Study On Early Warning Of Grape Downy Mildew Based On Gas Sensing Technology

Downy mildew is a common disease in grape production.In the grape orchard,once downy mildew occurs in the plant,it is easy to expand and spread,and difficult to control,thus affecting the field yield and fruit quality.Although grapes suffered from downy mildew,the plant will show the corresponding symptoms,but if only rely on the experience of producers or through the naked eye to identify and judge,there will be significant hysteresis,that is,plant growth and development has been more serious damage.Therefore,it is of great significance to actively explore some effective ways to effectively monitor the occurrence of grape downy mildew and timely make early warning.Compared with healthy leaves,the contents of volatile gases in the leaves infected by pathogens are different because of the changes of physiological metabolism and ultrastructure of the leaves.Based on this,this study combined with grape downy mildew incidence law and dynamic analysis of physiological and biochemical indicators,to use gas sensor technology（electronic nose technology）to monitor the occurrence of grape downy mildew and early warning analysis,to provide technical support for the scientific prevention of downy mildew in grape production.In this study,’Xiangyue’grape was used as experimental material,3×10²·m L^-1and3×10⁴·m L^-1concentrations of spores suspension inoculation,and healthy growth of grape leaves as a control.After inoculation,two studies shall be carried out simultaneously:（1）Dynamic analysis of physiological and biochemical indexes after grape leaves were inoculated with downy mildew;and（2）Research on early warning of grape downy mildew based on electronic nose detection technology.The main findings are as follows:1.Dynamic analysis of physiological and biochemical indexes of grape leaves inoculated with Downy Mildew bacteria.Peroxidase（POD）activity,superoxide dismutase（SOD）activity,soluble protein content,soluble sugar content and plasma membrane permeability indexes were measured and analyzed during the experiment.The results showed that on the first day after inoculation,the changes of each index were not obvious.On the 4th day after inoculation,POD and SOD activity indexes of leaves treated with 3×10²·m L^-1and 3×10⁴·m L^-1spore suspension were significantly higher than those of the control（P<0.05）,while soluble protein and soluble sugar contents of leaves were increased.However,the content of soluble protein and soluble sugar in leaves of3×10²·m L^-1inoculation group was not significantly different from that of the control group（P>0.05）.On the 7th day after inoculation,POD activity and SOD activity in 3×10²·m L^-1and3×10⁴·m L^-1spore suspension treatment groups decreased compared with the 4th day,but the two treatments were still significantly higher than the control group,while the contents of soluble protein and soluble sugar in leaves of inoculation group were relatively stable.On the 10th day after inoculation,POD activity,SOD activity,soluble protein content and soluble sugar content of leaves in both inoculation groups were decreased compared with that on the 7th day,among which POD and SOD activity decreased significantly.The change trend of plasma membrane permeability of leaves showed that from the 4th day after inoculation,the plasma membrane permeability of leaves of the two treatment groups inoculated with bacteria continued to increase,and both groups were significantly higher than the control.In conclusion,grape leaves after inoculation of pathogen,its physiological metabolism produced great changes,and the change trend of physiological and biochemical indexes is complex,so only by physiological and biochemical index of the test is difficult to accurately determine whether grape leaf disease and disease early warning,and physiological and biochemical index of testing workload is big,is not convenient to practice.2.Study on early warning of grape downy mildew based on electronic nose detection technology.Grape varieties were inoculated with spore suspension at 1d,4d,7d and 10d,and the occurrence of downy mildew in leaves of 3×10²·m L^-1and 3×10⁴·m L^-1were investigated in control group and control group respectively.The investigation results of downy mildew in grape leaves showed that there was no obvious morphological change in the early stage of downy mildew,so the artificial observation method could not be used to detect the disease effectively in the early stage.In order to explore an effective method for early warning of grape downy mildew,the odor information of healthy grape leaves and leaves inoculated with different concentrations of downy mildew spores(3×10²·m L^-1and 3×10⁴·m L^-1)was collected dynamically（1d,4d,7d,10d）by using the gas sensor（electronic nose）test system.Then,the mean value and maximum value of odor response data were extracted by MATLAB software as the characteristic quantity,and random forest（RF）,support vector machine（SVM）and BP neural network pattern algorithm were used to identify and analyze the downy mildew infection of grape leaves.After selecting a better signal eigenvalue（mean value）,the mean value is used for principal component analysis（PCA）and Fisher linear discriminant analysis（LDA）.However,the two linear pattern algorithms（PCA and LDA）were not satisfactory.Even on the7th day after inoculation,there was still a large overlap between healthy leaves and leaves with3×10²·m L^-1and 3×10⁴·m L^-1downy mildew pathogen concentration.It is difficult to make an accurate distinction.And data extraction based on electronic nose response mean value as the characteristic,adopts the model of three kinds of nonlinear algorithm（RF,SVM,BPNN）identification results show that compared with the linear algorithm（PCA,LDA）,the nonlinear model algorithm have higher recognition accuracy,and is also an advantage in achieve accurate identification of time,the SVM recognition effect is best,From the 4th day after inoculation,the grape healthy leaves and downy mildew infected leaves could be accurately distinguished,with an accuracy of 94.67%.After that,the identification accuracy was higher.In conclusion:The electronic nose detection technology combined with SVM mode algorithm can not only accurately identify grape healthy leaves and downy mildew infected leaves,but also has the advantages of fast,non-destructive and efficient compared with physical and chemical test detection methods.This study verified the feasibility of electronic nose detection technology in the early warning of grape downy mildew,so as to provide scientific basis for disease control in grape production.