Research On Sensor Signal Reconstruction Technology For Turbofan Engine Control System

Sensor signal reconstruction plays an important role in enhancing the reliability of aero-engine control system. The sensor signal reconstruction for a twin-spool high-bypass-ratio separate-exhaust turbofan engine was researched into based on component-level model which was simplified and improved to increase the computational efficiency.A nonlinear component-level model for high-bypass-ratio turbofan engine was built up which considered the influence of high bypass ratio as well as engine air bleeding and deflating. Steady and transient processes from idle to maximum power were simulated by the model. Simulation results show the validity of the model which lays the foundation for sensor signal reconstruction.Based on the component-level model, the improvement of computational efficiency was achieved through reducing single-step simulation time and flow path iterative times. The time consumed by each computation module was tested and analyzed using multimedia timer on the platform of VC++6.0. Key factors affecting real-time computation were identified. Interpolation table describing gas thermodynamic properties was built. Binary search method and a modified binary search method was used. The time consumed by single-step simulation is reduced 80% to 0.02 ms on the platform of 3.3GHz Intel CPU without loss of simulation accuracy. The iterative times of Broyden method and Newton-Raphson method were analyzed. Simulation results show that flow path iterative times of Broyden method is less than that of Newton-Raphson method.A new thermodynamic computation based method was presented to reconfigure sensor signals according to the signal reconfiguration requirement of aero-engine gas path sensors for the control system. The sensor signal was reconfigured through improving and simplifying component-level model using the effective sensor measurements. Sensor signal reconstruction was realized for a high-bypass-ratio turbofan engine through developing the corresponding reconfiguration structures for related sensor signals. Computation time is further reduced by means of simplifying flow path calculation and reducing the amount of iterative variables compared to the component-level model. On the platform of 168 MHz STM32F407, the reconstruction time is within 9ms. Compared with rig test data, the steady-state reconfiguration error is less than 2% and the transient-state reconfiguration error is less than 5%.