Research On Maize Yield Prediction Model Based On Machine Learning

The prediction of maize yield is related to national grain reserve,grain regulation,farmland management and agricultural decision-making,and is an important reference basis for grain security and grain policy making,as well as a key factor for regulating agricultural planting and agricultural management.In recent years,with the continuous development and improvement of artificial intelligence technology,the use of machine learning method,based on meteorological data to predict corn yield has gradually made new progress.In view of the limitation of traditional statistical methods to predict corn yield,this paper uses machine learning algorithm to predict corn yield,in order to improve the accuracy of prediction,so as to promote the progress of using meteorological data to predict crop yield again.This paper is based on 98 districts and counties under the jurisdiction of Jilin and Shanxi provinces and the corresponding satellite meteorological data as the research object.A total of 15years' data were collected from 2004 to 2018.After data preprocessing,significance verification and other steps,the maize yield prediction model based on machine learning was studied.In this paper,strong correlation factors were selected by Pearson correlation analysis,including correlation analysis between meteorological and maize yield,correlation analysis between meteorological and agricultural information,correlation analysis between agricultural information and maize yield,etc.On the traditional machine learning based on random forest,of two regression algorithm of support vector machine(SVM)model,the results show that the model fitting and forecast ability effect is good,then again deep learning of the neural network model is put forward in this paper,three time dimension respectively meteorological data joint training a network model,the specific data for days,months and years data to implement the maize yield prediction.After parameter adjustment,the prediction effect of each model is evaluated and analyzed by performance indexes such as mean square error,mean absolute percentage error and root mean square error.The results show that the neural network model based on multiple time dimensions has the best effect.It shows that the machine learning algorithm is feasible to predict maize yield under meteorological conditions,especially the neural network model with multiple time dimensions can effectively improve the prediction accuracy.