Examples of typhoon landing data.

Typhoons are natural disasters characterized by their high frequency of occurrence and significant impact, often leading to secondary disasters. In this study, we propose a prediction model for the trend of typhoon disasters. Utilizing neural networks, we calculate the forgetting gate, update gate, and output gate to forecast typhoon intensity, position, and disaster trends. By employing the concept of big data, we collected typhoon data using Python technology and verified the model’s performance. Overall, the model exhibited a good fit, particularly for strong tropical storms. However, improvements are needed to enhance the forecasting accuracy for tropical depressions, typhoons, and strong typhoons. The model demonstrated a small average error in predicting the latitude and longitude of the typhoon’s center position, and the predicted path closely aligned with the actual trajectory.

台风是一类发生频次高、影响显著的自然灾害，常引发次生灾害。本研究提出了一种台风灾害趋势预测模型。本研究借助神经网络（neural networks），计算遗忘门（forgetting gate）、更新门（update gate）与输出门（output gate），以此实现台风强度、位置及灾害趋势的预测。本研究依托大数据（big data）理念，通过Python技术采集台风数据，并对模型性能进行了验证。整体而言，该模型拟合效果良好，针对强热带风暴的预测尤为出色。但针对热带低压、台风及强台风的预测精度仍有待提升。该模型在预测台风中心位置的经纬度时平均误差较小，预测路径与实际轨迹贴合度较高。