The super rail-guided train (SRT) with four modules and six axles is one type of virtual track train (VTT). It has high carrying capacity and excellent curve passing flexibility. The core issue is achieving high path-tracking accuracy and low hinge forces simultaneously. This paper proposes a new control strategy. Firstly, the reconfigurable dynamics model of SRT is established by a two-step method, and the hinge force is quantified by vehicle state and control inputs. The control strategy adopts a layered framework. The upper layer calculates the generalized force at the centre of gravity (CG) of each module needed for tracking based on the improved model predictive control (MPC). The lower layer reallocates the CG forces, uses the ‘virtual axle’ method, and then distributes the wheel steering angle and driving torque according to the different actuator configurations of each module. The simulation results show that the proposed strategy can achieve high tracking accuracy and low hinge force and maintain robustness under different velocities and curve radii without increasing the computational press. Moreover, the control strategy can be applied to virtual track train (VTT) with different structures.