Prediction of global temperature based on the BP neural networks

Abstract

The problem of global warming is getting worse as global temperature has risen year by year to an alarming level. This paper is first based on historical data on global temperatures, two prediction models were developed using Gray Prediction Method and BP Neural Network model, whose results show that the root mean square error test value of BP neural network is about 0.03 lower than that of the gray prediction method (0.1627).Thus, the BP neural network prediction model is proved effective, and then the model was analyzed by gray correlation theory, which found that carbon dioxide concentration, forest cover area and human population are affecting the global temperature, and the Gray correlation of the fisrt two factors is as high as 0.8, which makes the two the main factors. In the end the paper gave measures to contain global warming.