Artificial Intelligence Techniques for Enhancing Thermo-Electro-Mechanical Responses of MEMS Resonant Accelerometers with an Attention-Guided Siamese Fusion Neural Network: Trends and Challenges

Abstract

Microelectromechanical systems (MEMS) resonant accelerometers have become essential sensing components in precision engineering applications such as inertial navigation, structural monitoring, aerospace systems, and autonomous vehicles, due to their high resolution and frequency-based digital output. However, their performance is significantly affected by complex thermo-electro-mechanical interactions, including temperature variations, electrostatic nonlinearities, damping effects, and fabrication inconsistencies, which are difficult to model using traditional analytical methods. This review examines recent advancements in artificial intelligence techniques aimed at improving these sensors, with particular emphasis on attention-guided Siamese fusion neural networks. These architectures enable effective multi-branch feature extraction and fusion, allowing the system to learn correlations between thermal, electrical, and mechanical inputs while filtering noise. Enhanced with attention mechanisms, they dynamically prioritize relevant features, improving robustness and accuracy. The review covers both simulated and experimental datasets across diverse applications, and discusses optimization approaches such as transfer learning, Bayesian tuning, and multi-task learning. It concludes that integrating AI with multiphysics modeling offers a pathway toward intelligent, self-adaptive MEMS sensors, while highlighting challenges like limited datasets, interpretability, and hardware constraints.