Vehicle platoon reinforcement learning model predictive control method with state-aware switching mechanism under dos attack
By optimizing MPC parameters and buffer mechanisms through reinforcement learning, the adaptiveness and coordination problems of vehicle queuing systems under DoS attacks in traditional methods are solved, and stable car-following control under communication interruption conditions is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHONGQING UNIV OF POSTS & TELECOMM
- Filing Date
- 2026-05-18
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional model predictive control methods lack parameter adaptation capabilities under DoS attacks, and vehicles cannot adjust optimization strategies in real time, leading to instability of the queuing system when communication is interrupted. Furthermore, reinforcement learning methods struggle to handle coupling constraints between vehicles and maintain coordination in complex environments.
Reinforcement learning is introduced to optimize MPC parameters. Combined with onboard sensor information and a buffer mechanism, the optimal state trajectory under attack-free conditions is stored. The optimization solution is maintained during the attack period. The optimization direction is guided by the assumed trajectory to ensure that the vehicle maintains car-following consistency under conditions of missing information.
It enhances the system's adaptability under multiple operating conditions, maintains the collaborative tracking function during attacks, mitigates performance degradation caused by missing information, and ensures rapid recovery of the collaborative state.
Smart Images
Figure CN122411033A_ABST