Hardware-in-the-loop neuro-based simulation for testing gas turbine engine control system

Abstract

Designing and testing gas turbine engine control systems requires hardware-in-the-loop (HIL) simulation to improve project time and guarantees safety. A HIL bench should provide real time calculations of object models. Thermodynamic gas turbine models are mostly not applicable for real-time computations due to solving constraints. Models should be accurate and easy-calculation for gas turbine engine modeling in the HIL. Those models can be created via neural networks. Thus, aim of this research is to design hardware-in-the-loop neuro- based simulation for testing gas turbine engine control system. The neural network model is based on JETCAT-P60 testing data. After network is synthesized, a code implementation is generated and integrated in MCU software. The regulator is implemented in another MCU-based electronic unit. The two units interact by simulating real system signals (PWM control and PFM frequency signal)$.\mathrm {I}\mathrm {n}$ result, the HIL-bench was verified by the JETCAT-P60 experiment and control system was tested.