This research presents an algorithm for modeling and controlling a trajectory tracking problem for a 6-dof (degree of freedom) parallel platform using inverse differential modeling aided by neural networks. The algorithm addresses the issues of model uncertainty and adaptation to changes and external disturbances within the surrounding environment. By employing the algorithm of Particle Swarm Optimization (PSO), the initial weight values of the neural network are determined, along with the optimal network structure required for estimating the differential model. This is done to achieve the best network performance with minimal training time. Subsequently, the network is trained using a comprehensive training dataset that covers the entire workspace of the platform, without having to know the geometric configuration or mathematical model of the robotic manipulator. The trained network is then utilized within the control loop to drive the parallel platform in a reference trajectory tracking problem. The acquired data during platform operation is used to retrain the neural network, enabling it to adapt to changes occurring within the system. The simulation results demonstrated the effectiveness of the proposed algorithm in the problem of reference path tracking by enhancing performance and adapting to changes occurring in the overall platform. The proposed model was able to reduce tracking error by an average of 92.2%, assuming differences in the length of one of the platform’s arms, as well as an average reduction of 27.3% assuming the presence of mechanical backlash within the platform actuators. Furthermore, the proposed algorithm exhibited robustness against noise and external disturbances.