Research On Intelligent Control Of Swing-up And Stabilization Of The Linear Inverted Pendulum

The Linear linearly inverted pendulum, can include multiple input and output controlling variables, and have the significant nonlinearity and instability. All these characteristics have made it as one of the reliable platforms to study the control theories and carry out various experiments. It has been widely used in the fields of aerospace, robotics and so on. At present, the investigation about the linear inverted pendulum mainly focuses on the swing-up control and stability control. As to the swing-up control, it is difficult to realize rapid swing-up control because of the limitations of the systems. As to the stability control, more research results come from the theory or algorithm of the intelligent control. There is little investigations considering multiple intelligent control methods or combining these methods with traditional PID method. In view of the above questions, further studies on the swing-up control and stability control of linear inverted pendulum are made based on hybrid intelligent control methods. The concrete contents are:Firstly, numerical models of linear and double inverted pendulum are established according to Newton-Euler and Lagrangian methods respectively. Based on the models, the state equations are derived and the controllability, observability and stability of the pendulums are analyzed.Secondly, an optimization swing-up control strategy which is based on Lyapunov energy feedback is designed and the real-time control of swing-up of linear single inverted pendulum is achieved. Considering of the range of pendulum lever switch angle, the stability pendulum angle and car offset, switching control strategy is designed and smooth transition from swing-up control to stability control of the line level inverted pendulum is realized.Thirdly, because the intelligent control method is not based on accurate numerical model, stability pendulum control strategies which combine BP, RBF, ANFIS, KOHONEN and other neural network with PID control are presented. These strategies take advantage of the performances of the neural networks’self-organizing, self-adapting and self-learning. The corresponding simulation and experimental research are completed.Finally, stability controller for double inverted pendulum is designed by using LQR algorithm and BP, RBF, ANFIS, KOHONEN, etc. The simulation and measurement results illustrate that the optimized pendulum control algorithm improves the swing-up speed of the pendulum.Therefore, it is proved that the proposed control algorithm is feasible and effective.