Fault Tolerant Model Predictive Control of a Turbofan Engine using C-MAPSS40k

摘要

Engine control is a crucial component for the safe and stable operation of gas turbine engines which are complex nonlinear systems. As engines have evolved to higher capabilities it is essential to update the control strategy. The generic commercial turbofan engine simulation C-MAPSS40k developed by NASA has a gain scheduled PID baseline controller with additional limit controllers for constrained operation of the nonlinear engine within its physical limits. In this paper we propose the replacement of the PID baseline controller with a Model Predictive Control (MPC), a control technique that can handle complex constrained dynamics. Due to its heavy computational burden, MPC has not been implemented practically for engine control purposes. However, recent developments in digital electronics have caused a transition from a centralized to a more distributed engine control architecture allowing the integration of higher processing capabilities onboard aircraft engines. During an engines operation occurrences of fan blade-shroud rubbing, presence of sand particles and structural wear and tear due to heating effects cause a decrease in the efficiency of engine components. This degradation is modeled using health parameters like fan efficiency, flow capacity etc. in C-MAPSS40k. In this paper we apply a single variable MPC based on a linearized time invariant (LTI) model of the C-MAPSS40k engine with on-line optimization. The performance of MPC is compared with that of the baseline PID controller in terms of handling the constraints and engine degradation for a power level angle (PLA) demand profile. Based on the results, it is concluded that MPC can replace the PID baseline controller without loss in performance and a MPC based on a degraded model can handle a fixed range of degradation in the actual linear time invariant engine.