Model Parameters Accurate Identification Of Cart Triple-pendulum System And Its Swing Up And Handstand Controller Design And Parameters Tuning

Cart multi-pendulum system is a typical of nonlinear, under-actuated complex system. The study of pendulum can be treated as the study of nonlinear, multivariable, under-actuated and absolutely unstable system. Cart pendulum system usually be treated as a examination platform of practice for kinds of nonlinear control theories to validate its correctness, validity and application problems in real control. Control problems from any barycenter supra and pivot below being in daily life to special aerocraft and kinds of servo-platform have great comparability between control problems of pendulum. So the study of pendulum not only have important academic meaning but also have important practice meaning.Movement control problem of multi-pendulum is a representative difficult problem in control domain, its production of real control is very little in the world. Professor lizushu and his team introduce their original human simulated intelligent control based on sensor-motor intelligent schema to control multi-pendulum, successfully achieve swing up and handstand control and switch control between balanceable poses of double-pendulum, also simulate swing up and handstand control of triple-pendulum. But it is hard to achieve realtime control, it exists great difficulties from simulation to realtime control. In the first place, it have a great difference between simulation model and real system, also it is hard to adjust parameter of swing up and handstand controller. Model parameters usually are determined by measure directly and traditional identification after the dynamic model is deduced by Lagrange method. It has great difference between dynamic response by simulation and real response, at the same time, human simulated intelligent controller that can achieve swing up and handstand control of cart triple-pendulum have the structure of multi-mode and multi-controller and have many parameters, it is very difficulty to adjust and optimize parameters.Aim at the two problems proposed above, evolutional thought is introduced in this paper to combine human simulated intelligent control theory and genetic arithmetic,so an improved genetic arithmetic is introduced. For model parameters identification and human simulated intelligent controller parameters adjustment we designed corresponding fitness functions, numerical simulation experiments indicate that we can achieve precise identification of dynamic model parameters and many groups valid or optimized parameters of human simulated intelligent controller.