Research On Global Average Annual Temperature Prediction Based On Genetic Neural Network

According to the latest data,since the beginning of the 20th century,the global average annual temperature has increased by about 1.4?,and the greenhouse effect is increasingly intensified.From the perspective of climate phenomenon,since the 20th century,the frequency of extreme weather(high temperature,cold wave,rainstorm,etc.)around the world has shown an increasing trend.Especially in recent years,extreme high temperature weather in summer has seriously affected people's production,life and health.Therefore,as one of the important indicators to study global warming,global average annual temperature has important research value.In this paper,we use the traditional time series(ARMA)model,BP neural network model and GRU neural network improved by genetic algorithm to forecast the global average annual temperature,to warn people to pay more attention to the global warming problem by the result.Also,by comparing the prediction result of traditional time series analysis model(ARMA),BP neural network model and GRU neural network improved by genetic algorithm and on the basis of the research and exploration on the data pretreatment,model structure,model parameter adjustment and so on,we look for better methods of model forecasting and provide a reasonable method for the study of similar time series data.Through the experimental comparison,the neural network improved by genetic algorithm shows good rationality and prediction results,and then we use the model to predict the future global average temperature.Finally,we study the main influencing factors and its effects of global warming by linear regression.Firstly,we construct a linear model between co2 concentration and the global average annual temperature,which shows a strong positive correlation between them.Then,we study the linear relationship between sea ice area,extreme weather frequency and global average annual temperature on the basis of the previous prediction of the future temperature to show the trend of the sea ice area and the frequency of extreme weather in the future so as to reflecting the influence of the greenhouse effect more directly.