Global Temperature Prediction Study Based On AMIMA, Grey Prediction and Neural Network Models

Abstract

Global warming has profoundly affected the survival and development of human beings and has become a worldwide problem, and the mitigation of global warming requires a more accurate prediction of global temperature development. In order to accurately predict global temperatures in future years and provide feasible solutions, this paper comprehensively compares the prediction results of three prediction models. Firstly, a time series model is chosen to map the past time series to describe the global temperature changes. In addition, this paper selects three models for analysis by comparing a time series ARIMA model, a grey prediction model and a neural network model, fits the model using data information on the global average temperature for 200 consecutive years, and trains and corrects the model to obtain the projected global temperature for the next 80 years. Considering the advantages and disadvantages of each model and combining the conclusions of previous studies, this paper concludes that the neural network model predicts the most accurate results. Meanwhile, the results indicate that in the future, if the continuous increase in global temperature is not controlled, global warming will continue to cause irreversible damage to the Earth's ecosystem, which in turn will have a negative impact on human health.