Dynamic modeling and improved neural network compensation sliding mode control of the rotating flexible load servo drive system with tip payload

Flexible manipulators are extensively utilized across various sectors because of their superior ability to adapt to intricate operational conditions. However, the tip payload of the flexible manipulator varies under different operating conditions, leading to system resonance during rotational motion and subsequently reducing operating accuracy and stability. This paper investigates the rotating flexible load servo drive system with tip payload (RFLSDS-TP) to address these issues. The dynamic model is developed using the assumed mode method (AMM) and Lagrange’s principle, incorporating flexible structure deformation and external friction torques. Furthermore, an improved sliding mode control (SMC) strategy is proposed to enhance system performance. This method integrates neural network compensation, leveraging the nonlinear approximation capability of radial basis function (RBF) neural networks to handle system uncertainties effectively. In addition, a saturation function is used in the control law design to reduce jitter, thereby improving tracking accuracy. Simulation results demonstrate that the proposed control strategy can effectively compensate for external disturbances in real time and significantly reduce tracking errors. Experimental validation on a physical prototype further confirms its effectiveness: compared with conventional PID and SMC approaches, the proposed method reduces the acceleration of the flexible manipulator by 61.1% and 38.3%, respectively. These findings highlight the substantial advantage of the proposed strategy in vibration suppression and precise motion control.