Research And Development Of Edible Rate Non-destructive Detecting Method And Grading Equipment For Pomelo

China has large production of pomelo,where pomelo has good quality,numerous varieties and wide planting area.Honey pomelo is a high-quality variety of pomelo,having the characteristics of rich nutrition and good taste,and is deeply loved by the majority of consumers.However,there is a lack of advanced technology in post-harvest commercialization processing of pomelo in China,thus missing the economic benefits brought by grading sales according to internal and external quality.As an important internal quality index of pomelo,the edible rate plays an important role in making sales strategy of pomelos and the full use of them.However,in practice,the traditional manual measurement method of edible rate is time-consuming,and has destructive damage to pomelo.Therefore,it is urgent to develop a method of detecting edible rate,which is nondestructive,rapid and having potential use in online detecting and grading.Two cultivars of pomelo,namely ‘Hongrou’ pomelo and ‘Guanxi’ pomelo,were taken as the object in this research.In this study,the physical parameters and X-ray image features of samples were used to predict the edible rate,then an online X-ray detection module and a set of grading equipment were designed according to the postharvest commercial processing requirements of pomelo.Firstly,the prediction effect of the model based on the different combination of physical parameters on the edible rate of pomelo was explored,and the initial grading of the edible rate was realized.Secondly,the prediction effect of the model based on the different combination of X-ray image features on the edible rate of pomelo was explored,and the model was optimized by fruit shape index.The quantitative prediction of the optimized model was studied,and the pomelos were classified by the edible rate.Finally,a set of online detecting and grading equipment for pomelo’s internal and external quality was designed,and the simulation of key mechanism and the design of online X-ray detection module were completed.The main research contents and results of this paper are as follows:(1)The prediction method of edible rate of pomelo based on physical parameters was studied.The results showed that density was the most important feature in the prediction of edible rate among the physical parameters,and the multiple linear regression model established by mass,volume and density had the best performance.Using this feature combination,the residual prediction deviation(RPD)value of‘Hongrou’ pomelo,‘Guanxi’ pomelo and hybrid samples prediction models were 2.08,2.05 and 2.17,and the root mean square error of prediction(RMSEP)were 2.86%,1.91%and 2.52%,respectively.The performance of the prediction model was good and could basically meet the requirements of qualitative discrimination.Then,the above three features were used to train a two-grade classifier to judge whether the pomelo was qualified(edible rate ≥ 60%)or not.The discriminant accuracy of ‘Hongrou’ pomelo,‘Guanxi’ pomelo and hybird sample in prediction set reached 85.71%,96.77% and90.80%,respectively.(2)The prediction method of edible rate of pomelo based on X-ray image method was studied.Firstly,image enhancement and feature extraction of X-ray images were completed.The features included 2D edible rate and 2D flesh content.2D edible rate was the area ratio of the flesh region to the whole fruit,and 2D flesh content was the logarithmic sum of the gray levels of each pixel in the pulp region.Secondly,in terms of the prediction of the edible rate,three features of density,axial 2D edible rate and transversal 2D edible rate were selected to complete the full combination of multiple linear regression modeling and analysis.The RPD values of the best models of‘Hongrou’ pomelo and ‘Guanxi’ pomelo were 2.52 and 2.62,and RMSEP were 2.08%and 1.46%,respectively.Then,in order to avoid the influence of pomelo shape on the axial 2D edible rate,the model was modified and optimized by the shape index.After the optimization,the feature combinations of the best model were shape index,axial2 D edible rate and transversal 2D edible rate.The results showed that the RPD values of the best models of ‘Hongrou’ pomelo,‘Guanxi’ pomelo and hybrid sample were 2.78,2.82 and 2.83,and the RMSEP were 1.99%,1.43% and 1.74%,respectively.In terms of the prediction of the flesh content,the feature combinations of the best model were axial 2D flesh content and transversal 2D flesh content.The RPD values of the best models of ‘Hongrou’ pomelo,‘Guanxi’ pomelo and hybrid sample were 2.92,2.05 and2.63,and the RMSEP were 53.0g,58.6g and 62.7g,respectively.Finally,the three features of the optimal model were used to train a two-grade classifier and a three-grade classifier for predicting the edible rate of pomelo.The two-grade classifier took 60% as the dividing line and the classification accuracy of ‘Hongrou’ pomelo,‘Guanxi’ pomelo and hybird sample in prediction set reached 93.10%,96.77% and 96.67%,respectively.The three-grade classifier took 60% and 70% as the dividing line,and the classification accuracy of ‘Hongrou’ pomelo,‘Guanxi’ pomelo and hybird sample in prediction set also reached 89.66%,93.55% and 91.67%,respectively.(3)A set of online detecting and grading equipment for the internal and external quality of pomelo was studied and designed.Firstly,according to the requirements of pomelo grading and ergonomics,the functional module layout and main parameters(height,speed of each conveying module,etc.)of grading equipment were determined.Secondly,the kinematic simulation of classification mechanism and free tray confluence mechanism was completed.Three classification mechanism including blocking type,hitting type and inclined guide wheel type were comparative analysised,and the confluence mechanism included two-channel confluence and free tray recycling confluence.Then,a set of online X-ray detection module with application potential was designed,and the parameter requirements of the module were presented.Additionally,the calculation of ray protection and the work flow introduction were completed.Finally,through the steps of detection module layout,grading channel design,parameter design and so on,the design of online detecting and grading equipment for the internal and external quality of pomelo was completed.