Copper and Iron Oxide Sensor Arrays Combined With SVM Neural Networks to Achieve Classification and Concentration Prediction of VOCs Gases

CuFe2O4/ $\alpha $ -Fe2O3 and CuO/ $\alpha $ -Fe2O3 heterojunctions were synthesized by solvothermal modulation of the Cu to Fe ratio, and highly sensitive volatile organic compounds (VOCs) sensing materials were successfully prepared for the self-building of sensor arrays. The measured responses of the sensor arrays were functionally fitted and later combined with a support vector machine (SVM) algorithm to classify and predict the concentration of VOCs gases. The optimal working temperature of the four types of sensors prepared in this article is $240~^{\circ }$ C, and the optimal working temperature of the previously working sensors is $260~^{\circ }$ C. Then, the dynamic corresponding test of VOC gas from 10 to 100 ppm at their respective optimal operating temperature showed good response results for the seven types of sensors. Among them, S500 has the best response to five VOC gases, and its response to ammonia, benzene, methanol, formaldehyde, and n-butanol at 100 ppm is 5.07, 5.08, 11.76, 10.10, and 46.67, respectively. The sensor array has good performances to VOCs gas in the simulated experimental environment, and the combination with SVM algorithm achieves good results in VOCs classification and concentration prediction.