Self Tuning On-board Real-time Engine Model For EHM

Aero-engine self tuning on-board real time model is the fundamental technology of engine health management. The self tuning on-board real time model for engine health management is studied in this paper.The methodology for establishing state variable model is studied based on the aero-engine nonlinear model. The effect of nonlinear modeling precision to state variable model is analyzed. The relation between the eigenvalues of system matrix and linearization range referring to nonlinear model is studied. The state variable model with more than three state variables is solved through composite least-square method proposed in this paper.A large-scale state variable model is founded through combining state variable model with associated tables of steady-state trim values. The augmented state variable model (ASVM) is founded after the health parameters are augmented as the new state variables of the state variable model referring to engine degradation. The engine self tuning on-board real time model is established based on Kalman filter. The observability of the Kalman filter system and the relationship of matrix condition number and estimation precision are also studied.In order to reduce the modeling error between the steady-state trim values and the real engine, a neural network is built into the self tuning on-board real time model. The neural network modification improves the precision of the engine self tuning on-board real time model. The self tuning on-board real time model founded in this paper can apply for the aero-engine fault diagnosis and health trend analysis.