Abrupt event oriented robust gas concentration detection method of MOX gas sensor arrays

Metal-oxide (MOX) gas sensor array and pattern recongnition method are the core compoents of machine olfactory systems. The accuracy and reliability of measurements of MOX gas sensor array directly influence the analysis performance of subsequent pattern recongnition method. With the expanding of appied range of machine olfactory systems, abrupt events, such as external interferences or gas sensor failures, greatly deteriorate the reliability of the measurements acquired by MOX gas sensor arrays in the complex work environment. In this paper, an abrupt event oriented robust gas concentration detection method based on principal component analysis (PCA) and sparse representation classification (SRC) is proposed to improve the reliability of the analysis reults of machine olfactory system. The proposed strategy is fully evaluated in a real machine olfactory system. The experimental results demonstrate that the proposed gas concentration detection method can effectively monitor the running status of each gas sensor mounted on the gas sensor array and improve the robustness of machine olfactory system when sensor faults or external interferences occur.