Optimal Robust Matching of Engine Models to Test Data

摘要

Status matching supports USAF turbine engine development, qualification, and maintenance test planning and diagnostics. Future USAF maintenance concepts will require that engine status decks be made more frequently than they are today, and thus the process must be faster and require less expert knowledge than the traditional approach. The research program developed an improved, automated process for calibrating turbine engine performance models. The Filtered Monte Carlo (FMC) and the Singular Value Decomposition (SVD) algorithms were found to meet the sponsor's requirements for a robust, fast process suitable for inexperienced users. Both methods were demonstrated to successfully match measured data with no prior knowledge of the engine. The methods are complementary in that an initial FMC analysis can identify for an inexperienced user which model tuning parameters are significant and can provide him or her with appropriate ranges for those parameters. The SVD method may subsequently be used to quickly determine the best value for each modifier. The engine status matching process is also applicable to calibration of other types of models. Similar methods have been used by the researchers for calibration of engine, aircraft, noise, and emissions models and for calibration of lower fidelity aero models to CFD models.