Research On Aero-engine Fault Diagnosis And Performance Recovery Control Technology

To increase the safety of engine and decrease the cost of maintenance, meanwhile ensuring the degenerated aero-engine can maintain adequate performance for flight safety; this paper develops researches on aero-engine fault diagnosis and performance recovery control technology.Firstly, aero-engine fault diagnosis has been studied. Due to wide range of aero-engine's work and parameters, a single point to establish Kalman filter cannot adapt to the fault diagnosis of flight envelope. Therefore, combined with improved Kalman filter, according to current altitude, Mach, fuel flow, through 3d-interpolation method, fault diagnosis of flight envelope in subsonic and supersonic is expanded. Performance parameters tracking of engine model is established based on neural network with estimated deterioration information. In view of the contradiction that engine sensors' number less than engine deterioration number, reduced-dimensional Kalman filter based on SVD(Singular Value Decomposition) has been designed to solve the problem of unpredictable performance parameters under the condition of insufficient sensors.Secondly, performance recovery control has been studied. The least squares fitting method is adopted to optimize the parameters of ALQR(Augmented Linear Quadratic Regulator) in the inner loop control. Thrust estimation model is proposed in the outer loop control based on the least square support vector regression. The input of the thrust estimation model is selected through the thrust feature selection algorithm. Furthermore, fuzzy-PI controller is designed for the outer controller. In that case, the rotational speed can be adaptively controlled so as to implement degenerate engine performance recovery. The simulation results show that degenerate engine can quickly recover to the rated thrust, fully extracting the potential performance of degenerate engine.Lastly, due to the changes of the turbine tip clearance considered as an engine turbine fault, clearance control is similar to the performance recovery control. Therefore, turbine tip clearance modeling and active clearance control have been discussed. Semi-infinite transient heat conduction equations combined with polynomial fits is proposed to estimate the distribution of temperature in the tip clearance calculation model. The average temperature can be used to compute the tip clearance. With the variation of tip clearance, the turbine performance parameters could be amended to reflect effects of static and dynamic performance with the changes of turbine tip clearance. Mechanical active clearance control system is designed to control the rapid change of tip clearance with PID controller based on neural network. The validity is verified by the simulation results during the transient process of ground acceleration and cruise deceleration.