Modeling memristor switching behavior through phase-delay cluster analysis

Memristors are promising components of modern microelectronics, yet accurately simulating their switching dynamics remains a challenge due to the complexity of underlying physical processes. This work introduces a novel framework for memristor evolution modeling, incorporating a first-of-its-kind clustering approach based on the state change rate. Our method integrates statistical estimation of the conditional probability for each cluster with the NMRG neural network model to generate realistic time-current switching sequences. We validate our approach using an anionic ZrO${}_2$/TaO${}_x$-based memristor, demonstrating that it produces physically plausible switching trajectories. The proposed model offers a promising tool for third-party simulation systems, enabling accurate characterization of memristor behavior.