Research And Development Of On-line Near Infrared Detection System And Its Application In Rapeseed Oil Content Detection

It is of great importance to obtain information on the chemical composition of certain cereals and oil crops in their harvesting,acquisition,storage and transport,processing and other stages in real time,and it is also one of the important technological means for improving the effective resource utilization of agricultural commodities and the level of deep processing.In this paper,an on-line near infrared spectrum detection system,which can be mounted on the production line,was developed.The detection system can be used to monitor the quality information of grain and oil crops in real time,and adjust the production process parameters in time,and improve the processing quality of the products.The main research contents are following:(1)Built a pre-experiment platform for the study of different spectral acquisition methods,and find the optimal spectral acquisition method by comparing the spectral stability of several spectral acquisition methods.The study revealed that the effect of the distance between the optical fibre head and the sample surface on the spectral stability showed that the optimum optical fibre acquisition range for soya beans was 90mm,the rapeseed was 70 mm and the flour was 70 mm.In addition,the analysis showed that the spectral stability of the Angle between the center line of the fiber head and the center line of the light source showed that the optical fiber detection Angle could be 10°while the angle of the light source was 10°.(2)Completed the design of the spectral collection module structure,the design of the lift platform structure and the selection of the conveying unit of the online near-infrared detection system,focusing on the selection of light sources,the near-infrared micro-spectrometer,the phase motor control,etc.When the device is in operation,the upper computer gives different pulse signals to the drive system in the spectrum acquisition module to monitor the rotation of the turntable,which enables easy swapping of dark field spectrum acquisition,regular plate spectrum acquisition,reference spectrum acquisition and sample spectrum acquisition.The light source is off-stage during the collection of dark field spectrum and the light source is on-site in other collection modes;the scheme to use the forward and reverse rotation of the stepper motor to accomplish the lifting of the lifting platform is planned and,in order to ensure the stable operation of the lifting platform,the two sets of safety security measures are planned.(3)The outcome of the evaluation of the optical performance parameters of the spectrum acquisition module show that the performance indexes of wavelength accuracy,wavelength repeatability,absorbance repeatability and baseline stability satisfy the practical requirements.Rapeseed is selected as the test sample,and the rapeseed oil content calibration model is developed.K-S method was used to partition the NIR spectra of 102 samples by a ratio of 7:3.Seven pretreatment methods including Raw,S-G,1Der,2Der,normalization,SNV and MSC are used,as well as five wavelength optimization methods including iPLS,GA,SPA,CARS and RC are used for comparison in tandem with PLS modeling.The result after 1Der pretreatment is the best and the R_C,RMSECV,R_P,RMSECP are 0.9221,0.7812,0.9136,0.8122respectively.As well as the results after the GA selected wavelength is the best,and the R_C,RMSECV,R_P,RMSECP are 0.9383,0.7121,0.9311,0.7652 respectively.Finally,this test system was used to detect the oil content of 10 non-modeling sets of rapeseed samples.The standard deviation was 0.7349,which was not more than the RMSECP value of the model.The model could be used for the testing in actual production line.