Analysis of critical control points in deviant thermal processes using artificial neural networks Original Research Article

The successful implementation of a scheduled thermal process is of paramount importance in ensuring the safety of heat processed foods. It is therefore important to consider all associated critical factors and their relative importance in process calculations. The objectives of this study were (i) to evaluate the relative order of importance of different critical control variables with respect to process calculations and (ii) to develop predictive models to compensate for their deviations. The critical variables studied were: retort temperature (RT), initial temperature (Ti), cooling water temperature (Tw), heating rate index (fh), heating lag factor (jh) and cooling lag factor (jc). Their ranges of deviation from a set point were selected as −2 to 2 °C for RT, −5 to 5 °C for both Tw and Ti, −2 to 2 min for fh, and −0.2 to 0.2 for both jc and jh. Artificial neural network (ANN) models were developed and used for analysis of different critical variables with respect to their importance on the accumulated lethality (F), process time (PT), cooling time (CT), and total time (TT) under the given processing conditions. The results indicated that within the deviation ranges, the relative order of importance of critical variables were as follows: for F, RT>fh>jh>TiTw>Ti>Tifh>RT Ti>jc>RTjh; for PT, RT>fh≫jh>Ti>jc>TiTw; for CT, jc>Tw>fh; for TT, RT>fh>jh>jc>Tw>Ti>Tijc>TiTw. When the desired F value was set at 6±0.5 min, the maximum acceptable deviation ranges of different variables were: ±0.3 °C for RT, ±4 °C for Ti, ±0.1 for jh, ±0.8, ±1, ±1.2 min for fh at fh=20, 40, 60 min, respectively, and ±0.4 for jc. Finally, the combination effect of deviation of multiple variables on F, PT and CT were analyzed. ANN models could be effectively used for identification of critical control points, and for correcting process deviations, both important from the point of view of implementation of HACCP approach in thermal processing.