Evaluation of Industrial Roasting Degree of Coffee Beans by Using an Electronic Nose and a Stepwise Backward Selection of Predictors

Online monitoring of coffee roasting in an industrial plant is becoming an important issue as the experience of the roast master still plays an important role. Despite several approaches have been tested, some limitations were not surmountable as difficulties in scalability from bench scale to industrial roaster, the use of expensive analytical instrumentation, and the need to handle a large dataset of variables. In this paper, response of an electronic nose sampling, the headspace of roasted beans, was correlated with brightness and mean density, using the generalized least square regression in combination with a stepwise backward selection of predictors. To avoid scalability issues, roasting took place in an industrial plant using two Arabica (Brazil and Costa Rica) and two Robusta (Vietnam and India) origins. Regression showed R2 ranging in the interval 0.994–0.999, with statistical significance p < 0.0001. The present approach has the potential to be used effectively instead of roast master, in the online monitoring of coffee roasting in industrial plants.