作者: H ZHANG , M CHANG , J WANG , S YE
DOI: 10.1016/J.SNB.2008.05.008
关键词:
摘要: Abstract In this paper, responses of a gas sensor array were employed to establish quality indices model evaluating the peach indices. The relationship between signals and firmness, content sugar (CS) acidity “Dabai” developed using multiple linear regressions with stepwise procedure, quadratic polynomial step regression (QPST) back-propagation network. results showed that represented good ability in predicting indices, high correlation coefficients ( R 2 = 0.87 for penetrating force CF; = 0.79 CS; = 0.81 pH) relatively low average percent errors (ERR) (9.66%, 7.68% 3.6% CF, CS pH, respectively). provides an accurate model, = 0.92, 0.87, 0.83 respectively) predicted measured values error (5.47%, 3.45%, 2.57% prediction. feed-forward neural network also = 0.90, 0.81, 0.87 (6.39%, 6.21%, 3.13% These prove electronic nose has potential becoming reliable instrument assess