Nonlinear Filtering Methods Of Gas Path Fault Diagnosis For Aeroengine

Aero-engine fault diagnosis is one of the most effective ways to reduce maintenance cost and ensure flight safety. The nonlinear filters is studied and applied to health parameters estimation for turbofan engine gas path fault diagnosis in this dissertation.Extended Kalman filter(EKF) algorithm and its fault diagnosis application are presented. The effects to estimate accuracy of four matrices in the EKF, such as Jacobian matrix A and C, the process noise covariance matrix Q and the measurement noise covariance matrix R, are discussed.In order to introduce the prior knowledge into gas path fault diagnosis, the constraint EKF algorithms are researched. This prior knowledge of component health should be transformed into the inequality constraints, and the EKF is separately modified by the least mean square(LMS) and the truncated probability density function(PDF). The constraint mean square error function is minimized the least mean square function, and at the same time, the method of Lagrange multipliers is utilized to solve the inequality-constraint equations. The prior inequality constraint is transformed into probability density function in the truncated PDF, and the standard normal distribution function is obtained.A method of linear combination of health parameters is proposed to solve the problem that the number of available sensors is less than that of health parameter. A transformation matrix is selected through minimizing the EKF's estimation errors. The matrix is used to reconstruct a tuning parameter vector which is a linear combination of all health parameters, and the reconstructed vector dimension is equal to the sensor number. The health parameters estimates are obtained through the tuning parameter vector, whose dimension is low enough to enable EKF estimation.Finally, the EKF and its improved algorithms above are tested both on digital simulation and the rapid prototyping platform. The simulation results presented that the modification of EKF with the LMS and the truncated PDF, the underdetermined EKF can be realized in the engine health management platform based on the Lab VIEW environment and the embedded controller CRIO.