Predictive Control Of A Double Inverted Pendulum Research Based On Neural Network

The inverted pendulum can show many abstract concepts of control, for example, the stability, the convergence and the ability of resisting interface. It is an ideal model in testing various of control theories. A lot of researchers of modern control theory are attracted by nonlinearity, strongly-coupling of the inverted pendulum. Especially, the inverted pendulum is on the focus of people. So far, the control methods, explored by studying the inverted pendulum, have widely applied to war industry, space flight, robots and general industrial processes.This paper focus on double link rotary inverted pendulum, using Lagrange energy equation to build the inverted pendulum mathematical model, which greatly reduced the flexibility of the system modeling course. On the foundation of the known inverted pendulum mathematical model, taking neural networks model for predictive control and combing three kinds of mechanism of predictive control, the predictive control based on neural networks is discussed. Because of the trouble of finding the global optimal solution in the cost function of rolling optimization of the algorithm, we can linear the predictive model round the operation to get the instantaneous model. The generalized predictive control algorithm is used to get an approximate solution first the final global optimal solution can then been found by optimization of the cost function. A simulation study of double inverted pendulum is also given.At final, this paper presents an H∞robust model predictive control (MPC) scheme for a double inverted pendulum with unknown but bounded persistent disturbances. Firstly, the input-to-state stability of discrete-time systems is analyzed; Secondly, the control action of MPC is designed and an estimate solution of its terminal constrained set is addressed. Then, the robust stability of the controlled system with the mentioned scheme is achieved. Finally, a simulation demonstrates the effectiveness of the proposed scheme.