tangent (tanh) and Gaussian activation functions, respectively. A flux-controlled memristor is introduced to emulate the electromagnetic coupling between neurons. Equilibrium and stability analyses reveal stimulus-dependent multistability and saddle-node bifurcations. Numerical investigations, including phase portraits, Lyapunov exponent spectra, and bifurcation diagrams, demonstrate that external stimulation can drive transitions from periodic to chaotic dynamics. The system exhibits rich dynamics, including coexisting attractors, transient chaos, and high stimulus sensitivity. The model is validated through both Multisim circuit simulations and breadboard implementation, and the experimental phase portraits are consistent with the numerical results. The proposed system is further applied to pseudorandom number generation. These findings offer theoretical insights into how external stimulation modulates the collective dynamics of heterogeneous neural systems.
Keywords:
Heterogeneous memristive Hopfield neural network
External stimulation
Non-autonomous system
Analog circuit implementation
Journal
IF:
6
Papers:
1.4W
Citations:
4.1W
Organization
Cited Papers