Study On Dehydration Characteristics And Hyperspectral Characterization Of Maize Seeds Under Different Freeze-drying Conditions

As a kind of life material,maize is especially important to people’s production and life,and the quality of maize seed is the key to ensure the yield of maize.Freeze-drying is a method of removing water by sublimating ice crystals from frozen materials.It can obtain higher quality seeds and increase yield.At present,Bloom spectrum imaging technology non-destructive testing is a trend.In this paper,hyperspectral imaging technique was used to detect the moisture content of freeze-dried maize seeds.Corn seeds were freeze-dried by vacuum freeze dryer.According to differential scanning calorimeter(DSC),the eutectic point of corn seeds was-11.63℃ and the eutectic point was 3.58℃.Five different freeze-drying parameters were set,namely,freezing temperatures(-25℃,-30℃ and-35℃)and drying time(120min),pre-freezing rates(20℃/min,30℃/ min and 40℃/ min),sublimation drying temperatures(3℃,5℃,7℃ and 9℃)and analytical drying temperatures(35℃,40℃ and 45℃).The physical properties and activity of frozen-dried corn seeds were measured by grinder,drying box and portable headspace analyzer.The results show that the moisture content is the lowest at-35℃,which indicates that the free water freezes completely and evaporates more at-35℃.The moisture content decreased with the increase of sublimation drying time.The higher the analytical temperature,the lower the moisture content,but the slower the decreasing trend.The moisture content first decreased and then increased with the rise of sublimation temperature.When the sublimation temperature was 5℃,the moisture content approached the melting point of maize seeds,and the sublimation was completely dry.The moisture content of maize seeds increased with the increase of pre-freezing rate.The other series indexes of maize seed were correlated with moisture content.When the moisture content is small,the muck rate is big,the particle size index is small,the hardness is big,the brittleness is small,the bulk density is big,the conductivity is related to the activity,the activity is big,the conductivity is small.The optimum freeze-drying process is as follows:pre-freezing rate is 40℃/min,freezing temperature is-35℃,sublimation drying time is 8h,sublimation temperature is 5℃,and analysis temperature is 35℃.The hyperspectral imaging technology in the range of 900-1700 nm was studied.ENVI5.3 analysis software was used to process the data,and the original spectral image of corn seeds was cropped,denoised and the average spectrum was extracted.The Unscrambler9.7 was used for four different pretreatments,and the spectral curves of different reflectivity intervals were obtained.The results show that the fluctuation trend of spectral reflectivity is similar in different freeze drying groups,but the range of reflectivity is different.The lower the freezing temperature,the higher the analytical temperature,the higher the sublimation temperature and the lower the prefreezing rate,the longer the sublimation drying time is,the higher the spectral reflectivity is.The hyperspectral imaging technology in the range of 900-1700 nm was used to collect the hyperspectral images of corn seeds after freeze-drying,and the original spectral data were pretreated differently.The characteristic bands were screened by principal component analysis and the prediction model of moisture content of corn seeds was established by combining partial least square regression method.The experimental results show that the PLSR model of water content established in different freezing temperature groups under S-G-SNV has the best accuracy.The PLRS model of water content established under MSC with different sublimation drying time groups has the best accuracy.The PLSR model of water content established by different analytical temperature groups under SNV pretreatment has the best accuracy.The water content PLSR model established by different sublimation temperature groups under S-G-MSC pretreatment has the best accuracy.The PLSR model of water content established under S-G-MSC pretreatment with different pre-freezing rates has the best accuracy.