Aero-engine Parameter Estimation In Performance-seeking Control

High performance, high efficiency and adaptive propulsion system has been thefocus of aviation. Aero-engine performance-seeking control (PSC) is a technologythat can ensure the engine’s security and stability as well as can improve it’s potentialfor the better performance. And estimated performance parameters for aero-engine area key technology to realize its advanced control methods.In this paper, a turbofan engine is regarded as the research object, and a furtherresearch on the aero engine performance estimation method using Kalman filteralgorithm is studied. We compared two different methods to build the engine model.The methodology for establishing state variable model (SVM) based on theaero-engine nonlinear model is discussed, which is analyzed for the way to set upmodel. The appropriate engine parameters to ensure the accuracy of the model areselected, which can reflect the true conditions for the requirements of PSC. Then, thispaper mainly focuses on how to use the Kalman filter estimator to design engineonboard adaptive model and the theory, characteristic and realization of Kalman wasintroduced as well.For the four selected component parameters，the situation of one degradation andmultiple parameters degradation was assumed and the amount of degradation wasestimated. The simulation results show that the Kalman filter is able to estimate theparameters with high accuracy. Finally, For the deviation of the aero-engine on-boardadaptive system model could not be completely eliminated and it may result in seriousestimated deviations and filtering divergence, a new Kalman estimation algorithmwith fading factor was proposed. Adjusting the weight of innovation covariance andincreasing the effect on realistic measurement data in state estimation were introducedto solve the problem, which ensured the accuracy of aero-engine parametersestimation. Compared with the conventional Kalman filter, simulation results showsthat the method proposed has the ability of restraining filtering divergence and canobtain the high accuracy of estimation and the short convergent time. Above all, thenew method is simple and the computation burden is low as well as engineeringapplications value is high.