| Near infrared spectroscopy(NIRS)is the fastest developing modern analytical technology with high efficiency,high-speed,high data output,low technical requirements,no damage and pollution since 1980 s.However,the current experience of fertilization and management can not meet the current situation of production,so a scientific and effective rapid detection method is urgently needed to provide a strong theoretical basis for walnut production.1.A quantitative prediction model for rapid and nondestructive detection of chlorophyll content,reducing sugar content and soluble sugar content in walnut leaves was established.In the experiment,the walnut leaves of "Wen 185" in South Xinjiang were selected as the research objects,and the original spectra were analyzed by near infrared spectroscopy,including multi scatter correction(MSc),standard normal variable transformation(SNV),first derivative and second derivative The PLS quantitative detection model of chlorophyll content,reducing sugar content and soluble sugar content in walnut leaves was established in the range of 10000-4000cm-1 in the whole band.The results showed that after the original spectrum was smoothed by MSC + 1std + SG,the best factor number was 10.The correlation coefficients(R)of chlorophyll content and soluble sugar content correction model were 0.93775,0.89809,0.124 and 0.0496,0.399 and 0.114,respectively.The model was ideal and accurate The results show that it is feasible to detect the chlorophyll content and soluble sugar content of "Wen 185" Walnut Leaves Based on near infrared spectroscopy.For the prediction model of reducing sugar content in walnut leaves,the correlation coefficient(R)value is 0.54534,the corrected root mean square(RMSEC)value is 0.0145,and the predicted root mean square(RMSEP)value is 0.0159.The accuracy of the model is poor.2.The correlation coefficients of available nitrogen model,available phosphorus model,available potassium model and RMSECV were 0.77941,0.95279,0.81847,16.5,42.3,19.1,19.1,30.6,31.3,0.86,1.38 and 0.61,respectively.The results show that the real value of the available P model is highly correlated with the predicted value,and the prediction result of the available K model is good,but the prediction result of the available N model is general.It is feasible to use this method to predict the available N,P and K in the soil,and it can be used for the rapid detection of the soil in the walnut plantation in South Xinjiang Provide reference and basis for measurement.3.Taking "Wen 185" walnut kernel in South Xinjiang as the research object,16 different chemometrics algorithms such as SNV,MSC,derivative processing and smoothing were used to preprocess the original spectrum.The results show that SNV combined with first derivative smoothing is the best method.Based on PLS and chemical analysis,the "wen185" walnut kernel protein and soluble sugar models were established with high accuracy and stability.The correlation coefficients(R)of the models were 0.89411,0.98268,RMSEC were 0.505,0.01650,RMSEP were 0.941,0.01650,respectively.The actual value and the predicted value of the validation set were 0.941 and 0.01650,respectively The results showed that the linear correlation was strong,which satisfied the determination of protein and soluble sugar content in walnut kernel.However,the model of reducing sugar content detection has over fitting phenomenon and poor accuracy,which needs to be further optimized. |
PDF Full Text Request