Hui Tian,∗ Wenshen Jia,∗,∗∗,∗∗∗,∗∗∗∗ Jie Ma,∗ Jihua Wang,∗∗,∗∗∗,∗∗∗∗ and Jianxiong Hao∗∗∗∗∗
Electronic nose, sensor, principal component analysis, bakepropaga-tion neural network∗ College of Automation, Beijing Information Science and Tech-nology University, Beijing, China; e-mail: [email protected];{jiawenshen; mjbeijing}@163.com∗∗ Department of Beijing Research Center for Agricultural Stan-dards and Testing, Beijing Academy of Agriculture andForestry Sciences, Beijing, China; e-mail: [email protected]∗∗∗ Department of Risk Assessment Lab for Agro-products(Beijin
To reduce huge economic losses caused by apple deterioration to growers and dealers, a high-sensitivity, low-cost portable electronic nose detection system used for rapid detection and early-warning of apple deterioration was designed. The system comprises an electro- chemical sensor array, data collection and transmission module and human–machine interaction software. Firstly, data collected by the electronic nose was pre-processed with smoothing filtering proce- dures, then a non-destructive detection model for refrigerated-stored apples was constructed on the basis of bionic olfactory technology in combination with bionics, principal component analysis (PCA), algorithms of kernel principal component analysis (KPCA), back- propagation neural network (BPNN) and support vector machine (SVM). The experiment results show that seven-point smoothing filtering can produce better outcomes, and models on the basis of backpropagation (BP), SVW, PCA+SVM and KPCA+SVM all reach a correct recognition rate of more than 90%, with correct recognition rates of PCA+SVM being higher than those of BP, SVM and KPCA+SVM. This research provides significant reference value for research focusing on electronic nose applied in non-destructive rapid detection and early-warning of apples.
Important Links:
Go Back
