Premium
Contamination Predictions of Cape Hake Fillets during Display and Storage by Artificial Neural Network Modeling of Hexadecanoic Acid
Author(s) -
Venter Pierre,
Swanepoel Hanita,
Lues Ryk J.F,
Luwes Nicolaas
Publication year - 2016
Publication title -
journal of food process engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.507
H-Index - 45
eISSN - 1745-4530
pISSN - 0145-8876
DOI - 10.1111/jfpe.12241
Subject(s) - hake , contamination , artificial neural network , food science , fish <actinopterygii> , chemistry , computer science , environmental science , artificial intelligence , fishery , biology , ecology
This study aimed to design an artificial neural network ( ANN ) that could distinguish between C ape hake fillets displayed and stored on ice that have been exposed to excessive contamination and those that were not. The selected variable was a biochemical indicator, hexadecanoic acid, a fatty acid. C ape hake fillets with and without excessive contamination was kept on ice and analyzed every 48 h over a period of 10 days. A novel ANN was designed and applied, which provided an acceptable prediction on the contaminated fillets based only on the hexadecanoic acid changes during day 8 ( T 4) and day 10 ( T 5). The ANN consisted of a multilayered network with supervised training arranged into an ordered hierarchy of layers, in which connections were allowed only between nodes in immediately adjacent layers. The network consists of two inputs, T4 and T5 connected to two neurones that are connected to one output neuron that indicates a prediction on contamination of the fillets. These two neurons are connected to one output neuron that indicates a prediction on contamination of the fillets. Practical Applications The model sets the stage for the development of alternative quality control measures for retailers and buyers of fish and other foods that contain fatty acids such as hexadecanoic acid to provide saver food.