Study On Walnut Quality Evaluation And NIR Predictive Model Establishment

The walnut is one of the most important special oils and woody oil crops,which is popular among consumers for its rich nutrition and unique flavor.During the acquisition and storage process,it is crucial to quickly and accurately evaluate the quality of walnut kernel.In this study,basic component content(oil content,protein content,moisture content)and oxidation index(acid value,peroxide value)of walnut kernel were studied during the storage process.After the analysis of factors influencing the near infrared detection,the calibration models of the walnut kernel quality indicator were established,which was expected to provide technical support for the rapid detection and control of oils and nuts quality.The qualitative model of walnut shell rate was established using near infrared spectroscopy coupled with chemometrics.The difference of spectra of empty shell and walnut with kernel was mainly distributed in the wavelength bands of 1450nm,2100?2150nm and 2305?2345nm.The optimum measurement position for the sample was walnut belly(lay flat).The different states of walnut shell(smooth shell and wrinkled shell)had a greater impact on the repeatability and stability of the spectrum,while the different varieties of walnuts had relatively little influence.It was feasible to detect percentage of empty grain by near-infrared spectroscopy.The optimized modeling conditions were determined as follows:detection wavelength band 780?1100nm,SNV preprocessing,and PLS model,which can obtain the recognition rate 100%.Through the analysis of the quality change of basic component content in walnut kernel,it can be seen that the order of the influence of storage conditions on the content of basic components of walnut kernels is:water>fat>protein.The optimum detection conditions of the walnut kernel basic component content were 20 meshes and 20 mm thickness.By comparison,the optimized modeling conditions of oil content and protein content were determined as follows:detection wavelength band 780?2500nm,SNV+1^st pretreatment,and PLS model.Under these conditions,the root mean square error of prediction(RMSEP)of the prediction model were 0.0069 and 0.0039,respectively.The optimum detection conditions of moisture content were as follows:detection wavelength band 1349?1490nm,SNV+1^st pretreatment,and PLS model.Under these conditions,it can obtain the RMSEP of the moisture content 0.0016.Through the analysis of the quality change of oxidation index in walnut kernel,it can be seen that the order of the influence was:temperature>light.The proper storage conditions were low temperature,dark,and in-shell storage.The influence of moisture content and sample pretreatment on the near-infrared detection of walnut kernel oxidation index was investigated in detail.For 1.5%change in moisture content,the average coefficient of variation of spectrum increased to 0.023.Drying treatment can both reduces the change in moisture content,and increases the acid value and peroxide value by 20.7%and 5.9%,respectively.Thus,it was feasible to detect walnut kernel acid value and peroxide value by near-infrared spectroscopy.The optimized modeling conditions of acid value were determined as follows:detection wavelength band 1800?2200nm,SNV+De-trending pretreatment and PLS model.After computation,the R²of low acid value and high acid value were 0.7663 and 0.9539,respectively.The optimized modeling conditions of peroxide value were determined as follows:detection wavelength band 780?2500nm,normalize pretreatment,and PLS model,resulting in the RMSEP of the peroxide value 0.0067.