Aiming to address the issues of complicated operation and insufficient control accuracy in current micro-nanopositioning systems, a model-free adaptive predictive control scheme based on a semi-physical real-time simulation control platform is proposed. To overcome the limitations of poor adaptability in traditional experimental platforms, a control platform with real-time data exchange functionality is developed by integrating a real-time simulation environment with a data acquisition toolbox and utilizing equipment such as a laser range finder and a voltage amplifier. This significantly enhances the versatility and flexibility of the experimental system. In terms of control, to overcome the limitation that traditional offline models fail to fully account for the influence of dynamic nonlinearity, a model-free adaptive control scheme is proposed, and predictive control technology is introduced. This approach effectively addresses the frequency-dependent hysteresis nonlinearity problem, enabling real-time trajectory tracking. Experimental results show that the model-free adaptive predictive control algorithm can quickly and accurately track a given displacement, with a settling time of less than 1.5ms, and the tracking accuracy remains within 1.3μm.