SYSTEM FOR PREDICTING TRAIN MOVEMENTS AND REDUCING DELAYS BASED ON ARTIFICIAL INTELLIGENCE IN RAILWAY TRANSPORT LOGISTICS.

Abstract

This article investigates the application of artificial intelligence and machine learning technologies for predicting train delays in the Uzbekistan railway transport system. The study comparatively analyzes Random Forest, Gradient Boosting, and LSTM neural network models. The experimental data includes train movement statistics for the Tashkent—Samarkand—Bukhara route during 2022—2024, weather data, and infrastructure condition indicators. Results demonstrate that the Gradient Boosting model can predict delays with 87.3% accuracy, showing 23% higher efficiency compared to traditional statistical methods. The research findings can be applied within the framework of the “Digital Uzbekistan – 2030” strategy for railway system digitalization.