Research On Characterization Factors Of Postharvest Quality Change And Low Temperature Shelf Life Prediction Model Of Six Apple Varieties

In this research,the mid-ripening apples "Jonagold","Granny Smith","Golden Delicious",and late-ripening apples "Fuji","Ruiyang" and "Qin Guan",which are widely planted in Shaanxi Province,were collected as the research objects.The quality change of six apple varieties during 0°C shelf and ’Jonagold’ and ’Fuji’ apples with different cold time during 25°C shelf were measured periodically.Analyze and compare the difference in the change trend and change rate of each quality variable between different varieties;Correlation analysis,principal component analysis and cluster analysis were used to select quality change characterization factors;Recursive feature elimination,feature selection Relief F and sparse principal component analysis were used to rank the quality attributes,and combination of quality attributes were established respectively.The BP-ANN shelf-life prediction model was established between the quality attributes combination,quality characterization factors and apple shelf life during 0°C,respectively.The mean relative error,root mean square error and determination coefficient of the predicted and experimental shelf-life were used as the evaluation criteria for the accuracy of the model.The model with high accuracy was selected to build software.The main results are as follow:(1)Six apple varieties("Granny Smith","Golden Delicious","Jonagold","Fuji","Ruiyang" and "Qinguan")stored under the conditions of 0～1℃ and 85%～90% relative humidity crown",the end of shelf life was evaluated by sensory scoring method.The maximum shelf-life were 120,180,180,270,270,and 270 days,respectively.The maximum shelf life of late-ripening apples was higher than mid-ripening varieties.(2)The goodness of fit of the equation established by weight loss,firmness,titratable acid,C value,starch content and 0 °C shelf life of 6 apple varieties was significant.There was no significant difference in the variation rate(slope)of starch content among the three latematuring apple of apple("Fuji","Ruiyang" and "Qinguan"),which was significantly lower than the middle-maturing varieties of "Jonagold" and " Golden Delicious ",and the latter two were significantly lower than "Granny Smith".The change rate(slope)of other quality variables showed significant differences at 4 – 6 levels.(3)The weight loss(about 4%)of medium ripe varieties("Granny Smith","Golden Delicious","Jonagold")can be used as the evaluation standard for the end of shelf life at low temperature.The four quality attributes of solid acid ratio,a* value,L* value and C value can be used as the common quality characterization factors of 6 varieties apple("Granny Smith","Golden Delicious","Jonagold","Fuji","Ruiyang" and "Qinguan")stored at low temperature.(4)"Jonagold" and "Fuji" apples were stored at 0 ～ 1 ℃ and 85% ～ 90% relative humidity for 0,40 and 80 days,0 and 170 days,respectively,and then stored at 25 ℃.The maximum shelf life was 48,48,42,54 and 42 days,respectively.The longer the apple’s cold time,the shorter its 25℃ shelf-life.(5)The change rate of quality attributes of "Jonagold" and "Fuji" apples was different during 25℃ shelf with different cold time.The longer apples are refrigerated,the higher the rate of change of a* value(absolute value of slope)was,the smaller the rate of change of titratable acid,C value and △E value(absolute value of slope)was.The △E value and solid acid ratio of "Fuji" apples with different storage time showed the rule that the later the storage time was,the larger the change rate(slope)was.The other indicators did not show the regular influence of cold time.(6)The quality characterization factors of " Jonagold " apples with different storage time can be selected as weight loss or titratable acid or starch or firmness,solid-acid ratio,L * value,a * value or b * value and △ E value,or solid-acid ratio,weight loss or titratable acid or starch or firmness,a * value or b * value,C value and △ E value.The quality characterization factors of "Fuji" apples with different storage time can be selected as weight loss or titratable acid or starch or firmness or △ E value,solid-acid ratio,a * value or b *value,or solid-acid ratio,L * value,a * value or b * value,weight loss or titratable acid or starch or firmness or △ E value.(7)Sparse principal component analysis was used to rank the quality attributes,and the combination of quality indicators was used as the input variables of BP-ANN.The shelf-life prediction models of six apple varieties("Granny Smith","golden Delicious"," Jonagold ","Fuji","Ruiyang" and "Qinguan")were constructed.The accuracy of validation set was higher than 90%,and the combination of indicators were 1.(a* value,starch,L* value,weight loss,firmness,soluble solids content),2.(a* value,starch,L* value,weight loss,firmness,soluble solids content,titratable acid),3.(a* value,starch,L* value,weight loss,firmness,soluble solids content,titratable acid,b* value),4(a* value,starch,L* value,weight loss,firmness,soluble solids content,titratable acid,b* value,solid acid ratio,△E value,C value).(8)In order to facilitate the follow-up use of the model,the model with high prediction accuracy is embedded in the app.There are four models,which users can choose by themselves.Since this research is only one year’s data,in order to improve the prediction accuracy,the internal parameters of the model should be further adjusted by using years of data if conditions permit.