Research On Grey Neural Network Prediction Model

The grey neural network prediction model is a hot topic in the research of combined forecasting models in recent years.This paper studies the combined forecasting model and constructs a new model for single variable and multivariate small data sample predictions.The main results are as follows:(1)For the grey neural network combination model,the number of modeling data samples cannot be determined specifically.This paper defines the comprehensive prediction accuracy and proposes an model named SGMBP(1,1),which based on the comprehensive prediction accuracy of the optimal modeling sample.The sequence is set to be dynamically variable,and the optimal prediction sample number is determined by integrating the principle of maximum prediction accuracy.The model was applied to the prediction of soil moisture content in Xinxiang City,Henan Province.The results show that the prediction accuracy of the model is relatively high and the number of modeling data of the grey neural network combined forecasting model is solved.(2)For the sample data with large data volatility,and which have missing some data.An non-equidistant sequence GMP(1,1,N)model with predictable non-homogeneous exponential law sequence is proposed.At the same time,in order to reflect the impact of missing data on the modeling results.The non-equidistant sequence and the GMP(1,1,N)model's prediction results are taken as the input of the BP neural network at the same time,then a neural network is constructed.The model was applied in soil moisture prediction in Dengzhou City,Henan Province.The results show that the combined model improved the prediction accuracy of GMP(1,1,N)model.(3)Based on the unpredictability of catastrophic problems,construct a multivariable soil moisture prediction model.In order to better reflect the influence of related factors on system factors,three-parameter interval grey numbers are used for statistical data samples.Based on the advantages of gray GM(0,N)model and BP neural network model,the BPSGM(0,N)model is constructed.The model was applied to the prediction of soil moisture content in Xinzheng City,Henan Province.The results show that the BPSGM(0,N)model had a good prediction effect on soil moisture content.Based on the characteristics of measured data in soil moisture prediction problems,this paper proposes three different forecasting methods.The results show that the model has higher prediction accuracy,and it can provide useful reference for regional water-saving agriculture decision-making.