Iranian Research Organization for Science and Technology (IROST) Innovative Food Technologies 2783-350X 9 4 2022 07 23 Investigation and modeling of lemon properties with different coatings under storage conditions with artificial neural network and regression models Investigation and modeling of lemon properties with different coatings under storage conditions with artificial neural network and regression models 289 307 1157 10.22104/ift.2022.5534.2097 FA Fatemeh Moradi Ganjeh Former student of Mechanics of Biosystems Engineering Department, Faculty of Agricultural Rasoul Meamar Dastjerdi Assistant Professor, Department of Mechanics of Biosystems Engineering, Faculty of Agricultural Engineering and Rural Development, Khuzestan Agricultural Sciences and Natural Resources university, Mollasani, Iran. Mohammad Hadi Movahednejad Assistant professor of water and soil Department, Agriculure Faculty, Shahrood University of Technology 0000-0002-0833-828X Mokhtar Heidari Department of Horticulture, Faculty of Agriculture, Khuzestan Agricultural Sciences and Natural Resources university, Mollasani, Iran. Journal Article 2022 03 08 The application of edible coatings is one of the most effective ways to maintain the quality of fruits. In this study, the quality of sweet lemon fruit was evaluated in a completely randomized factorial design with edible coating of chitosan-clay nanocomposite in three levels, olive oil and carnauba wax in comparison with uncoated samples during refrigeration for 4 months. During the storage period of sweet lemon fruits, chemical characteristics (acidity, brix and ascorbic acid) as well as mechanical properties (maximum skin tensile force, fruit skin punch modulus) were measured. The results showed carnauba wax and chitosan-clay nanocomposite had better performance in maintaining lemon quality and mechanical properties than samples without coating. Moreover, among the coatings used, 5%chitosan-clay nanocomposite coating had higher preference than other coatings. In this research, Artificial Neural Networks (ANN) , linear and nonlinear regression method were used to predict the quality of lemon The results showed that the ANN has a better forecasting accuracy to predict the lemon properties compared to linear and nonlinear regression models and the LM learning algorithm with tansig transfer function had the best result. The best fit for the qualitative parameter was acidity with the coefficient of determination (R|) of 95%. The result of sensivity analysis indicated that the highest sensitivity coefficient was obtained for the punch modulus against the time feature with 47.96%. The application of edible coatings is one of the most effective ways to maintain the quality of fruits. In this study, the quality of sweet lemon fruit was evaluated in a completely randomized factorial design with edible coating of chitosan-clay nanocomposite in three levels, olive oil and carnauba wax in comparison with uncoated samples during refrigeration for 4 months. During the storage period of sweet lemon fruits, chemical characteristics (acidity, brix and ascorbic acid) as well as mechanical properties (maximum skin tensile force, fruit skin punch modulus) were measured. The results showed carnauba wax and chitosan-clay nanocomposite had better performance in maintaining lemon quality and mechanical properties than samples without coating. Moreover, among the coatings used, 5%chitosan-clay nanocomposite coating had higher preference than other coatings. In this research, Artificial Neural Networks (ANN) , linear and nonlinear regression method were used to predict the quality of lemon The results showed that the ANN has a better forecasting accuracy to predict the lemon properties compared to linear and nonlinear regression models and the LM learning algorithm with tansig transfer function had the best result. The best fit for the qualitative parameter was acidity with the coefficient of determination (R|) of 95%. The result of sensivity analysis indicated that the highest sensitivity coefficient was obtained for the punch modulus against the time feature with 47.96%. https://jift.irost.ir/article_1157_a8240cb8235e9c493a0c30607586166c.pdf