Study On Prediction Model And Algorithm Of Grassland Environmental Pollutants Based On IGWO-RBF

China has a vast territory.Grassland,as an important land resource in the north,is of great significance to agricultural development and ecological environment protection,the grassland plays an irreplaceable role in human production activities and the improvement of the natural environment.In recent years,with the mushrooming of population and the high speed developmen of economy,the atmospheric environment in the grassland area has been seriously polluted,which has seriously damaged the ecological environment of the grassland.This paper will use neural network to predict the concentration of pollutants,but the research of pollutant prediction is mainly aimed at nonlinear data.In this context,the study on the learning parameters of RBF neural network(Radial Basis Functionin)grassland environmental pollutant prediction was carried out,aiming to improve the generalization ability and accuracy of air pollutant prediction,and to provide scientific guidance for pollutant treatment.The research content will be divided into the following:First of all,this paper will start from the research on pollutant prediction,introduces the progress of pollutant prediction research at home and abroad,describes the steps of the prediction model and the evaluation criteria of the performance comparison of the prediction model,and then introduces the network structure,basic learning algorithm and advantages and disadvantages of the RBF neural network algorithm in detail.Because the traditional RBF neural network has the problem of insufficient parameters,the grey Wolf swarm algorithm is introduced to optimize the RBF neural network to solve the problem of insufficient learning encountered in nonlinear prediction research.Then introduces the basic working principle of GWO algorithm,as well as the optimization process of the algorithm,advantages and disadvantages.Secondly,the problem of insufficient parameters of RBF neural network is analyzed,and the improved algorithm is used to adjust the network parameters of RBF.The nearest neighbor clustering algorithm is used to further adjust the central parameters of the RBF network to reduce the dependence on the initial value of the clustering center.Then,the nonlinear adjustment of convergence factor and position update adjustment with weight coefficient are used to improve the grey Wolf colony algorithm,so as to improve the development and exploration ability of GWO algorithm.The Improved grey Wolf colony algorithm IGWO(Improved GWO)is used to adjust the weight parameters of RBF network.To this end,a combined neural network model(IGWO-RBF network model)is designed,and then the function is divided into blocks.Then,the design of IGWO-RBF prediction algorithm is introduced in detail,and the process of establishing the prediction model is explored.Finally,the IGWO-RBF algorithm is used to design the pollutant prediction model.Using the historical data of urban environmental monitoring stations in Sahantala Grassland of Baotou City as experimental data,the model was trained according to the three modules of IGWO-RBF algorithm for pollutant prediction model research,and the result of experimental study is analyzed.Finally,through comparative experiments,prediction is made from different model algorithms under the same experimental data and environment,and error comparison is made on the predicted results.It is concluded that the performance of pollutant prediction model of IGWO-RBF algorithm is better than that of RBF algorithm and GWO-RBF algorithm.Therefore,IGWO-RBF algorithm is more suitable for nonlinear prediction of grassland pollutants.