Abstract
Rapid test methods with portable devices along with standard chemical tests are necessary to determine raw syrup quality in the sugarcane agro-industries. On this account, a special e-nose device was developed to test the sugarcane syrup and its association with the odor emitted from it to determine the amount of sucrose (purity) in the sugarcane syrup. Samples were obtained from the farms of Hakim-Farabi agro-industry, including four varieties (CP57, CP69, IRC99-02, and CP48). Experiments included chemical tests to determine the percentage of purity (PTY) and refined sugar (RS) plus an electronic nose test. Partial least squares (PLS), principle component regression (PCR), multiple linear regression (MLR), and artificial neural network (ANN) methods were used to evaluate the correlation between the gained signals from the sensor array and chemical analysis results of the samples. In the case of PTY, among 8 sensors, MQ3, MQ5, and MQ9 had the highest response compared to the others, while regarding RS, all the sensors except for MQ8 indicated a great contribution. Also, all models for PTY and RS showed a good prediction performance. The results revealed that ANN model, with topology 8-1-2, outperformed others for prediction of the quality indices of sugarcane, with high correlation coefficients (R(2) = 0.96 for RS; 0.99 for PTY), and relatively low RMSE values of 0.33 for RS; 0.4 for RTY. Finally, findings indicated that e-nose technique has the potential to become an authentic tool to assess chemical features of sugarcane syrup from e-nose system signals.