Energy-Efficient Train Control Incorporating Realistic Traction Characteristics Using Convex Optimization

The study proposes an energy-efficient train control (EETC) model that utilizes the convex optimization (CO) method to incorporate more practical and complex traction system characteristics. These characteristics, referred to as RMeB-RP, include regenerative and mechanical braking at low speeds and reduced power at relatively high speeds. In many previous studies on the EETC problems, the traction system characteristics often incorporated the simplified pure regenerative braking and constant power characteristics during high speeds, known as RB-CP, to reduce the complexity of the model. We applied convexification and linearization to model such nonlinear characteristics during traction and braking operation over various speeds and applied convex optimization to achieve the optimal solution efficiently. Our simulation results showed that the train with RMeB-RP has a higher total energy consumption, cruising speeds, and speeds at the beginning of the braking stage than the train with RB-CP. Furthermore, the train with RB-CP in the EETC model often leads to underestimating energy consumption since the partial loss of traction and regenerative braking effort. As for the real-world scenario in this model incorporating RB-CP, energy consumption was underestimated by 17% compared to incorporating RMeB-RP, with a difference of 0.9136 kWh/km in energy consumption per kilometer. In addition, applying CO to quickly address the EETC problems incorporating the complex traction system characteristics has shown to be effective in the real case, in which the total computational time of interstation operation by using the CO model is only within the range of [0.52, 1.2] seconds.