Flexible Capacitive Memristor with Intrinsic Frequency-Domain Perception for Intelligent Mechanical Fault Diagnosis.

Flexible capacitive memristors with frequency-sensitive electrical responses are promising for device-level edge-intelligent sensing and industrial mechanical fault diagnosis. Here, a flexible Ag/SnNb2O6/ITO capacitive memristor is demonstrated, exhibiting stable bipolar resistive switching, pronounced capacitive-coupled behavior, reliable durability over 500 cycles, and retention exceeding 104 s under various conditions, including over 103 bending cycles, different compliance currents (10-104 nA), varying sweep rates (1-128 V/s), and varying bias voltages (±0.2 to ±1.6 V). Notably, under identical voltage and duty-cycle conditions, the device demonstrates fast frequency-domain sensing ability across the range from 1 Hz to 100 kHz. For frequency-dependent mechanical fault diagnosis, combined with fast Fourier transform (FFT) feature extraction and a convolutional neural network (CNN), the device achieves 95.5% recognition accuracy for four slewing bearing states. Mechanistic analysis reveals that the capacitive resistive switching originates from capacitor charging/discharging and ion-modulated Schottky barrier modulation. These results highlight the potential of a flexible capacitive memristor for conformal sensing and intelligent fault diagnosis.