RBF-ARX Modeling And Cloud Inference Controller Of Triple Inverted Pendulum

Inverted pendulum is a complex nonlinear system. Firstly, it needs to establish the precise model of inverted pendulum before control it. The traditional modeling method adopts mathematical method. This method ignores secondary factors to establish the linear model through the linearization. In order to achieve better control effect, it needs to establish a more accurate nonlinear model. In this paper, RBF-ARX(radial basis Function-Auto Regressive e Xogenous) is proposed. The triple inverted pendulum nonlinear model is established based on testing data, and its control system is also established, respectively, using T-S fuzzy neural network optimized by Cloud Genetic Algorithm(CGA) and T-S cloud inference neural network optimized by traditional genetic algorithm(GA) as the controller. This system realizes the stable control of the triple inverted pendulum by simulation.Firstly, RBF-ARX model is in-depth study, finding that the original SNPOM optimization method has such disadvantages like computational complexity, slow operation, and difficult to spread in the field of engineering application. The genetic algorithm(GA) instead of the SNPOM is used to optimize all the parameters of the model. So, the nonlinear model of triple inverted pendulum based on GA-RBF-ARX is established. The new scheme is not simplified optimization process only, but also established more accurate nonlinear model for controller.Secondly, the nonlinear model of triple inverted pendulum based on GA-RBF-ARX provides the basic for the T-S fuzzy neural network control system. CGA is introduced into the control system as optimization algorithms for T-S fuzzy neural network controller, which can overcome the shortcomings of GA, like slow convergence and easily trapping into the local convergence. CGA is a combination of cloud model and GA optimization algorithm. Therefore, it is superior to the traditional genetic algorithm at the aspect of convergence speed and optimization results. The simulation shows that the CGA fuzzy neural network control system based on T-S fuzzy neural network is an effective control system, which has a very good control effect.Finally, the cloud theory is in-depth study based on the T-S fuzzy neural network. T-S cloud inference neural network is proposed by combining cloud model and T-S fuzzy neural network. It is used as the controller of triple inverted pendulum. Triple inverted pendulum control system is established based on T-S cloud inference neural network optimized by GA. After simulation, the results show that the proposed method has the advantages of both fuzzy neural network and cloud model, and the method increases the ability of the network to deal with the problem of uncertainty, and provides a new way of thinking for the control of complex systems.