Deep-learning assisted biomimetic self-powered wireless electronic noses system enabled by triboelectric discharge

Electronic noses, inspired by the mammalian olfactory system, have revolutionized various industries including food, environment, medicine, and rescue. They detect gases by using sensor arrays and provide real-time information. However, current electronic noses demand batteries as the power source, limiting their applications and requiring frequent maintenance. Wireless transmitter modules can also increase battery dependence and introduce high energy consumption. Moreover, gas sensors used in electronic noses lack multidimensional information identification ability. Here, we have developed a self-powered, wireless electronic nose system based on triboelectric discharge. By converting mechanical energy into high-voltage power, this system can initiate gas discharge and produce wireless signals that contain valuable gas information. With deep learning, it can recognize up to 60 different types of gas atmospheres with 93.8% accuracy, including 5 gas components with 12 different gas pressures. The system with a visualization program is designed for wireless, real-time gas monitoring toward applications in health monitoring, precise manufacturing, and high-tech fire rescue operations.