Linear Inverted Pendulum Stability Control, Adaptive Neural Network Fuzzy Inference System Studies

Inverted pendulum is a typical model of multi-variable, nonlinear, essentially unsteady system. The only effective means can realize stability of the system. In practical engineering, many control systems resemble inverted pendulum. Thus, the stability research of inverted pendulum system has the profound meaning both in theory and in practice.In this dissertation, the characteristics of inverted pendulum and the research meaning of stable control is analyzed. Besides, research actuality of inverted pendulum is introduced. According to Lagrange equation, model of single, double and triple inverted pendulum are constructed. The stability, controllability and observability of inverted pendulum system is analyzed. Based on this, the dissertation presents Adaptive-Neural-Network-based Fuzzy Inference System to realize the balance of inverted pendulum. ANFIS, which can modify fuzzy rules and membership functions, has advantage of both fuzzy logic and Neural-Network. Considering multi-variable of inverted pendulum, state synthesis is adopted to reduce the number of input variable dimension and the number of control rules, in order to avoid the explosion of rules.The simulation results show inverted pendulum based on ANFIS is feasible. Compared with LQR control arithmetic, inverted pendulum based on ANFIS control has good dynamics performance and anti-interference capability. When parameters of model are changed, ANFIS controller has strong adaptability. To more approve that ANFIS is feasible in the control of inverted pendulum practicality, single and double inverted pendulum practicality realize balance based on Real-Time Workshop. From the control result of inverted pendulum practicality, inverted pendulum practicality based on ANFIS realizes balance in the direction of uprightness. And cart precisely reaches target position and anti-jamming of whole inverted pendulum system is well.