home // Aviation // Aircraft

pdf download Nonlinear Model Predictive Control of an Aircraft Gas Turbine Engine

 
Nonlinear Model Predictive Control of an Aircraft Gas Turbine Engine Source: oodgeroo.ucsd.edu
File Size: 109.02 KB
Category: Aviation
Last Download : 7 days 14 hours 29 minutes ago

Share this info:


Short Description: applied to a high-fidelity turbojet aircraft engine model. Strong ... a high-fidelity model of a military aircraft engine, showing that ...

Content Inside: Proceedings of the 41st IEEE Conference on Decision and Control Las Vegas, Nevada USA, December 2002 FrP08-7 Nonlinear Model Predictive Control of an Aircraft Gas Turbine Engine Brent. J. Brunell, Robert R. Bitmead , Allan J. Connolly Abstract complexity (8 states), and performance focus, this is a demanding task. This application includes the use of an Extended Kalman Filter state estimator as the adjunct to a full- The feasibility of constrained nonlinear model predictive state NMPC. The analysis here explores the use of a simplified control (NMPC) with state estimation is investigated and model as the predictive core of the NMPC and of the EKF. The applied to a high-fidelity turbojet aircraft engine model. Strong performance is based on application to a higher fidelity nonlinearities are present in turbojet aircraft engines due to the component level model. large range of operating conditions and power levels experienced during a typical mission. Also, turbine operation is restricted due to mechanical, aerodynamic, thermal, and flow 2 Component Level Model limitations. NMPC is selected because it can explicitly handle the nonlinearities, and both input and state constraints of many This engine is an aerodynamically coupled, dual rotor machine variables in a single control formulation. Due to the wherein a low-pressure rotor system (fan and low-pressure computational requirements of NMPC and the fast dynamics of turbine) is mechanically independent of a high-pressure (core aircraft engines, a Simplified Real Time Model (SRTM) is engine) system. Air entering the inlet is compressed by the fan created that captures all of the relevant dynamics while and then split into two concentric streams. One of these then executing quickly. An Extended Kalman Filter (EKF) is enters the high-pressure compressor and proceeds through the applied to estimate the states in the presence of noise and main engine combustor, high-pressure turbine, and low-pressure limited sensor data. This output feedback controller is tested on turbine. The other is directed through an annular duct and then a high-fidelity model of a military aircraft engine, showing that recombined with the core flow, downstream of the low-pressure NMPC based on the simplified model has the potential to turbine, by means of a convoluted chute device. The combined achieve better performance than the production controller. streams then enter the augmenter to a convergent-divergent, variable area exhaust nozzle. Here the flow is pressurized, 1 Introduction expands, and accelerated rearward into the atmosphere, generating thrust. Gas turbines can be used for propulsion as aircraft engines and The plant model is a physics based component level model for power generation in land based power systems. The gas (CLM) of this turbine configuration, which was developed by turbine model considered is a low bypass, two rotor, turbojet GE Aircraft Engines. This model is very detailed, high-fidelity, with a variable exhaust area typical of military aircraft and models each component starting at the inlet, through the applications. During normal operation this turbine experiences fan, compressor, combustor, turbines, and exhaust nozzle. large changes in ambient temperature, pressure, Mach number, and power output level. For each of these variations the engine 3 Simplified Real Time Model dynamics change in a significant nonlinear manner. Careful attention must be paid by the controller during engine operation Since NMPC is a model-based control, an internal model is to ensure that the mechanical, aerodynamic, thermal, and flow needed to predict the future responses of the plant to control limitations of the turbo machinery are maintained. In addition, inputs. As the CLM is a very large and complicated model, a the control authority is restricted by the actuator rate and new model was developed to be used in the NMPC that has a saturation limits. Current technology solves this nonlinear small number of states, executes quickly, can be analytically constrained problem using many SISO linear controllers in linearized, and is accurate to within 20 percent in transient and 5 concert that are gain scheduled and min/max selected to protect percent steady state over the area of the flight envelope that is against engine limits. While this method has many merits we most used. The SRTM has two control inputs; fuel flow propose solving the problem using NMPC, which handles the demand (WFDMD), and exhaust nozzle area demand MIMO nonlinearities and constraints explicitly and in a single (A8DMD), as well as ambient condition inputs; altitude (ALT), control formulation. Mach (XM), and ambient temperature deviation from ISO (DTAMB). The outputs are; percent core speed (PCN25), NMPC is model-based, has the capacity to accommodate percent fan speed (PCN2), engine pressure ratio (PP), constraints, and relies on on-line open-loop optimization via a compressor discharge static pressure (PS3), high pressure receding-horizon formulation. It is inherently multi- turbine exit temperature (T4B), fan stall margin (SM2), core input/multi-output and can be tuned for performance and stall margin (SM25), and thrust (FNAV). stability [1,2]. Vroemen [3] has also considered the application of NMPC to a laboratory gas turbine. Here we develop these Following the model structure proposed by Shaoji [4], a SRTM ideas further by using an Extended Kalman Filter for state of an aircraft engine along with the main fuel-metering valve estimation and a nonlinear model of the dynamics. (MFMV) and variable exhaust nozzle (A8) actuators is developed that meets the above specifications. The model is The approach of this paper is to use this application of NMPC designed to replicate both transient and steady state to explore the development of a constrained MIMO controller performance. as a replacement for multiple single-loop closures. The objective is for performance while satisfying operational For validation the SRTM is run open loop versus the CLM. The constraints. Because of the sampling speed (10 ms), model input profiles for the validation are a large step increase in fuel The authors gratefully acknowledge the US Air Force funding to support IHPTET Phase III Control Technologies. GE Global Research Center, One Research Circle, Niskayuna, NY 12309 USA. {brunelbr,connolly}@crd.ge.com Mech & Aero Eng, University of California San Diego, La Jolla, CA 92093-0411, USA. rbitmead@ucsd.edu 0-7803-7516-5/02/$17.00 ©2002 IEEE 4649 Document OutlineMAIN MENUPREVIOUS MENU---------------------------------Search CD-ROMSearch ResultsPrint

Advertiser

download pdf for free Go to Download Page

Related PDF Files