Study On Time Series Dynamics And Out-of-Warehouse Quality Model Of Apple Under Controlled Atmosphere Storage

Our country has a huge output of apples.It is difficult to sell all of them immediately after the harvest period.In order to meet the long-term supply needs of the fruit,it must be stored.The main storage method has changed from traditional refrigeration to controlled atmosphere storage.The controlled atmosphere storage can adjust various environmental parameters,making it better than traditional refrigeration in terms of freshness preservation ability,which is the trend of fruit and vegetable storage in the future.The quality of fruits stored after harvest is the key factor to determine apple flavor and improve economic value.As storage time continues to extend,the internal and external quality of apple will change physiologically.The main factors affecting the accuracy of apple quality model during storage include maturity,environment and time.In this study,under the condition of ensuring that the sample maturity is basically the same,the standard controlled atmosphere environment is simulated,and the quality prediction model of the controlled atmosphere storage period is studied in depth.This study provides a theoretical basis and method for apple industry upgrading,quality grading at ex-warehouse and dynamic regulation of ex-warehouse sales during controlled atmosphere storage.The main research contents and results are as follows:(1)The physiological changes and spectral data of apples in different controlled atmosphere storage environments were analyzed.By building platform and designing experiment,the physiological and spectral changes of apples in two long-term storage environments were obtained and explained,which provided a theoretical basis for the establishment of dynamic time-series quality model during controlled atmosphere storage period.The study found that the change rules of the samples in different environments were different.As storage time continues to extend,the variety of color index did not change obviously,and the soluble solids contents(SSC)showed a trend of increasing first and then decreasing.Firmness,titratable acid(TA)and weight loss all showed decreasing trends,and the rate of change of the index of the sample at 4°C is faster than that of the sample at 1°C.The spectral data of the samples from two different environments were compared and principal component analysis was performed.The environmental changes caused the physiological changes of postharvest apples.As storage time continues to extend,the post-ripening process of the fruit accelerated,the starch was hydrolyzed,the soluble pectin increased,and the change of cell structure finally affect the spectral information of apple.(2)Non-destructive testing model for apple quality during controlled atmosphere storage was established.The study found that the generality of the quality model between different environments is poor,and the model of the harvest period cannot effectively predict the samples of the storage period.The PLS method was used to establish prediction model for apple SSC,firmness and weight loss rate.The CARS-SPA method was used to extract the characteristic wavelengths.In the Vis-NIR band,the prediction results are Rp=0.776,RMSEP=0.727 for SSC,Rp=0.605,RMSEP=0.852 for firmness,Rp=0.809,RMSEP=0.819 for weight loss rate,in the LWIR band,the prediction results are Rp=0.913,RMSEP=0.502 for SSC,Rp=0.664,RMSEP=0.883 for firmness,Rp=0.856,RMSEP=0.842 for weight loss rate.Model accuracy needs to be improved.Since PLS fails to consider the influence of time on the model,nonlinear feedback autoregressive neural network NARX is used to establish dynamic time series quality model for controlled atmosphere storage period.The stability of the model is better than PLS linear model.In the Vis-NIR and LWIR bands,the RMSE of model is significantly reduced.(In the Vis-NIR band,R=0.807,RMSE=0.294 for SSC,R=0.885,RMSE=0.376 for firmness,R=0.922,RMSE=0.608 for weight loss rate.In the LWIR band,R=0.831,RMSE=0.268 for SSC,R=0.824,RMSE=0.472 for firmness,R=0.915,RMSE=0.581 for weight loss rate.)(3)Prediction model for the ex-warehouse quality of apples during the controlled atmosphere storage period was established.Based on the analysis of the existing industry standards,it is determined that the best quality standard of apples out of the warehouse is that the firmness is not less than 6.5kg/cm~2,the weight loss rate is not more than 5%as the minimum reference value,the range of SSC higher than 12.5%during storage was selected as the best quality range for sell.Using the TOPSIS method based on the entropy weight method to weight and evaluate the important quality indicators of apples.The entropy weight method can objectively determine the weight and is not affected by subjective factors.Using PLS can basically predict the ex-warehouse score(Vis-NIR:Rc=0.912,RMSEC=0.051,Rp=0.812,RMSEP=0.039,LWIR:Rc=0.937,RMSEC=0.041,Rp=0.933,RMSEP=0.039).Using the PLS and NARX models to predict the storage time of apples,NARX predicts the storage days error within 9 days,and builds the NARX network time series model to predict the ex-warehouse quality.Realize multi-index output and accurate prediction of outbound quality(SSC:R=0.725,RMSE=0.413,firmness:R=0.857,RMSE=0.363,weight loss rate:R=0.914,RMSE=0.432,storage day:R=0.971,RMSE=9.183).Provide accurate delivery methods for the apple industry,ensure the quality of fruit products and increase economic benefits.