Fuzzy controller based E-nose classification of Sitophilus oryzae infestation in stored rice grain. (15th June 2019)

• Record Type: Journal Article
• Authors: Srivastava, Shubhangi; Mishra, Gayatri; Mishra, Hari Niwas
• Is Part Of: Food chemistry. Volume 283(2019)
• Page Start: 604
• Page End: 610
• Publication Date: 2019-06-15
• Subjects: Electronic nose -- Fuzzy analysis -- Infested rice -- Principal component analysis -- Sensors; Food -- Analysis -- Periodicals; Food -- Composition -- Periodicals; 664
• DOI: 10.1016/j.foodchem.2019.01.076
• Languages: English
• ISSNs: 0308-8146

Abstract

Highlights: Fuzzy controller algorithm via E-nose was used for S. oryzae infestation detection. Out of 18 metal oxides sensors in E-nose, 6 were selected by fuzzy triplet score. Defuzzified data obtained was subjected to PCA and MLR techniques. MLR was cross validated with reference analysis methods of protein and uric acid. PCA (84.75%) and MLR (97.2, 99.7%) values confirmed reliability of algorithm. Hybrid fuzzy E-nose is authentic tool for detection of insects in food grains. Abstract: Fuzzy controller artmap based algorithms via E-nose selective metal oxides sensor (MOS) data was applied for classification of S. oryzae infestation in rice grains. The screened defuzzified data of selective sensors was further applied to detect S. oryzae infested rice with PCA and MLR techniques. Reliability of data was cross validated with reference methods of protein and uric acid content. Out of 18 MOS, 6 sensors namely P30/2, P30/1, T30/1, P40/2, T70/2 and PA/2 showed maximum resistivity change. Defuzzified score of 62.17 for P30/2 and 59.33 for P30/1 MOS further confirmed validity studies of E-nose sensor response with reference methods. The PCA plots were able to classify up to 84.75% of rice with variable degree of S. oryzae infestation. The MLR values of predicted versus reference values of protein and uric acid content were found to be fitting with R 2 of 0.972, 0.997 and RMSE values of 2.08, 1.05.