Prediction of Traction Energy Consumption for Urban Rail Transit Trains in Relative Speed Mode

［Objective］It is aimed to accurately predict the traction energy consumption of urban rail transit trains operating in relative speed mode using support vector machine（SVM）regression and genetic algorithms, ultimately enhancing energy efficiency during train operation. ［Method］First, the dynamics characteristics of urban rail transit train traction operation are analyzed to obtain dynamics physical indicators that reflect the real-time operational state of the trains. It then models the relative speed and position changes between trains to establish a train operation model based on relative speed mode. On this basis, key train operational indicators that directly influence train traction energy consumption are extracted, and SVM regression combined with genetic algorithm is employed to analyze these indicators, enabling precise train traction energy consumption predictions. ［Result & Conclusion］The experimental results demonstrate that the proposed method effectively predicts train traction energy consumption. The prediction accuracy ranges from 92.0% to 99.6%, with a maximum relative error of 2.36%, an average relative error of 1.75%, and a root mean square relative error of 1.52%, outperforming other prediction methods by every indicator value. The prediction results of the design method show minimal fluctuation throughout the entire prediction range, indicating excellent overall prediction stability and strong practical applicability.