Actuator Fault Adaptive Compensation Control For A Class Of MIMO Nonlinear Systems

In this thesis, fault-tolerant control for a class of multi-input multi-output (MIMO) nonlinearsystems with actuator failures are studied, the problems are studied in the aspects of actuator failurestype, system dynamic performance, system uncertainty and actuator grouping. Based on differentialgeometry feedback linearization, backstepping control, model follow adaptive control and multi-modeswitching techniques, a set of control and adaptive compensation control methods are proposed.Firstly, an adaptive compensation control law is designed for a class of MIMO nonlinearminimum phase systems with actuator failures. Based on the differential geometry feedbacklinearization method, prescribed performance bound (PPB) and backstepping technique, an adaptivecompensation tracking control scheme is designed for the system with actuators lock in space and lossof effectiveness failures. The proposed control law can guarantee that the closed-loop system withactuator failures is stability and asymptotically tracks the given reference signals, and the systemtransient performance is guaranteed.Secondly, a neural adaptive compensation control scheme for a class of MIMO uncertainnonlinear systems with actuator failures is proposed based on PPB transient performance whichcharacterizes the convergence rate and maximum overshoot of the tracking error. Radial BasisFunction (RBF) neural networks are used to approximate the error of plant model, the control lawproposed can guarantee the asymptotic output tracking and closed-loop signal bounds.In order to enlarge the set of compensable actuator failures, an actuator grouping scheme basedon multiple model switching and tuning (MMST) is proposed for the nonlinear MIMO minimumphase systems with multiple actuator failures. Then an adaptive compensation scheme based on PPBis designed for the system to ensure closed-loop signal boundedness and asymptotic output trackingdespite unknown actuator failures.Then, based on the previous research, the application object of MMST grouping method isexpaned to uncertainty of MIMO nonlinear systems, due to the uncertainty of system will increase thedifficulty of identification models tracking, so use neural network to estimation uncertainty ofidentification model in order to eliminate the effects of uncertainty. And an adaptive Control law isdesigned to ensure closed-loop signal boundedness and asymptotic output tracking despite unknownactuator failures. The control scheme is applied to an aircraft dynamics model, and the simulationresults demonstrate the effectiveness of the proposed method.Finally, Aircraft control simulation software was designed by using Microsoft Visual C++(VC) to design the control interface of simulation platform, using MATLAB to achieve flight control modelto calculate, applying network programming technology to achieve the data transmission betweenvarious parts, and using the MATLAB engine for VC controlling MATLAB model. The designedflight simulation platform is capable of real-time flight simulation, and laying the foundation for theengineering application of theoretical research results.