柔性力学传感器及其基于神经网络算法的应用

Flexible Mechanical Sensors and Their Applications Based on Neural Network Algorithms

摘要:

柔性力学传感器作为可穿戴电子产品不可或缺的一部分，近年来受到了广泛关注，特别是在医疗健康、人机交互、电子皮肤和物联网等领域中的应用。为满足可穿戴设备的需求，柔性传感器在敏感材料、传感器结构设计和制备方法等方面进行了大量研究。同时，神经网络算法作为人工智能的分支，可以高效处理复杂数据，实现多传感器或多模态数据的分析处理，为柔性力学传感器在复杂环境中的应用提供了强大工具。综述了柔性应变和压力传感器及其基于神经网络算法的应用，详细讨论了应变传感器和压力传感器的传感机制和不同类型的传感器的具体应用，如压阻式、压容式、压电式和摩擦电式。神经网络算法显著提高了来自传感阵列以及复杂传感系统的大传感数据的处理效率，并且可以更好地显示感知信号与信息事件之间的关系。随着力学传感器和可穿戴设备的发展，传感器收集到的大量身体活动等生理信号数据集，可以有效训练神经网络算法，从而进一步增强可穿戴设备的性能。此外，还讨论了神经网络算法增强力学传感器在语音识别、手势识别、物体/材质识别和人机交互等方面的应用，并对基于神经网络算法的柔性力学传感器的发展进行了展望。

Abstract:

Flexible mechanical sensors are important components of wearable electronics and have gained significant attention due to their potential applications in various fields such as medical health， human-computer interaction， electronic skin， and the Internet of Things （IOT）. In recent years， efforts have been made to improve the performance of wearable devices by analyzing and constructing flexible sensors in terms of sensing materials， structural design， and preparation methods. Meanwhile， neural network algorithms， as a subset of artificial intelligence， can efficiently process complex data and analyze multi-sensor or multi-modal data. These features make them powerful tools for the application of flexible mechanical sensors in complex environments. This paper provides a comprehensive review of the current state of development of flexible strain/pressure sensors and their combined neural network algorithms. We discuss the sensing mechanisms of strain and pressure sensors， including piezoresistive， piezocapacitive， piezoelectric， and triboelectric types， and provide a detailed overview of their potential applications. Additionally， we introduce neural network algorithms for strain and pressure sensors and discuss their specific applications. Neural network algorithms significantly improve the processing efficiency of large sensing data and can better display the relationship between sensed signals and information events. With the development of mechanical sensors and wearable devices， large datasets of physiological signals such as physical activity can effectively train neural network algorithms and further enhance the performance of wearable devices. In addition， applications of neural network algorithms to enhance mechanics sensors for speech recognition， gesture recognition， object/material recognition， and human-computer interaction are also discussed. Finally， an outlook on the development of flexible mechanical sensors based on neural network algorithms is provided. We believe that with continuous development， neural network algorithms can be effectively used to enhance the performance of wearable devices in medical health， human-computer interaction， and electronic skin applications.