Research on Typhoon Classification and Path Prediction Based on Multi-model Fusion

Typhoons are prone to disasters such as gales, heavy rains and storm surges. Studies have shown that the formation of typhoons is linked to the meteorological conditions of tropical oceans, and the trajectories of typhoon movements are highly complex and nonlinear. The aim of this paper is to establish a fusion model based on multi-model classification and evaluation prediction, to analyze the relationship between the characteristic parameters of typhoons and factors such as temperature, air pressure, monsoon, etc., and to predict the path of typhoons and their impacts after landfall. In this paper, Spearman correlation is first used to analyze the relationship between the characteristic parameters of typhoons and factors such as temperature, pressure and monsoon. Then, the typhoons are clustered using K-means. Based on the classification results, a random forest algorithm is used to classify the typhoon indicators, and the data are brought in for training and typhoon path mapping. Subsequently, taking ‘Bebija’ as an example, this paper chooses the LSTM model to predict its route, and uses the dynamic time regularization algorithm to compare with the actual travel route.