The cross sensitivity of metal oxides semiconductor（MOS）to volatile organic compound（VOCs）is a common problem

so the mixed gases can not be analyzed accurately by single sensor.In order to solve this problem

a sensor array is designed based on a variety of semiconductor metal oxides and a hardware system platform is built

as well as using three algorithms to analyze and show the 35 groups of mixed gas data of benzene and formaldehyde.The experimental results show that the metal oxide array combined with random forest algorithm has the best classification prediction effect with an overall prediction accuracy of more than 95 %.