study on firmness and texture changes of pear fruit when loading different forces and stored at different periods using artificial neural network

This study evaluated the effect of different dynamic and static loadings and different storage periods on the firmness of pear fruit. pear fruit was first segregated into three groups of 27 pear in order to undergo three loadings: static thinedge compression loading, static wideedge compression loading and dynamic loading. all loaded pears were stored in accordance with three storage period designs: 5day storage, 10day storage, and 15day storage. following each period, the variations of pear texture were scanned by using the ctscan technique as a nondestructive test. then, the firmness of pear texture was measured using a penetrometer. data were simulated and evaluated using mlp and rbf artificial neural networks. the results showed that with increasing storage time and loading force , the firmness significantly decreased (1% level) in all three types of loading, in addition, pear texture was destructed under dynamic compression loading in order to compare with other two loadings. best value artificial neural network for wide edge loading (12 neuronrbf) was (r2 wide edge= 0.9738– rmse wide edge=0.3419 mae wide edge =0.268) and for thin edge loading (4 neuronrbf) was (r2thin edge = 0.9946– rmse thin edge =0.170977 mae thin edge =0.133), also for dynamic loading (8 neuronrbf) was (r2 dynamic loading = 0.9933– rmse dynamic loading =0.230 mae dynamic loading= 0.187).

Study on Firmness and texture changes of pear fruit when loading different forces and stored at different periods using artificial neural network

This study evaluated the effect of different dynamic and static loadings and different storage periods on the firmness of pear fruit. Pear fruit was first segregated into three groups of 27 pear in order to undergo three loadings: static thinedge compression loading, static wideedge compression loading and dynamic loading. All loaded pears were stored in accordance with three storage period designs: 5day storage, 10day storage, and 15day storage. Following each period, the variations of pear texture were scanned by using the CTScan technique as a nondestructive test. Then, the firmness of pear texture was measured using a penetrometer. Data were simulated and evaluated using MLP and RBF artificial neural networks. The results showed that with increasing storage time and loading force , the firmness significantly decreased (1% level) in all three types of loading, In addition, pear texture was destructed under dynamic compression loading in order to compare with other two loadings. Best value artificial neural network for wide edge loading (12 neuronRBF) was (R2 Wide edge= 0.9738– RMSE Wide edge=0.3419 MAE Wide edge =0.268) and for thin edge loading (4 neuronRBF) was (R2Thin edge = 0.9946– RMSE Thin edge =0.170977 MAE Thin edge =0.133), also for dynamic loading (8 neuronRBF) was (R2 Dynamic loading = 0.9933– RMSE Dynamic loading =0.230 MAE Dynamic loading= 0.187).