Research On The Inverted Pendulum System With Neural Networks Adaptive Control

Inverted pendulum system is widely used in aerospace and robotics,therefore the control of inverted pendulum is received more attentions.As the inverted pendulum system is a typical multivariable strongly coupled uncertain nonlinear system,it is difficult to establish accurate mathematical model,so neural network adaptive control is used to solve the dynamic system,which is difficult to establish mathematical model.Most of the existing neural network adaptive control methods are based on uniform asymptotic bounded theory rather than finite time bounded theory,and finite time control is more meaningful.Based on the characteristics of inverted pendulum system,the neural network adaptive finite time constraint control of inverted pendulum system is investigated in this paper.Firstly,based on the nonlinear characteristics of the inverted pendulum system,the pure feedback uncertain nonlinear system finite time adaptive control is investigated.Through the mean value theorem,the pure feedback nonlinear systems is converted to strict feedback nonlinear systems,and radial basis function neural network approximate nonlinear system.The barrier Lyapunov functional and adaptive law of the weights of neural networks is designed to realize the nonlinear systems neural networks adaptive state constraint control.Secondly,based on the convergence rate of neural networks control,and the finite-time adaptive control is designed to realize the nonlinear systems finite-time state constrained neural networks adaptive control.By selecting barrier Lyapunov functional and finite time neural networks weight adaptive law,the closed-loop system is proved to be finite time bounded stable based on Lyapunov stability theory,and the feasibility of the algorithm is verified by MATLAB simulation experiments.Finally,the hardware and software of the rotary inverted pendulum system is designed,and the neural network controller is used to realize the stability of the inverted pendulum.Neural network controller adopts state constrained finite time adaptive control to ensure the performance of swapping-up process and stabilization process.A barrier Lyapunov functional is designed to prove that the inverted pendulum segment can track the target trajectory within a bounded time,and the inverted pendulum can balance vertically in a finite time with small error.Finally,experimental results are given to verify the effectiveness of the proposed method.