A parameter-optimized reinforcement iterative learning control method for enhancing passenger pressure comfort and air quality in high-speed trains traversing extreme tunnels

When high-speed trains traverse tunnels with characteristics such as high altitude, steep gradients, and substantial lengths, intense tunnel pressure waves are generated. These pressure waves induce pressure fluctuations within the carriage, adversely affecting passenger pressure comfort. However, when trains pass through extra-long tunnels or tunnel groups, maintaining a balance between passenger pressure comfort and air quality becomes challenging. This is because the existing passive control strategy—closing ventilation ducts for a certain duration—fails to match the tunnel pressure wave characteristics. Therefore, an iterative learning control (ILC) algorithm that matches the tunnel pressure waves is employed to suppress interior pressure fluctuations. However, the learning parameters in existing ILC algorithms for interior pressure control are fixed and cannot adapt based on control performance, thereby lacking adaptive tuning capabilities. To tackle these issues, this paper integrates reinforcement learning (RL) into the ILC framework and proposes a parameter-optimized reinforcement iterative learning control (PORILC) scheme for managing interior pressure fluctuations within the vehicle. Specifically, the RL-based adaptive gain adjustment scheme is developed. A performance-related weighting factor is introduced to combine the decreasing gain of ILC with the powerful exploration capability of RL. Finally, simulation results demonstrate that, compared with traditional passive control and conventional ILC methods, the proposed PORILC scheme ensures both passenger pressure comfort and air quality within the carriage, with enhanced tracking accuracy. Experimental results suggest that the PORILC scheme contributes to improved pressure comfort for vehicle occupants under extreme tunnel pressure wave excitation.