Prediction and Analysis Model of Tunnel Evacuation Based on Random Forest Algorithm

This paper presents an evacuation time prediction model tailored for tunnel scenarios, utilizing the random forest algorithm. The significance of quickly and accurately predicting evacuation times is underscored, as it can greatly enhance safety measures during emergencies. The model was developed through extensive training of the random forest framework, achieving an impressive average Mean Absolute Percentage Error (MAPE) of approximately 7.8 % and a high <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> value of 0.92. This level of accuracy indicates the model's capability to effectively forecast evacuation times under varying conditions, specifically influenced by parameters such as the width and interval of evacuation passages. Furthermore, the analysis conducted using the random forest model identifies key factors influencing evacuation times within tunnels. The number of lanes in the tunnel emerged as the most significant contributor, with a contribution factor of 0.34, reflecting its critical role in facilitating the evacuation process. Other influential parameters include those related to the evacuation setup within the tunnel, such as the interval and width of the evacuation passageway, which have contribution factors of 0.26 and 0.25, respectively. Interestingly, the ratio of vehicle types present in the tunnel has a comparatively minor effect on the evacuation time, with a contribution of only 0.15. The findings of this study highlight the practical application of machine learning techniques, particularly the random forest algorithm, in enhancing evacuation strategies. By integrating crucial parameters that affect evacuation dynamics, the model provides a valuable tool for rapid predictions, thereby contributing to improved safety protocols in tunnel environments. Overall, this research not only sheds light on the mechanics of evacuation but also establishes a foundation for future investigations into optimizing evacuation strategies in various settings.