Optimal tuner selection using Kalman filter for a real-time modular gas turbine model

In this study, a real-time flexible modular modeling approach for the simulation of gas turbine engines dynamic behavior based on nonlinear thermodynamic and dynamic laws is addressed. The introduced model, which is developed in Matlab-Simulink environment, is an object-oriented high speed real-time computer model and is capable of simulating the dynamic behavior of a broad group of gas turbine engines due to its modular structure. Moreover, a Kalman filter-based model tuning procedure is applied to decrease the modeling errors. Modeling errors are defined as the mismatch between simulation results and available experimental data. This tuning procedure is an underdetermined estimation problem, where there are more tuning parameters than available measured data. Here, an innovative approach to produce a tuning parameter vector is introduced. This approach is based on seeking an optimal initial value for the Kalman filter tuning procedure. Three simulation studies are carried out in this paper to demonstrate the advantages, capabilities and performance of the proposed scheme. Furthermore, simulation results are compared with manufacturer’s published data, and with the experimental results gathered in either turbo-generator or turbo-compressor applications. Computational time requirement of the model is discussed at the end of the paper.