Application of a Fuzzy Logic Controller for speed control on a Small-Scale Turbojet Engine

Having a small-scale turbojet engine operate at a desired speed with minimum steady state error, while maintaining good transient response is crucial in many applications, such as UAVs, and requires precise control of the fuel flow. In this paper, first the mathematical model of a Small-Scale Turbojet Engine (SSTE) is obtained using system identification tests, and then based on this model, a classical PI controller is designed. Afterwards, to improve on the transient response and steady state performance of this classical controller, a Fuzzy Logic Controller (FLC) is designed. The design process for the FLC employs logical deduction based on knowledge of the engine behavior and iterative tuning in the light of software- and hardware-in-the-loop simulations. The classical and fuzzy logic controllers are both implemented on an in-house, embedded Electronic Control Unit (ECU) running in real time. This ECU is an integrated device carrying a microcontroller based board, a fuel pump, fuel line valves, speed sensor and exhaust gas temperature sensor inputs, and starter motor and glow plug driver outputs. It mainly functions by receiving a speed reference value via its serial communication interface. Based on this reference, a voltage is calculated and applied to the fuel pump in order to regulate the fuel flow into the engine, thereby bringing the engine speed to the desired value. Pre-defined procedures for starting and stopping the engine are also automatically performed by the ECU. Further, it connects to a computer running an in-house comprehensive Graphical User Interface (GUI) software for operating, monitoring, configuration and diagnostics purposes. The designed controllers are used to drive a generic SSTE. Reference inputs consisting of step, ramp and chirp profiles are applied to the controllers. The engine response using both controllers are recorded and inspected. The results show that the FLC exhibits a comparable performance to the classical controller, with possible opportunities to improve this performance.