اميرحسين گل ابادي

Grade crossings are critical locations where railways and roads intersect. Accidents at these locations often result in fatalities and economic and social damages due to their impact on both road and rail users. Improving the safety of rail and road transportation requires the collection, classification, and analysis of accident data to better understand the status of grade crossings. In general, it is impossible to completely eliminate accidents at any critical point. Therefore, necessary measures must be taken to prevent and reduce the severity of accidents. This proposed research aims to identify the factors affecting the severity of accidents at rail-road crossings using data mining techniques and machine learning algorithms. For this purpose, data from the Federal Railroad Administration of the United States from 2012 to 2022 was used. Then, algorithms were developed to determine the various factors that increase accident severity at grade crossings. For this purpose, both unsupervised and supervised machine learning were used, including Prototype-K clustering and three models of random forest, K-nearest neighbors, and support vector machines for prediction. The accuracy of all three models was eva‎luated individually and in combination with clustering. The results show that the random forest algorithm provides the best accuracy among the different models examined.

داده كاوي , گذرگاه همسطح ريلي جاده اي , شدت تصادفات

level crossing , data mining , severity of accidents

Amirhosein Golabadi

Mohsen Pourseyedaghaii