The research on the real-time and full-enveloped model identification of turbofan engine's hydro-mechanical actuator

In order to add the static and dynamic characteristics of the actuator full-enveloped in the design of turbofan engine control system and hardware in loop simulation, the actuator of turbofan engine real-time model identification research is conducted. Firstly, on the basis of the controlling bandwidth range of engine control system, multiple semi physical experimental data as the initial value are selected, so that the identification model can be performed the full envelope characteristics of actuator; And then, the kernel function extraction to initial feature can reduce vector dimension for ensuring the identification of the global optimal solution; Finally, the least squares support vector machine (LS-SVM) for identification of turbofan engine hydro-mechanical main control system. In the process of identification, the introduction of vector base learning (VBL) can improve the model speed and ensure the real time of the model. Compared hardware in loop simulation results with the data of semi physical experiment, the comparison results show what as is followed: The error between the simulation and the semi physical experiment was less than 2%; the real-time calculation period of the model was less than 4ms, which has a high instantaneity. And those indicated that the model can satisfy the design of turbofan engine control system and the need of hardware in loop simulation.