AI-driven ocean monitoring with multimodal triboelectric nanogenerator: Self-sustainable real-time wave warning and forecasting system

Abstract

Conventional ocean monitoring systems utilizing single-mode triboelectric nanogenerators (TENGs) are fundamentally limited by their dependence on unimodal signal acquisition, which results in a critical lack of recognition accuracy and early warning reliability. To address this, we propose a highly integrated, multimodal self-powered AI-enhanced monitoring system (SAMS) for diverse ocean state monitoring. SAMS combines solid-solid and liquid-solid TENG modes, incorporating three distinct triboelectric conversion mechanisms. SAMS features a spherical framework with a freestanding-layer electret generator on its lower surface, detecting subtle wave vibrations through continuous liquid-solid contact. The upper surface features a double-electrode electret generator, enhanced via oxygen plasma treatment, which sensitively captures intermittent liquid-solid interactions (e.g., splashes and scours) under high-intensity waves, producing signals up to 80 V. Internally, a spiral electret generator with a dual-spiral structure generates in-plane and out-of-plane vibrations, delivering outputs of up to 100 V and significantly expanding detectable wave motion ranges. The triple-modal design of the SAMS enables simultaneous generation from three signal channels. Assisted by deep learning, the SAMS achieves a substantial improvement in wave level recognition accuracy, from 41.25 % (single-mode) to 96.25 % (triple-mode). This work advances multimodal TENGs for intelligent marine monitoring and enables real-time energy harvesting and state monitoring in complex marine environments.