| Abstract |
: |
A pneumatic conveying recirculated dryer is very suitable for sago starch drying and thus the present research has designed a PCRD machine. It aimed to develop an Artificial Neural Network (ANN) model to estimate the fineness modulus of sago starch dried using a PCRD machine. The designed ANN model structure used consisted of 12 input neuron, three variations of hidden layers, and 1 output neuron, with three topological variations, namely 12-5-5-1-1, 12-10-10-1-1, and 12-15-15-1-1, as well as used the learning algorithm backpropagation. The validity test for the ANN model generated an R2trained value by 0.859 or 85.9%, and an R2testing value by 0.576 or 57.6%. This suggests that the ANN model is valid enough to be employed to estimate the fineness modulus of sago starch dried using a PCRD machine. Optimization of the ANN model for the training and testing process generated the lowest MRE and MAE values for the variable vu, namely by 3.785% and 0.0004%, and LAcrb, namely by 13.214% and 0.0012%, respectively. This indicates that variables with a significant effect on determination of the fineness modulus are the speed of the air of the dryer and the length of the upper outlet pipe in the recirculation cyclone. |
