With the rapid development of electrification, both productivity and human civilization have made rapid progress and electrification has led to more and more fires. Therefore, it is of great significance to strengthen the investigation and prevention of fires, especially the confirmation of the origin of fire. Considering the post-combustion residual characteristics of common flammable production raw materials (carbon dioxide and dust) at the scene of the fire, this paper designs a detection system based on odor sensor for rapid detection, data analysis and recognition. The system designs a multi-function sensor array for detecting dangerous production raw materials (carbon dioxide and dust), and develop a good method for spot sampling. It introduces embedded technology and uses it together with chromatographic technology and sensor array, achieving the accurate detection of the combustion residues of dangerous production raw materials (carbon dioxide and dust) is realized. The detection signals are post-processed by feature extraction technology and pattern recognition algorithm. The research results are helpful for rapid analysis and recognition of residues, providing certain technical support for the recognition for the cause of fire.