Modeling of turbojet Fuel Control Unit using NARX-neural network

Fuel control unit (FCU) is one of the most essential parts in a gas turbine engine; therefore it is necessary to be studied as an important part of the fuel control system. This paper report the use of Nonlinear Auto Regressive with eXogenous input (NARX) neural network model for modeling of the jet engine FCU. Therefore، To measure and recording data from the FCU inputs and output، the test bench including hydraulic system، data acquisition system and induction motor control system are designed and constructed. This setup is a mechatronic collection which includes mechanical design، discharge and pressure sensors، tachometer، control unit and piping systems. The process of modeling is carried out in MATLAB software. The identified model is evaluated with validation data and its response is compared with the real system response. Results demonstrate the effectiveness of the NARX neural network model and show that the real system is estimated by the NARX neural network model accurately.