Suspension Control At Both Ends Of Wind Turbine Nacelle Based On RBFNN Decoupling

This paper take the wind magnetic levitation yaw system as the research object,focusing on the suspension control part of the wind magnetic levitation nacelle.Due to the harsh suspension condition of the nacelle and the time-varying wind direction,the nacelle is prone to pitch,roll,axial oscillation,which seriously affects the operation safety of the wind turbine.How to improve the axial suspension stability of the nacelle and effectively suppress the pitch of the nacelle is the key to the suspension stability of the nacelle,for this reason,a suspension model with axial and pitching two degrees of freedom motion at both ends of the nacelle is constructed,and the influence of mechanical coupling,electromagnetic force coupling and structural parameter mismatch between the suspension systems at both ends of the wind engine room on the suspension stability of the engine room is analyzed in depth,based on the infinite approximation ability of Radial basis function neural network(RBFNN)to the uncertain system,combined with the model reference adaptive control concept,the suspension control at both ends of the wind turbine nacelle based on RBFNN decoupling is proposed to improve the performance of suspension tracking,interference suppression and suspension synchronization performance at both ends of the nacelle.Firstly,the RBFNN adaptive direct decoupling control of the wind turbine nacelle suspension system is proposed,and the cross-coupling terms,axial disturbance terms and pitch disturbance terms in the suspension models at both ends of the nacelle are attributed to external disturbances,with the help of RBFNN,the external disturbances are infinitely approximated.The RBFNN controller is designed based on the levitation air gap tracking error,and the main levitation tracking controller is designed using the state feedback method to achieve effective independent decoupling control of levitation tracking and levitation synchronization;Secondly the RBFNN adaptive decoupling control of the cabin levitation system based on model reference is proposed,combining model reference adaptive control with RBFNN,with the help of the strictly linear and non-coupling characteristics of the decoupling model and the infinite approximation capability of RBFNN,the RBFNN controller is designed based on the model deviation,so that the suspension model at both ends is infinitely approximated to the linear decoupling model,So as to achieve complete decoupling,the linear tracking controller takes the decoupled suspension system as a reference to complete the cabin suspension tracking control;Thirdly The RBFNN adaptive synchronization decoupling control of the cabin suspension system based on model reference is proposed to suspend both ends of the cabin The uncertain term in the model is divided into axial interference and synchronization interference.With the help of the strictly linear uncoupling decoupling model and the infinite approximation capability of RBFNN,the difference between the suspended air gaps at both ends of the nacelle is taken as the synchronization error,and the synchronization error and its derivative are introduced.Synchronization performance indicators,based on the deviation from the model,design the RBFNN adaptive synchronization controller,and cooperate with the RBFNN adaptive axial disturbance compensator and the levitation tracking controller to complete the cabin levitation control.The suspension simulation experiment platform at both ends of the wind turbine nacelle was built,and the suspension tracking,1000 N pitch disturbance suppression and 1000 N axial disturbance suppression experiments of the above three suspension decoupling control strategies were carried out respectively.The simulation results show that the suspension tracking start time for strategy 1 is 0.43 s,steady state error is 0.056 mm,overshoot value is 0.163 mm,pitch disturbance maximum drop value is 0.275 mm,regression time is 1.7s,axial disturbance maximum drop value 0.07mm;The Suspension tracking start time for strategy 2 is 0.2s,steady state error is 0.0099 mm and no overshoot,pitch disturbance maximum drop value is 0.044 mm,regression time is 0.2s,axial disturbance maximum drop value is 0.0194mm;The suspension tracking start time for strategy 3 is 0.38 s,the steady-state error is 0.009 mm,the pitch disturbance maximum drop value is 0.035 mm,the regression time is 0.25 s,and the axial disturbance maximum drop value is 0.0192mm;it can be seen that the suspension start time and impact Disturbance fall regression time for strategy 2 is the shortest,and the pitch disturbance suppression and axial disturbance suppression for strategy 3 are the best.With the help of RBFNN infinite approximation capability and strict linear uncoupling decoupling model,it effectively improves cabin levitation tracking,disturbance suppression and levitation synchronization at both ends performance.