Study On The Diagnosis Of Soybean Diseases And Pests In Cold Region

The contradiction between the increasing demand for agricultural knowledge and the scarcity of agricultural experts is one of the main contradictions in China’s rural areas.In order to solve this contradiction,it is important to realize the intellectualization of agriculture.In China,in many soybean planting regions in China,the existence of pests and diseases often caused direct economic losses of more than 10%,the individual area will reach more than 30%,and the yield and quality of various forms has greatly restricted China’s soybean production.Artificial intelligence technology has been widely used in the field of disease diagnosis.The application of artificial neural network in the diagnosis of crop diseases and insect pests has become a popular trend.This paper intends to soybean diseases and pests accurately determine the selection of the fuzzy neural network model is established,and automatically generate and adjust the membership function into the analysis method of AHP,to explore the feasible way of combining fuzzy neural network and AHP layer analysis method for pest diagnosis.The simulation results show that using fuzzy neural network analysis and AHP phase model for soybean pest diagnosis has the advantages of high speed,high diagnostic accuracy and good generalization ability,it is a good choice.Specific contents are as follows:Firstly,choose the North China representative of 7 kinds of pests as output borer.182 soybean pest samples were diagnosed according to the 8 kinds of characters,such as harm way and harm symptom,136 soybean pest samples were selected as the training set,and 46 samples were used as the prediction set.Based on the collection and analysis of soybean diseases and insect pests,The input vector is used to digitize and the input is analyzed by AHP,The data processed by two methods are used as input vectors of neural network.Secondly,three kinds of neural network models for network training and simulation are established.This paper analyzes the influence of the number of hidden layer nodes in the BP neural network,the training target,the learning rate and the training times on the performance of the network;Effect of density parameters of radial basis that RBF neural network in the training result;and the response of the fuzzy neural network hidden layer nodes and the number of training parameters of model.Finally,the accuracy rate of three kinds of neural network for the diagnosis of soybean diseases and pests was compared.The results show that the fuzzy neural network has the best effect on soybean diseases and insect pests.The experimental results show that the method of analytic hierarchy process on the input processing and modeling of fuzzy neural network prediction,in 46 samples,a total of 44 samples were predicted,the recognition rate as high as 95%,proved that the method to distinguish the soybean pest is feasible.