The Research On Multi-model Based Inverse System Of Active Fault- Tolerant Control

This paper take the nonlinear system as the object of study, proposed in the earlier period in the multi-model active fault-tolerant control thought's foundation , will go the inverse system approach to unify with the multi-model method, uses the least squares support vector machine (LS-SVM) establishment inverse system model, and does have the model-free adaptive control and internal model control to carry on the fault-tolerant control to the system, mainly has carried on the following several aspect research work:1) The failure for a certain prior knowledge of the nonlinear system, in order to solve the multi-model active fault tolerant control system failure in response to calculate the fast control law is proposed which is based on LS-SVM built inverse system modules within the multi-model fault tolerant control methods. First, the system uses a variety of normal and fault conditions prior to model the inverse system, and thus construct the inverse model library of system, and then inverse system method based on inverse model in series with the controlled system, the formation of pseudo-linear system, and the design a good robustness of the internal model controller. System online operation, monitoring sole system performance, decision-making mechanism based on the calculation of indicators, called inverse model matching, so that it always remains the same in series of pseudo-linear system, and no need to change in the internal model control the case of control of nonlinear systems to achieve fault tolerance of the fast and reliable.2) For prior knowledge of the actual system failure and the actual system is not easy to get, further studied one kind based on the data driven inverse multi-model acive fault-tolerant control method which is based on data-driven model of inverse system of the mlti-model active fault tolerant control method. In this method, LS-SVM on the system to normal conditions and a variety of prior establishment of the inverse model of fault conditions, and thus construct the basic inverse model library system, the actual run-time system performance based on switching function, the line calculation and analysis, to determine the operating mode of the system, known to occur when the system fails, call the inverse model matching, so that it always remains the same in series of pseudo-linear system, and within a fixed mode controller for nonlinear system on the rapid known fault tolerance; An unknown when the system fails, switch to data-driven technique based on model-free adaptive controller (MFAC) for the transition fault tolerant control, in ensuring the stability of the system while taking advantage of fault-tolerant inputs and outputs during the transition data to establish the system inverse model of the current fault conditions and added to the model library, it has a self-learning ability, enables the system to deal with the unknown fault.3) The research method has carried on the valid simulation confirmation in the paper, its result showed that regardless of whether there is the system to the fault apriori knowledge, in the paper does state the method has the correct judgment, the active learning and good robustness fault-toleran ablity to the outside perturbation and the parameter perturbation fault system. This has provided the available method for nonlinear system's active fault-tolerant control.