Research On Adaptive Neural Network Anti-Sway Control Of Crane Based On Partial State Feedback

With the development of science and technology,cranes have become an indispensable part of modern industrial production because of their powerful material handling capabilities.The hoisting mechanism of the crane generally uses flexible cable as the traction mechanism.In the process of crane operation,the hook and payload will swing unavoidably due to the inertia force generated by the acceleration and deceleration of the crane.It will not only seriously reduce the efficiency of the crane,but also lead to payload collision,falling and crane metal structure damage and other safety accidents occur.For the problem of anti-sway control of double-pendulum crane,this paper discusses the double-pendulum crane system and control methods.The main results are as shown below:(1)Combining Lagrange energy equation of the second kind and the principle of virtual work,a planar double-pendulum nonlinear dynamics model of overcrane and gantry cranes was eatablished.Then the crane model was linearized at the payload equilibrium point using Taylor expansions.And the model was divided into two parts,the trolley subsystem and the hook-payload subsystem.(2)In the analysis of the crane nonlinear dynamics model,it is found that when the crane system is under the action of appropriate control forces,if the system is stable,the hook and the padyload will reach the equilibrium point simultaneously.A partial state feedback sliding mode controller without the feedback of the swing state information of the payload was proposed to address the problem that it is difficult to measure the swing state of the apyload in real time.The asymptotic stability of the trolley and hook states under the action of the controller was proved based on Lyapunov stability theory.Combined with the fact that the hook and the payload can be stabilized at the same moment,the asymptotic stability of the padyload was proved.Numerical simulation results show the feasibility of the controller and its robustness to changes in system parameters and external disturbances.(3)In order to reduce the maximum control force and the hook swing angle of the partial state feedback sliding mode controller and improve the performance of the controller,a neural network adaptive sliding mode controller with online updating of the weights was proposed,where radial basis neural networks are used to approximate the nonlinear dynamic terms in the system and adaptive laws are used to estimate the weight parameters.Then the asymptotic stability of the controller,sliding mode surface and system states was proved using Lyapunov stability theory and Barbalat’s Lemma.Numerical simulation results show the feasibility of the controller and its robustness to changes in system parameters and external disturbances.Simulation results show a great improvement in controller performance.(4)In order to reduce the trolley positioning time and solve the actuator saturation problem,an input-constrained neural network adaptive tracking controller that doesn’t require system parameters and the payload swing state was proposed.The crane system model is combined and transformed with the saturation module,and then the controller is designed to converge in a fixed-time according to the transformed model.The asymptotic stability of the sliding mode surface and system states was proved based on Lyapunov stability theory and La Salle invariance principle.Numerical simulations show the feasibility and superiority of the input-constrained neural network adaptive tracking controller and its robustness to changes in system parameters and external disturbances.