CONSTRUCTION OF A METHOD FOR PREDICTING THE NUMBER OF ENTEROBACTERIA IN MILK USING ARTIFICAL NEURAL NETWORKS

It is now established that artificial neural networks (ANNs) provide better simulation and prediction of the number of microorganisms in raw materials and foodstuffs. In this case, ANNs could be used as informative, fast, and cost-effective means. According to the European requirements to food products, basic microbiological indicators are the total number of microorganisms and bacteria from the Enterobacteriaceae family, since they are most commonly associated with food-borne diseases and poisonings. The aim of this work was to devise a method for predicting the number of bacteria from the Enterobacteriaceae family in raw milk at its chilled storage and to estimate the predictive capability of ANN. Construction of the method included 4 stages. At the first stage, we examined the number of enterobacteria depending on the physical-chemical composition of raw milk, temperature and duration of storage in a refrigerator. At the second stage, we compiled a base of experimental data obtained from research models. At the next stage, we introduced the received database to ANN. And at the last stage we assessed effectiveness of the predicting technique. The constructed ANN consists of three layers: an input layer (5 parameters: milk storage temperature (4, 6, 8, and 10 °C), duration of milk storage (from 1 to 48 hours); the acidity of milk (17‒20 %), the fat content (3.2; 3.6; 4.0; 4.5 %) and protein content (2.9; 3.0; 3.3 %) in milk; hidden layers (with 30 neurons) and the output layer (the projected number of bacteria). In order to train and optimize the ANN, we used 1,200 experimental data, which revealed that the prediction had the highest rate of deviation of 2.497 % (or 370 bacterial cells per 1 ml). Thus, the devised predicting method could be used to predict the number of bacteria taking into consideration the complex of environmental variables in different food products. In addition, a given approach could be employed as artificial intelligence when assessing microbiological risks and for quick monitoring of food safet