The performance of gas turbines degrades over time due to deterioration mechanisms and single fault events. While deterioration mechanisms occur gradually, single fault events are characterized by occurring accidentally. In the case of single events, abrupt changes in the engine parameters are expected. Identifying these changes as soon as possible is referred to as detection. State-of-the-art detection algorithms are based on expert systems, neural networks, special filters, or fuzzy logic. This paper presents a novel detection technique, which is based on Bayesian forecasting and dynamic linear models (DLMs). Bayesian forecasting enables the calculation of conditional probabilities, whereas DLMs are a mathematical tool for time series analysis. The combination of the two methods can be used to calculate probability density functions prior to the next observation, or the so called forecast distributions. The change detection is carried out by comparing the current model with an alternative model, where the mean value is shifted by a prescribed offset. If the forecast distribution of the alternative model better fits the actual observation, a potential change is detected. To determine whether the respective observation is a single outlier or the first observation of a significant change, a special logic is developed. In addition to change detection, the proposed technique has the ability to perform a prognosis of measurement values. The developed method was run through an extensive test program. Detection rates >92% have been achieved for changed heights, as small as 1.5 times the standard deviation of the observed signal (sigma). For changed heights greater than 2 sigma, the detection rates have proven to be 100%. It could also be shown that a high detection rate is gained by a high false detection rate (∼2%). An optimum must be chosen between a high detection rate and a low false detection rate, by choosing an appropriate uncertainty limit for the detection. Increasing the uncertainty limit decreases both detection rate and false detection rate. In terms of prognostic abilities, the proposed technique not only estimates the point of time of a potential limit exceedance of respective parameters, but also calculates confidence bounds, as well as probability density and cumulative distribution functions for the prognosis. The conflictive requirements of a high degree of smoothing and a quick reaction to changes are fulfilled in parallel by combining two different detection conditions.

This paper presents a novel technique for automatic change detection of the performance of gas turbines. In addition to change detection the proposed technique has the ability to perform a prognosis of measurement values. The proposed technique is deemed to be new in the field of gas turbine monitoring and forms the basic building block of a patent pending filed by the authors [1]. The technique used is called Bayesian Forecasting and is applied to Dynamic Linear Models (DLMs). The idea of Bayesian Forecasting is based on Bayes’ Theorem, which enables the calculation of conditional probabilities. In combination with DLMs (which break down the chronological sequence of the observed parameter into mathematical components like value, gradient, etc.) Bayesian Forecasting can be used to calculate probability density functions prior to the next observation, so called forecast distributions. The change detection is carried out by comparing the current model with an alternative model which mean value is shifted by a prescribed offset. If the forecast distribution of the alternative model better fits the actual observation, a potential change is detected. To determine whether the respective observation is a single outlier or the first observation of a significant change, a special logic is developed. Studies have shown that a confident change detection is possible for a change height of only 1.5 times the standard deviation of the observed signal. In terms of prognostic abilities the proposed technique not only estimates the point of time of a potential limit exceedance of respective parameters, but also calculates confidence bounds as well as probability density and cumulative distribution functions for the prognosis.

This paper presents a method to improve the process of gas turbine diagnosis in case of a single fault event by calculating both deviations of performance parameters from their nominal values and fault probabilities. The approach is based on a thermodynamic model of the monitored gas turbine that is used to generate an Influence-Coefficient-Matrix (ICM) describing the gas turbine’s behaviour at the desired operating point in a linearised form. The ICM is used to calculate deviations of performance parameters by least squares estimation. As it is not known a priori how many components are affected by a fault, the ICM is evaluated with respect to all possible fault numbers and combinations leading to sets of analyses. The nature of least squares estimation method limits the number of detectable faults (and thus the number of analyses sets) to the number of available measurements. In a second step a fuzzy logic system is applied to calculate the probability of each performance parameter to be actually affected by the fault. This is done by applying fuzzy logic rules taking into account the frequency and the mean deviation of every performance parameter within every set of analyses. The third and last step is to sum the probabilities over all sets in order to obtain a global probability for every performance parameter. The proposed technique has been applied to several simulated test cases with encouraging results. The main benefit of the technique is that the diagnostic result contains both magnitude and probability of the deviations of performance parameters. Furthermore, robustness in terms of measurement noise is achieved by using a statistical evaluation method.

No abstract