Application Of A Kind Of Improved T-S Type Fuzzy Neural Network Control In The Planar Single Inverted Pendulum

The inverted pendulum is a kind of typically multi-variable, nonlinear, unstable, close coupling and fast system. It is considered as a multiuse and comprehensive instrument. The studing on it can be thought of the studing on multi-variable, nonlinear system. Its control method and thought have general directive significance in dealing with the general industrial process. With the development of intelligent control, the new control method is appearing constantly, people attempt to examine the validity of the new control method by use of the inverted pendulum which is a strict control object. Therefore, as the representative and complex research object of controlled field , the research of the inverted pendulum system has important engineering background and profound significance.Moreover, in the sense of the control theory and control engineering, comparing with the linear inverted pendulum, the planar inverted pendulum system's variable, nonlinear degree and unstability increase manyfold that cause its realization of control is much more difficult than that of the linear inverted pendulum system.At first, the paper illustrates the background and scintific significance of the reseach for controlling of the inverted pendulum system. It gives a briefly introduce about the inverted pendulum system, and discussed reseach actuality of the inverted pendulum in the world and development trend of the intelligent control.Secondly, the paper introduced the physical structure of Google GPIP2001 planar single inverted pendulum system, and deduced mathematics model of the planar single inverted pendulum detailly. The deducing course entirely is succinct and clear, avoided some overelaborate contents in newton mechanics. On this basis , it adopt LQR algorithm based on the modern control theory to control the planar inverted pendulum and got a comparatively satisfactory control result.Finally, the paper analyses the existing intellectual control strategy on the inverted pendulum carefully and catch the idea that the current situation both at home and abroad of fuzzy neural network control theory. There are some defects in the T-S type fuzzy neural network, such as: its structure immobilizes, the rule library is unable to be concise and that when input variable's dimension is too great or cut apart fuzzily detailly, network structure will be very huge, the memory space will be too big, the speed of studied and controlling will be slow. So, the paper take a kind of improved Takagi-Sugeno type fuzzy neural network into the parallel control of the planar inverted pendulum, its basic thought is: Set up fuzzy neural network controller to x direction and y direction of the inverted pendulum separately, both the controllers describe system with fuzzy language; optimize parameter with neural network; adjust nodes of the network through establishing threshold value to consequent parameter that is obtained at first study and achieve the goal of optimizing the structure of neural network, then study again the sample data, adjuste secondly the membership function of every variable, concise the control rulesand make the structure of controller simple, easy to realize.The simulation of control on the planar inverted pendulum has proved the feasibility and effectiveness of this strategy which is used in the kind of nonlinear multivariable and unstable system, it provides a new train of thought to the fuzzy neural network to control this kind of problem.