Computerization of Stumbo's method of thermal process calculations using neural networks

The four heat penetration parameters in Stumbo's method of thermal process calculations were correlated using artificial neural networks (ANN). The pr ocess involved the development of two different artificial neural network m odels, one named ANNG for the parameter g (the difference between the retor t and food center temperatures) and the other named ANNFU for the parameter f(h)/U (the ratio of heating rate index to the sterilizing value). Both th ese models replace the 57 tables developed by Stumbo for assessing steriliz ing effects. The ANNG model deals with estimating the process time for a gi ven process lethality and involves g as the dependent (output) variable whi le f(h)/U, z (representing the temperature interval difference that causes a tenfold change in decimal reduction time), and j(cc) (the cooling rate la g factor) are taken as the independent (input) variables. The ANNFU model i nvolves the prediction of the lethality of a given process with the f(h)/U being taken as the dependent variable and z,j(proportional to) and g as the independent variables. In developing each of the ANN models, several confi gurations were evaluated: (i) the input and output parameters were taken on a linear scale, (ii) the input and output parameters were taken after the transformation of some or all the input and output parameters using a logar ithmic scale to the base 10, and (iii) all input and output parameters were transformed using a logarithmic scale to the base two. The optimum ANN mod els, ANNG and ANNFU, were those of the third configuration. ANNG involved a network with six neurons in each of the three hidden layers while ANNFU in cluded 16 neurons in each of the two hidden layers. The two optimal ANN mod els are capable of predicting the g and f(h)/U parameters in the range give n in Stumbo's tables. In each instance, the predicted values were in close agreement with those listed in the tables. In addition, the developed ANN m odels can predict the intermediate values of any combination of inputs. The refore, they eliminate the need for excessive storage requirements of table s and interpolations while computerizing thermal process calculations using Stumbo's method. (C) 2000 Elsevier Science Ltd. All rights reserved.