Power quality solutions for rail transport using AI-based unified power quality conditioners

This paper proposes an artificial intelligence-based Unified Power Quality Conditioner (AI-UPQC) and a unique Dynamic K-factor PI (DKPI) algorithm for PI-UPQC to address power quality issues in electrified railway systems. The widespread use of power electronic converters in modern traction drives has exacerbated problems such as harmonics, voltage fluctuations, and resonance phenomena. The proposed AI-UPQC utilizes artificial neural networks (ANNs) to generate optimal reference signals for controlling series and shunt active power filters. A detailed 25 kV, 50 Hz traction power supply system model is developed in MATLAB/Simulink to evaluate the AI-UPQC's performance. Simulation results demonstrate that the AI-UPQC significantly outperforms conventional PI-controlled UPQCs in reducing voltage and current total harmonic distortion (THD), improving power factor, and providing faster response times under varying load conditions. The AI-UPQC reduced source current THD from 25.16 to 1.12% and load voltage THD from 6.62 to 2.07%. Sensitivity analysis further validates the robustness of the proposed system across different operating parameters. The AI-UPQC shows promise as an effective solution for enhancing power quality in modern electrified railway networks.