Research On Fault Diagnosis For Dissimilar Redundant Actuation System Of More Electric Aircraft

The more electric aircraft drive system uses electric system to partially replace the original hydraulic,pneumatic and mechanical systems.Compared to conventional aircraft drive systems,more electric aircraft reduces dependence on hydrocarbon fuels and lower carbon emissions.The drive system of more electric aircraft is a dissimilar residual actuation system by combining different types of actuators.The probability of fault increases with the increase of system complexity and number of components.Therefore,it is important to conduct research on the fault diagnosis and remaining useful life(RUL)prediction for the dissimilar redundant actuation system of more electric aircraft in terms of improving the reliability and safety of the residual actuation system.This thesis proposes a fault diagnosis and remaining useful life prediction method based on bond graph for the dissimilar redundant actuation system of more electric aircraft.Firstly,the bond graph theory is used to model the dissimilar redundant actuation system of more electric aircraft,and the linear fraction transformation(LFT)method is used to describe the uncertainty of parameters to obtain Bond Graph-Linear Fraction Transformation(BG-LFT)model.Then the interval adaptive threshold of the system is obtained based on BG-LFT,combined with interval analysis.Compared with the adaptive threshold,the interval adaptive threshold improves the performance of fault detection.Secondly,the model decomposition method is used to decompose the bonding diagram model of the dissimilar redundant actuation system of the multi-electric aircraft into several local subsystem models.The distributed analytical redundancy relations(DARRs)are derived based on the local subsystem models,and the fault signature matrix(FSM)is established based on the DARRs.The DARRs and FSM are used to achieve distributed fault detection and isolation.Next,in order to determine the true faults of the system from the isolated possible faults,it is necessary to use the method of fault estimation to jointly estimate the possible fault parameters and system state variables.After the fault estimates are obtained,the nominal values of the parameters are compared to determine the true faults.This thesis introduces the extended Kalman filter algorithm commonly used in parameter estimation methods,and then proposes an improved extended Kalman filter algorithm based on fuzzy logic inference,and applies the algorithm to subsystems to estimate fault parameters.After the parameter estimation is completed,the remaining useful life prediction method based on the T-S fuzzy neural network model is proposed.Compared with the traditional BP neural network,the T-S fuzzy neural network can approximate the more complex nonlinear function by designing a small number of fuzzy rules.Finally,the effectiveness and feasibility of the method are verified by simulation.