MODELING OF SQUID PROTEIN HYDROLYSIS: ARTIFICIAL NEURAL NETWORK APPROACH

The processing of squid generates nearly 50% of raw material as a waste by-product rich in protein. Use of this by-product as a raw material in the manufacture of a squid protein hydrolysate would add considerable product value to the industry. Enzyme hydrolysis of squid waste protein is a complex process because of a number of inherent simultaneous inhibition and enzyme inactivation reactions which occur during hydrolysis. Artificial neural networks (ANNs) are very effective in developing predictive models for processes involving complex reaction kinetics that would otherwise be difficult to develop by more traditional deterministic approaches. The objective of this work was to develop a kinetic model to describe the kinetics of enzyme hydrolysis of squid waste protein using ANNs. A series of enzyme hydrolysis experiments were carried out on samples of squid waste under specified conditions of temperature, pH, and initial enzyme and substrate concentrations. Experimental data in the form of substrate concentration over time were taken as real time course data. These data were fitted with a cubic spline procedure for use in an ANNs training process to estimate reaction rates for the model. The effect of a number of hidden processing elements on the error in prediction was studied, and the ANN showing necessary prediction performance was constructed. Maximum differences between experimental and predicted values of substrate concentration at any point in time ranged from 0.098 to 0.29 g/L (34%). Correlation coefficients between predicted and experimental results were also, respectively, high, ranging from 0.988 to 0.992.