Fuzzy Sliding Mode Control For Magnetic Levitation System Based On Neural Network

The improvement of the machining accuracy is limited by the friction existing between the moving platform and the stationary rail in platform-moving type CNC machine tools. In order to eliminate friction completely, traditional support system is instead of platform magnetic levitation system which is no mechanical contact, it can achieve precision work in this thesis. The main goal of this study is that advanced sliding mode variable structure control strategy is used to ensure fast track and strong robustness of single electromagnetic levitation system of vertical direction support part.Firstly, on the basis of the analysis of the suspension principle of support parts in platform magnetic levitation system, mathematical model of single electromagnetic levitation system is established. In order to facilitate the design of the controller, for nonlinear characteristics of the magnetic levitation system, its mathematical model is converted to affine nonlinear mathematical model with the third-order single-input and single-output by using coordinate transformation theory.Secondly, for the problem of the steady-state error of the magnetic levitation system by the sliding mode variable structure without integral part, integral sliding mode variable structure control schemes is designed. By the design method of changing sliding mode surface, achieve the purpose of weakening the steady-state error,however, the output response curve of the system has a big chattering. Then, for the problem of the decline of the control accuracy of suspended air gap by chattering of the sliding mode variable structure control, the fuzzy inference method is introduced, and the fuzzy integral sliding mode variable structure control schemes is designed, achieve weakening chattering of the sliding mode variable structure control effectively.Thirdly, the system can achieve the control requirements, but it is difficult to achieve the best control effect, because fuzzy rules and fuzzy inference process are based on experience of experts. From this problem, neural fuzzy integral sliding mode variable structure control scheme is designed, and achieve the control effect better by neural network learning constantly and improving the network structure weights.Finally, MATLAB simulation software is used to separately verify the effectiveness of above three advanced sliding mode variable structure control strategies for single electromagnetic levitation system, and simulation result show that the above control methods can enhance the control performance of the suspension system in vary degrees and ensure fast track and strong robustness of single electromagnetic levitation system.