Research On Intelligent Control For Vibration Of Multiple Flexible-body System

The functional requirements of aerospace make the structure large and complex.There are coupling connections in many flexible intelligent structures,which is easy to produce residual vibration without effective control,and the operation accuracy will be reduced,which even leads to more serious consequences.Therefore,it is necessary to study the vibration characteristics and active control of multiple flexible-body system.In this paper,multiple flexible beams are taken as the research object,and the elastic coupling structure of multiple flexible beams and rigid-flexible coupling structure of multiple flexible beams are studied.The work is summarized as follows:The finite element model of flexible multi-body system with elastic coupling and rigid flexible coupling is established,and the structural characteristics are analyzed to provide guidance for controller design.For the elastic coupling structure of multiple flexible beams,it is simplified to some extent,and the structure has strong coupling,which is characterized by dense frequency.Then the modal model is truncated and identified;For the rigid flexible coupling structure with multiple flexible beams,the structure has weak coupling,close frequency characteristics but not obvious,and the vibration is affected by gravity.In order to suppress the vibration of flexible multi-body system quickly,a neural network controller and reinforcement learning controllers are designed and simulated for the elastic coupling structure of multiple flexible beams.The neural network controller uses pre-learning and online updating of network weights;The reinforcement learning controller has good robustness.Using off-line learning and parameter transfer,fuzzy reinforcement learning(FRL)and proximal policy optimization(PPO)are designed.For multiple flexible beams rigid flexible coupling structure,the vibration suppression of trapezoidal trajectory motion and fixed-point disturbance is studied.According to the characteristics of multiple flexible beams rigid-flexible coupling structure motion process,combined with high-order linear extended observer,a fuzzy auto disturbances rejection controller is designed.Two kinds of multiple flexible system experimental platforms,elastic coupling and rigidflexible coupling,are built.The interactive interface of the experimental platform is developed based on C + + language.Experiments are carried out to verify the effectiveness of the algorithm control.For multiple flexible beams elastic coupling structure,the free vibration and frequency characteristics under different excitation forms are analyzed;The experimental results of the neural network controller and the reinforcement learning controller show that for the proportional differential control,the large amplitude value decays faster,but the small amplitude control is slower.The designed intelligent controller can suppress the small amplitude vibration more quickly,and the overall control effect is better.For the rigid flexible coupling structure of multiple flexible beams,the vibration of each beam under trapezoidal trajectory and fixed-point disturbance is studied and the control experiment is carried out.The experiment shows that fuzzy auto disturbances rejection control(FADRC)can achieve better control effect than linear ADRC,which reflects the nonlinear characteristics of fuzzy control.