To address the challenge of fault-tolerant control in steer-by-wire systems for four-wheel steering vehicles, We introduce a model-free adaptive fault-tolerant control method based on reinforcement learning. A dual-freedom dynamic framework for the four-wheel-steering vehicle was developed. The correlation between inputs and outputs in model-free adaptive control is established through comprehensive examination of the vehicle’s dynamic behavior model. Subsequently, the vehicle dynamic model is dynamically linearised, and reinforcement learning is integrated to achieve autonomous learning and optimisation of the MFAC parameters. Consequently, an RL-MFAC fault-tolerant controller is designed. Matlab/Simulink serves as a final validation tool for simulations. The simulation results show that in the event of a fault in the steer-by-wire system, This method has been demonstrated to be an effective means of enhancing the vehicle’s fault-tolerant performance in the event of sensor faults.