n Maftoonazad - Department of Agriculture Engineering, Research Center of Agriculture and Natural Resources, Zarghan, Fars, Iran,
y Karimi - Department of Food Science, McGill University, Macdonald Campus,۲۱,۱۱۱ Lakeshore Ste-Anne-de-Bellevue
H. S. Ramaswamy - Department of Food Science, McGill University, Macdonald Campus,۲۱,۱۱۱ Lakeshore Ste-Anne-de-Bellevue
S.O Prasher - Department of Food Science, McGill University, Macdonald Campus,۲۱,۱۱۱ Lakeshore Ste-Anne-de-Bellevue

Hyperspectral observation were performed to characterize spectral features and Artificial Neural Network (ANN) models were used for predicting quality changes in coated and non-coated avocados during storage at different temperature. Avocados werecoated using a pectin-based coating and stored at different temperatures (10,15, 20ºC), along with control samples. At different intervals, avocados were removed from storage and respiration rate, total color difference, texture and weight loss were measured.The most effective spectral data were chosen by Principal Component Analysis to design multilayer neural network models forprediction of respiration quality parameters. The optimal configuration of neural network model was obtained by varying the main parameters of ANN: transfer function, learning rule, number of neurons and layers, and learning runs. Results indicated thatcompared to conventional mathematical models, ANN has more feasibility to predict of quality changes in avocado fruits. Models developed for firmness, weight loss and total color difference had better fitness than respiration rate

Avocado, pectin-based edible coating, texture, total color difference, respiration rate, weight loss, hyper spectral image analysis, Artificial neural Network