کد مقاله | کد نشریه | سال انتشار | مقاله انگلیسی | نسخه تمام متن |
---|---|---|---|---|
225180 | 464480 | 2008 | 8 صفحه PDF | دانلود رایگان |
![عکس صفحه اول مقاله: Prediction of brightness and surface area kinetics during coffee roasting Prediction of brightness and surface area kinetics during coffee roasting](/preview/png/225180.png)
The amount of heat transferred to green coffee beans is essential in the coffee roasting. During this process, several parameters can be used as indicators to determine the degree of roasting (color, aroma, volume, bean temperature). Consequently, two predictive models, using artificial neural networks, are proposed to determine the quality of coffee roasting. A bean brightness model took account of bean temperature (simulated by a physical model) and roasting time. A second model predicting the bean surface area, focused on roasting air temperature and roasting time. The color changes affecting coffee beans during the process were studied experimentally in a pilot roaster equipped with a CCD video camera and a lighting system consisting of two small optic fiber spotlights. Arabica green coffee beans of Colombian origin were roasted using different air temperatures (190, 200, 210, … 300 °C), for 10 min. Two separate feedforward networks with one hidden layer were used to brightness and surface kinetics. The best fitting training data set was obtained using three neurons in the hidden layer, which enable prediction of brightness and bean surface kinetics with an accuracy that was at least as good as the experimental error, over the entire experimental range. Using the validation data set, simulations and experimental data were in good agreement (R2 > 0.98). This study showed that real-time simulations were possible so that the roasting process could be stopped when the simulated brightness was similar to a target fixed by the roast master. The model thus developed could contribute to the on-line estimation of product quality, thus providing a parameter to control coffee roasting.
Journal: Journal of Food Engineering - Volume 89, Issue 2, November 2008, Pages 156–163