Dynamic model identification of gas turbines

In this paper the recorded data from normal engine operation have been exploited for dynamic model identification. This would usually require test inputs to be applied to excite sufficiently the engine dynamics. However, in this work a novel approach is adopted utilising existing random disturbances created by the controller and operating environment during normal engine operation as the test source. A key advantage of this passive identification approach is that special engine test-runs and monitoring equipment are not required. An optimization technique of spectral analysis is derived providing high accuracy of identification. Optimal smoothing provides the compromise between the bias and variance of spectral estimates. An example of the turbo-fan engine model identification is considered demonstrating the viability of the proposed method