MEMS Gas Sensor Array Fault Diagnostic Unit for Microsystem Applications

With the rapid development of the Internet of Things and environmental sensing systems, intelligent microsystems based on MEMS (Micro-Electro-Mechanical Systems) sensor arrays for environmental sensing have broad application prospects and advantages in fields such as environmental monitoring, medical diagnostics, and battlefield perception. These systems, particularly those aimed at machine olfaction, integrate MEMS gas sensor arrays with pattern recognition algorithms to monitor gases in the environment in real-time. The performance of these systems heavily depends on the accuracy of the MEMS gas sensor array signals. However, due to the characteristics of the sensitive materials and integrated manufacturing processes, environmental changes can cause failures in the MEMS gas sensor arrays, leading to abnormal operation of the entire microsystem. Therefore, this paper designs a fault diagnosis unit for MEMS gas sensor arrays tailored for microsystem applications, measuring 19mm x 19mm x 3mm. This unit can identify five types of sensor faults: shock, bias, constant output, power loss, and precision degradation. It also features fault isolation and localization capabilities, providing a basis for data recovery in case of MEMS gas sensor failures. The fault diagnosis unit employs a multi-task fault diagnosis algorithm based on 1D CNN-LSTM, achieving a fault identification accuracy of 99.93%, fault isolation accuracy of 99.43%, and fault localization accuracy of 98.12%. It occupies 194k of FLASH and 20k of SRAM, with an average running time of 234ms and a power consumption of 11.58mW.