Abstract:Oxidation tank is important equipment for gold bio-oxidation extraction. In the northwestern region in China with obvious season features, various operating failures of temperature control system of oxidation tank running in extreme weather environment arise inevitably. Failure of timely finding out will lead to drop of gold extraction rate and waste of mineral resources. Therefore, a fault classification method for oxidation tank temperature control system based on improved EM algorithm is proposed, which can quickly and accurately identify specific faults. The simulation results show that improved EM algorithm can make full use of advantages of industrial big data and has higher classification accuracy rate than that of traditional EM algorithm algorithms and BP neural network algorithm.