Gas Classification System Based on Hybrid Waveform Modulation Technology on FPGA

Abstract

With the growing global emphasis on environmental protection, new energy vehicles have become essential for reducing carbon emissions in the transportation sector. However, safety issues related to lithium-ion batteries, particularly thermal runaway, remain a critical concern. Different stages of thermal runaway produce distinct gas compositions, necessitating sensors with high selectivity for targeted detection of specific gases or gas categories. Dynamic measurement technology using temperature modulation can enhance the selectivity of semiconductor gas sensors. However, most dynamic measurements yield limited data features for gas categories, complicating subsequent classification algorithms and making them less suitable for deployment in embedded devices. To address these challenges, this study proposes an electronic nose system based on hybrid waveform modulation technology. By employing multi-waveform superposition heating, this approach enriches data features corresponding to gas responses and optimizes sensor technology and data processing algorithms using ARM+FPGA architectures, significantly improving system accuracy. The system collects gas sensor data via a sensor array and achieves a recognition rate of 95.82% using the MLP algorithm, successfully deployed on Xilinx’s System-on-Chip (SoC) platform.