Development Of Portable Potato Late Disease Detector Based On Spectrum Technology

Potato is the third largest economic crop in the world and one of the four staple foods in China.China's total planting area has ranked first in the world,but the average mu yield is far below the world average.Potato late blight is the primary factor restricting potato production.Traditional potato disease detection methods are difficult to accurately and quickly detect and diagnose potato disease severity and physical and chemical indicators.Therefore,it is necessary to explore a method for detecting diseases that is accurate,fast and non-destructive,accurately identify late blight and carry out accurate disease grading,and provide a theoretical basis for rapid prevention and control.In this study,potato leaves were used as research objects.The spectral information of the corresponding areas of potato leaf samples was collected by designing and constructing a portable detector based on fiber spectrum,and combined with stoichiometry to physicochemical values of potato leaves infected by late blight.Detection and disease grading studies.Through the research results,the software system of the portable detector was designed and optimized,and finally the diagnosis of the late blight period of potato leaves and the disease grading function of the disease stage were realized.The main work of this paper is as follows:(1)The shell of the portable potato late blight detector based on spectral technology was designed and a hardware system was built.Among them,the hardware system is mainly composed of a power supply module,a control module,a display module and an acquisition module,and hardware selection and connection are performed according to the functions of each module.The connected hardware is mounted in the detector housing to complete the prototype of the portable detector.Using the Python language to write programs in the main controller Raspberry Pi,the instrument can initially realize the spectral data collection function.(2)Research on the detection of POD activity and SPAD value by portable detector.The PLS prediction model of optimal full-spectrum data and POD activity values was established by comparing the modeling effects of various pre-processing methods.The PLS and MLR prediction models of spectral data and POD activity values under optimal pretreatment were established by extracting the characteristic wavelengths by x-LW method.Among them,the PLS model of spectrally transformed full-spectrum data and POD activity value has the best prediction effect,Rp is the maximum of 0.944 in all models,and RMSEp is the smallest of all models and is 2.764.The PLS prediction model of optimal full spectrum data and SPAD values is established by comparing the modeling effects of various preprocessing methods.The PLS,MLR and BP-ANN models of spectral data and SPAD values under optimal preprocessing were established by using SPA to extract feature wavelengths.The study found that the average smoothed-processed full-spectrum data and the SPAS value of the PLS model predicted the best,Rp was 0.950 in all models,and RMSEp was the smallest of all models and 2.786.(3)Classification and identification of potato leaf diseases under late blight stress.By analyzing the kinetic equations and the change curves of reflectance value,lesion size and physical and chemical values with time,two disease classification and recognition models based on physical and chemical values and reflectance values of potato leaf late blight were established.(4)Designed and developed the software system and human-computer interaction interface of potato late blight detector.The programming of the main system,spectral acquisition system,spectral display system,computing system and disease grading system was completed by Python language,and the graphical interface was designed by Tkinter toolkit.Through the graphical interface program,the above five systems are connected and implanted into the Raspberry Pi of the portable detector,and finally realize the functions of data acquisition,prediction physical and chemical value,disease identification and grading of the instrument.