مجلة الاربعين المحكمة
Effective crowd management is essential for maintaining public safety during large-scale events. Traditional approaches often prove inadequate in addressing real-time changes and the complex dynamics of high-density crowds. This challenge is particularly evident during the Arbaeen pilgrimage in Karbala, Iraq—one of the largest annual human gatherings—where millions of pilgrims converge within confined areas. The absence of predictive, data-driven systems to monitor and guide crowd movement increases the risk of congestion, delays, and potential casualties. This study introduces an intelligent framework that leverages machine learning and real-time analytics to enhance crowd management efficiency. An artificially generated dataset was constructed to simulate real-world conditions, including features such as visitor numbers, crowd pressure, average speed, environmental conditions, and emergency events. The system calculates a dynamic «risk degree» index, classifying areas into three behavior categories: Normal (73.7%), Suspicious (5.0%), and High Risk (21.2%)—enabling early detection and intervention. To support strategic planning, a linear regression model is employed to forecast trends in visitor numbers, pressure, and risk levels. While the model effectively captures overall patterns, it is limited in predicting abrupt fluctuations, underscoring the importance of real-time data monitoring. For optimal path planning, Dijkstra’s algorithm is applied with risk-weighted edge costs. The resulting smart route significantly outperforms the traditional path in both safety and efficiency, with a total cost of 11.18 compared to 85.70, and an average risk of 4.92 versus 8.26. An interactive web-based dashboard supports visualization and decision-making through real-time alerts, heatmaps, and exportable analytical tables.