Research on a Machine Learning-Based Prediction Model for Tear Gas Smoke Dispersion

In this study, a machine learning-based prediction model for tear smoke dispersion was constructed, incorporating Random Forest, Gradient Boosting Tree and LSTM networks. Detailed meteorological, geographical and topographical data were collected, and data cleaning and feature engineering were performed. The model was found to outperform traditional models in terms of prediction accuracy, computational efficiency and adaptability, and to show excellent performance in different time scales, spatial scales and environmental conditions.