The Research On Multi-model Predictive Active Fault-tolerant Control Method Based On SVM

Multi-model predictive active fault-tolerant control method based on LS-SVM is proposed for the nonlinear systems with variety of known or unknown faults, which is based on former multiple models active fault-tolerant control. The main research work on it is done as follows:1) For the nonlinear systems with prior fault knowledge, a new nonlinear active fault-tolerant control method based on LS-SVM multiple models predictive control is presented for dynamic process, which directs shortcoming that it is difficult to determine the network structure, slow convergence and easy to fall into local minimum in BP neural network and directs shortcoming that it is complicated to calculate control law in nonlinear predictive control method. This method linearizates SVM kernel function to simplify the complexity of control low. Simulations are given to demonstrate effectiveness of the proposed approach.2) For the non-linear systems, dynamic data driven is lead into LS-SVM multiple models active predictive fault-tolerant control, which achieves fault-tolerant for the system with unknown faults. When unknown faults occur, it calculates predictive control law based on new LS-SVM models which set up online. Simulations are given to demonstrate that it is effective for both known faults and unknown faults.3) There is some brief analysis on the performance of the LS-SVM multiple models active fault tolerant control system. Some simulations on systems which contain some uncertainty due to the environment in the thesis are analyzed. The simulations indicate the system has some robustness in the condition of a limited disturbance and small faults, which means that the supervision does not have false judge and false controller reconfiguration and the controller itself has some robustness.