Research On Sensor Failure Diagnosis And Signal Accommodation Of Aero-engine Control System

Sensors used by the aero-engine control system are easy to fail. Sensor failure detection, isolation and accommodation are important to the control system and aero-engine safety. 33.28 engine control systems item of airworthiness regulation clearly regulates the safety of civil aero-engine control system. Sensor failure detection, isolation and accommodation of a civil geared turbofan(GTF) engine was studied.The nonlinear thermal-dynamic model of GTF engine was developed based on component-level modeling method for the purpose of sensor failure diagnosis. The particular gear box was processed in the modeling application according to the gear ratio and the test data of GTF engine designing point. The component models, such as fan and compressor, were established. Then the steady simulation of GTF engine was studied since the steady model was established and solved. Comparing the simulation result and test data of the engine designing point, all parameter error except thrust is less than 3%, so the steady model of GTF engine was validated. And the dynamic simulation of GTF engine was studied since the dynamic model was established and solved. After the fuel step was applied on the designing point and cruise point of GTF engine, the simulation results indicates that the dynamic model reflects the dynamic characteristic of the engine, which proves the validity of the dynamic model.The nonlinear model of GTF engine was linearized at designing point and cruise point. Then the linear state variable model was developed. The linear state variable model matches with the nonlinear thermal-dynamic model when the step of the same control variable was applied so that the model validation was achieved. The continuous state variable model was discretized to meet the requirement of computer processing.The sensor failure detection, isolation and accommodation simulation was studied based on state variable model of GTF engine. The sensor failure diagnosis and signal accommodation simulation platform was developed based on a bank of Kalman filter. The detection threshold was determined by the failure indication signal of no sensor failure. The sensor failure, including hard failure, soft failure, hard failure adding soft failure and successive soft failure, was studied. The simulation results show that four kinds of sensor failure were detected, isolated and accommodated, which ensures the safety of controls system. The effect of different failure magnitude on detection, isolation and accommodation was analyzed. It could be concluded that all sensor failure that insistently exceed 1.8% and whose failure indication signal exceed the detection threshold could be detected and isolated. And the smaller the detectable sensor failure is, the accommodation signal is closer to the real value.