Applied back propagation neural network and machine vision for modelling and controlling turmeric (Curcuma domestica Val.) drying process

Turmeric is the largest herbs and spices potentials of Indonesia. However, the postharvest technology turmeric is still inadequate. Drying is one of the post-harvest processing technologies, to reduce the water content of a food with thermal energy such as sunlight or mechanical equipment. This technology is a new method for predictive modeling of turmeric drying process for on-line monitoring and controlling of this process. It can optimize the drying process to increase value. A back propagation neural network (BPNN) was developed to predict the model of turmeric drying process in a hot air dryer. BPNN inputs were read mean, blue mean, and green mean at time and output was water content at time t + Δt. The results showed that used BPNN model had better performance than conventional model. The best BPNN graphic model is 0,004 MSE and 25,33% ARE for training set and 0,003 MSE and 20,25% ARE for validation set that built 0.6 of learning process and 0.3 of momentum rate. This model could predict the water content of turmeric at time t + Δt by knowing the input data at time t. Also, this BPNN model can used for on-line control of the turmeric drying process.