Modeling of pH and acidity for industrial cheese production

A three-layer feedforward neural network was successfully used to model and predict the pH of cheese curd at various stages during the cheese-making p rocess. An extended database, containing more than 1800 vats over 3 yr of p roduction of Cheddar cheese with eight different starters, from a large che ese plant was used for model development and parameter estimation. Neural n etwork models were developed with inputs selected among 33 quantitative and qualitative process variables for final pH of cheese, pH at cutting, and a cidity at whey drawing-off and at pressing. In all cases, very high correla tion coefficients, ranging from 0.853 to 0.926, were obtained with the vali dation data. A sensitivity analysis of neural network models allowed the relative import ance of each input process variable to be identified. The sensitivity analy sis in conjunction with a priori knowledge permitted a significant reductio n in the size of the model input vector. A neural network model using only nine input process variables was able to predict the final pH of cheese wit h the same accuracy as for the complete model with 33 original input variab les. This significant decrease in the size of neural networks is important for applications of process control in cheese manufacturing.