Non-destructive Detection Of Kiwifruits,peaches And Pears Quality Based On Near-infrared Spectroscopy Analysis

Fruit is the third largest agricultural products besides food crops and vegetables in our country.The fruit yield of our country is highest in the world, but fruit quality is uneven and low degree of commercialization.These factors make low exports and seriously influence the development of fruit trade industry. At present,domestic fruit quality testing mainly rely on traditional detection methods.There are many non-destructive testing resear_ches,but are mainly limited to laboratory phases with fussy process.In order to explore a quick method in detecting fruit quality,a near infrared diffused spectroscopy(12000~4000cm-1) technology was used to study the internal relations between fruit quality indexes and spectrum with kiwifruits(Huayou and Xixuan no.2),peaches(Shahong,Beijing no.8 and Laishanmi) and pears(Dangshansu and Xuehua).Partial least squares (PLS) regression was carried out to analyze the spectra.Mod-els of kiwifruits during cool storage,pears during growth and post-havest were establis-hed.SSC,firmness and acidity models of three kinds of peaches were also established. In addition, qualitative analysis was tried to use to identify the breeds of kiwifruits,peaches and pears.This thesis mainly include three parts:1.analyze the spectrogram of kiwifruits during cool storage,establish SSC,firmness and pH models of known samples and forecast unknown samples;2.establish SSC and firmness quantitative models of three kinds of peaches,screen out optimal modeling conditions;3.establish SSC and firmness models of pears during growth and posthavest,analyze the model applicability in forecasting;4.identify the breeds of three kinds of fruits by clustering analysis and di-scriminant analysis methods.The main resear_ch results were as follows:(1)To a certain extent,spectra changes of kiwifruits could reflect quality changes during cool storage.Spectra measure point modeling was better than whole sample modeling.(2)The best SSC,firmness and pH calibration models of Xixuan no.2 kiwifruits c-ould be obtained by the original spectra with the correlation coeffiecient of calibration (r_c) of 0.951,0.971,0.882 and the root mean square error of calibration(RMSEC) of 0.622,1.243,0.196.The three quality indexes of unknown samples were predicted by the cali- bration models with the correlation coeffiecient of prediction(r_p) of 0.866,0.939,0.816 and the root mean square error of prediction(RMSEP) of 1.040,1.711,0.238.(3)To the three kinds of peaches,the SSC quantitative analysis results were better than that of the hardness quantitative analysis.To Shahong and Laishanmi peaches,the best calibration molels of SSC could be obtained by the MSC spectra with r_c of 0.964 and 0.928,RMSEC of 0.265 and 0.371.Some unknown samples were predicted by the models with r_p of 0.886 and 0.900,RMSEP of 0.351 and 0.671.To Beijing no.8 peaches,these indexes were 0.937,0.346,0.920 and 0.668 respectively.Moreover,the hardness quantitative analysis was not ideal.The best calibration models of firmness in Shahong and Beijing no.8 could be obtained by original spectra with r_c of 0.760 and 0.836,RMSEC greater than 2.The results in Laishanmi was better with r_c of 0.875 and RMSEC 2.364.In the three models,forecasting precision of Laishnmi was better than the other two,with r_p of 0.769 and RMSEP of 2.765.Shahong and Beijing no.8 models neither could predict firmness exactly.(4)The precision of SSC model was very high during growth and posthavest in Dangshansu pears, give the r_c of 0.936 and 0.949, the RMSEC of 0.152 and 0.174,the r_p of 0.894 and 0.889,the RMSEP of 0.242 and 0.271 respectively.The growth SSC model was appropriate for predicting unknown samples a month or so before maturity.The precision of posthavest firmness model was not high to predict unknown samples exactly.(5)Clustering analysis and discriminant analysis method were used to identify breeds of kiwifruits, peaches and pears.The results showed that two methods could both identify kiwifruits succeedfully.Clustering analysis could identify peaches and pears, but with low precision.Discriminant analysis could not identify peaches and pears.This study provided a rapid detection method for fruit quality during havest,storage and posthavest,but also provided a certain theoretical direction for grading and shelf life of fruits.