Methods Study On Early Warning Of Facility Vegetable Diseases Based On Disease Triangle Theory

Facility agriculture is an important form of vegetable production in China,the industry of facility vegetable in china is developing rapidly,and its proportion of the cultivation area and output is increasing year by year,China has become the first major country in the production of facility vegetable,but the situation of vegetable disease seriously reduced the yield and quality of vegetables,which brought great economic loss to our country,therefore,the effective early warning of vegetable disease is very important.There are many problems in the early warning methods of plant diseases,such as single algorithm,incomplete model,model timeliness and accuracy.To solve these problems,the relationship between the three factors(environment,pathogenic microorganism and host plant)was analyzed based on the basic theory of vegetable disease triangle in this paper,with iot sensor technology,electronic microscopy and optical technology for technical support,accurately access to the environment,pathogen spores image and vegetables leaf spectral information,and combined with image processing and machine learning algorithm,an early warning model of plant disease was constructed.This model improves the comprehensiveness,timeliness and accuracy of early warning,provides timely and effective vegetable diseases forecasting for farmers,provides method support for the early warning research of agricultural diseases for China.The main research results of this paper are as follows:(1)From the angle of time,according to the influence sequence of the three elements of disease on facility vegetable,analyzed the influence of environment on the pathogen infection,the pathogenic mechanism of the pathogen and the mechanism of the disease resistance of vegetables.Through monitoring environment,the number of spores and the spectrum information of leaves can provide methods for early warning of vegetable disease and it also make the early warning more comprehensive.(2)An early warning model of plant diseases based on environmental information was proposed.Adverse environment is an important factor of vegetable diseases,based on the environmental mechanism of vegetable diseases,firstly,an environmental information acquisition scheme based on iot sensor technology was designed,secondly,the collected environmental data were combined with the mean based recursive method,finally,the prediction model of plant disease damage was constructed based on SVR combining with the disease index and temperature humidity data,the correlation coefficient R of the model was 97.15%,the mean error was 13.65%,and the accuracy was higher.(3)An auxiliary early warning model for plant diseases based on the microscopic image of pathogen spores was proposed.Infection of pathogenic spores is a direct cause of vegetable morbidity,firstly,using electron microscopy to obtain the image of pathogen spores,using wiener filtering to reduce the image of noise,then the contour extraction was completed by the edge detection of prewitt operator and global threshold segmentation;secondly,using the watershed algorithm and morphology to remove impurities and complete the segmentation of adhesion spores;finally,using the target marker to complete the automatic counting of pathogen spores.The average relative error was 5.66%,which can replace the artificial count,increase the counting efficiency and provide auxiliary reference for facility vegetable disease early warning.(4)An identification and early warning model of plant disease damage based on leaf spectral information was proposed.Through spectroscopy can get infected leaves spectral information facilities vegetables,firstly,using the principal component analysis to reduce the dimensionality of high-dimensional spectral data;secondly,selecting the first three principal components PC1-PC3 as the new input variables;finally,based on the discriminant results of the diseased leaves and healthy leaves as the output,the classification and identification and early warning model of facility vegetable diseases based on SVM was constructed.When the radial basis kernel function was selected,the classification recognition model had the highest classification accuracy of the leaves of healthy leaves and powders,100%and 96.25%respectively,using confusion matrix to evaluate the model,its overall recognition accuracy was 98.125%,can realize the recognition of infected leaves and healthy leaves.(5)Taking cucumber powdery mildew as an example,an example of plant disease warning was carried out.It turned out that this study had a good effect for facility vegetable disease early warning,early warning research provided a new reference for agriculture.