Effects of sampling decimation on a gas turbine performance monitoring

Abstract

Monitoring the performance of gas turbine engines (GTEs) by sampling the operating parameters of the GTEs is the central part of the GTEs health management program. The rate of data sampling and the consequent analyses of the sampled data are restricted to the available resources. It especially appears as a principal constraint where the data is manually logged by the operators. In a recent research work, a physics-based approach and resulting performance indicators, i.e., “Heat Loss index” and “Power Deficit index” were introduced by the authors to monitor the health state of the gas turbines using only the readings from the GTE operating system. Statistical estimation approach was taken to establish prediction models for performance indicators. This study provides a quantitative analysis for the effect of sampling decimation on the accuracy of the developed predictor within a time window. Consequently, it provides an insight into the performance prediction uncertainty, in connection with the sampling frequency and the length of the observation window on which the model is established.