Hybrid Modeling Based Fault Diagnosis And Fault Tolerant Control For Quadrotor Helicopter

Quadrotor helicopter is a typical modern complex system. Compared with fixed-wing aircrafts, quadrotor helicopters have in general more complicated aerodynamic characteristics and quite special flight attitudes. Accordingly, they have urgent requirements on more accurate system models and more robust controller designs to assure flight quality and safety.Hybrid modeling strategy is one of the most effective ways to get accurate and reliable models. In the existing hybrid modeling methods, it is difficult to clarify the scope of prior models and nonparametric models. Besides, there is lack of a uniform criterion for selection of proper linearization methods when analyzing the nonparametric models. Aiming at the above issues, a hybrid modeling method based on physical effect analysis and nonlinearity measure is proposed. The experimental results can show the accuracy and reliability of the developed hybrid model.With the developed hybrid model, a principal component analysis(PCA) based double-granularity fault detection and diagnosis(FDD) method is proposed. The coarse-granularity-level FDD is built on the prior model which can isolate the faulty channel(s)(i.e., pitch, yaw and roll); while the fine-granularity-level FDD is built on the nonparametric model which can isolate the faulty components in the faulty channel. The experimental results show that the proposed double-granularity FDD method can effectively detect and diagnose the structural damages in a quardrotor helicopter.Finally, a dynamic feedback fault tolerant controller scheme is designed, which can assure the stability analysis and the tracking performance of the close-loop system in the presence of structural damages. The simulation results show that the proposed fault tolerant control method has better fault tolerant performance and less energy consumption.