A computational model for optimum process parameters based on factory data and overall liquor rating of black tea

This paper presents a model to find the optimum process conditions for tea manufacturing as well as to predict the black tea quality by implementing a network based tea process parameters monitoring and data logging system. Here, the developed instrument is first calibrated and then implemented to collect the process parameters of tea fermentation and drying. The corresponding tea quality also termed as overall liquor rating (OLR) is collected from tea tasters. Principal component analysis (PCA) is carried out to visualize the pattern of the process parameters. The first two principal components stored 93% useful information whereas more than 6% useful information are stored in the 3rd principal component. It is found from the PCA that maximum samples are clustered in well-defined manner. To study the correlation of the process parameters with OLR, a computational model based on Artificial Neural Network (ANN) has been developed. Non Cross validation (NCV) ANN and Tenfold cross validation (TFCV) ANN models have been trained and tested. Process conditions and corresponding OLR are taken as input and target for the model. 74% classification rate with root mean square error (RMSE) of 0.13 is obtained from the study. The optimum process conditions are found out from the model.