Prediction And Analysis Platform Of Invasive Insects Based On SOM Neural Network

China is a large agricultural country,invasive insects have seriously restricted the bumper harvest of crops,caused huge losses to the agricultural economy,and had a significant impact on the national economy.The prediction of invasive insects makes the prevention and control of invasive insects purposeful and planned,which is not only an important part of invasive insect management,but also the basis for effective prevention and control of the occurrence of invasive insects.However,the accuracy,effectiveness and practicability of prediction of the invasive insects have always been the bottleneck restricting the effective control of agricultural pests in China.SOM is a neural network model which can extract information from environmental data and map it into fewer dimensions,and can analyze complex non-linear data.Strong clustering ability and prediction analysis ability make the prediction results more accurate and reliable SOM neural network was selected to analyze the environmental characteristics of invasive insects and build a prediction model in this study R,a free and open source software that can build prediction SOM model and spatial visualization analysis,was used to complete the construction of real-time online visual invasion insect prediction and prediction platform in this study.Based on the research of SOM neural network and R,this study constructed a shinyweb application program with friendly interface,flexible and online real-time interaction through Rshiny,which realized the functions of preprocessing of invasive insect data,construction of prediction model,visual analysis of space and results and download.The results of the platform displayed through interactive text,tables and visual graphics,and all the results could be downloaded and saved according to the needs.This platform chose the free cloud server provided by shinyapps.io deployed in R,which is open to all researchers.In order to test the practicability of the application of the platform,a field suvry data for Spodoptera litura was taken as an example to test the prediction platform,The data was collect in Baoshan city,Yunnan Province from 2020 to 2022,inculduing Spodoptera litura abundance and local meteorological data.The results showed that the real-time online visual invasion insect prediction and prediction platform can effectively analyze complex nonlinear data,solve the problem of dimensionality reduction of high-dimensional data,extract effective information of data and cluster,construct SOM prediction model and show the predicted number of insects based on meteorological data through visual analysis.This study results could be a solution and suggestion for the agricultural production management and decision-making for the invasion insect research.