Model-Based Control of an Aircraft Engine using an Optimal Tuner Approach

Abstract

This paper covers the development of a model-based engine control (MBEC) methodology applied to an aircraft turbofan engine. Here, a linear model extracted from the Commercial Modular Aero-Propulsion System Simulation 40,000 (CMAPSS40k) at a cruise operating point serves as the engine and the on-board model. The on-board model is updated using an optimal tuner Kalman Filter (OTKF) estimation routine, which enables the on-board model to self-tune to account for engine performance variations. The focus here is on developing a methodology for MBEC with direct control of estimated parameters of interest such as thrust and stall margins. MBEC provides the ability for a tighter control bound of thrust over the entire life cycle of the engine that is not achievable using traditional control feedback, which uses engine pressure ratio or fan speed. CMAPSS40k is capable of modeling realistic engine performance, allowing for a verification of the MBEC tighter thrust control. In addition, investigations of using the MBEC to provide a surge limit for the controller limit logic are presented that could provide benefits over a simple acceleration schedule that is currently used in engine control architectures.

Towards Run-time Assurance of Advanced Propulsion Algorithms

“…30,31 The tracking filter includes an optimal tuner estimation routine based on a Kalman filter design. 32 This advanced controller directly feeds back an estimate of thrust to control thrust directly, rather than indirectly through other sensed engine parameters.…”

Runtime Assurance Protection for Advanced Turbofan Engine Control

“…Therefore the total reference input for experiment #2 was the nominal acceleration schedule, r, minus the nominal output data from experiment #1, 1 y , the original closed-loop acceleration response. The output of experiment #2, 2 y , is the closed-loop transfer function acting on the tracking error signal: 01 () T r y  , which is used to compute the tracking error gradient.…”

“…This section presents an initial investigation of RTA protection applied to a particular high fidelity turbofan engine model developed by NASA and its contractors known as the Commercial Modular Aero-Propulsion System Simulation 40k (C-MAPSS40k) [11,12]. This model was used to develop an advanced model-based engine control (MBEC) system for experimental study [13]. Since thrust cannot be directly measured, traditional engine control indirectly controls thrust through related parameters, such as fan speed or engine pressure ratio (EPR), which is the low pressure turbine discharge pressure P50 divided by the inlet pressure, P2.…”

“…Improved tracking responses were observed with no change to the existing protection logic control architecture.2 conditions are ensuing due to undiscovered errors in the untrusted advanced system, then the RTA system activates one or more recovery actions to mitigate the adverse conditions and return operation to a safe/correct state.Because of the need to increase the operating performance of turbofan engines for new and future aerospace applications, there is currently wide interest in using more advanced control algorithms to achieve these new capabilities. The current investigation is motivated by a number of recent advanced control approaches in the propulsion community to provide better performance over the life cycle of the engine [1,2]. However, it is recognized that certification of such advanced controllers may be difficult because of their inherent complexity with intelligent and possibly nondeterministic elements.…”

“…Because of the need to increase the operating performance of turbofan engines for new and future aerospace applications, there is currently wide interest in using more advanced control algorithms to achieve these new capabilities. The current investigation is motivated by a number of recent advanced control approaches in the propulsion community to provide better performance over the life cycle of the engine [1,2]. However, it is recognized that certification of such advanced controllers may be difficult because of their inherent complexity with intelligent and possibly nondeterministic elements.…”