The aim of this study is to evaluate gas path diagnostic techniques using a principle of variable structure classification applied to cover possible fault scenarios in gas turbine maintenance. This principle allows creating more versatile and realistic fault conditions relative to existing studies such as complex fault classifications, a new boundary for fault severity, and real deviation errors. The techniques analyzed are included into a special procedure that repeats a diagnostic process many times and computes for each fault class a probability of correct diagnosis. Using this probability averaged for all the classes as the evaluation criterion, the techniques are tested under the conditions of four comparative studies. The results show that (a) there is no single technique significantly outperforming all others over the full range of diagnostic conditions even if engine operating modes, fault simulation data, fault classifications, multiple-class boundaries or the scheme of deviation errors are varied; (b) the common level of diagnosis accuracy greatly depends on the fault classification used; (c) significant influence of fault severity boundary is found. The boundary proposed makes the level of accuracy much more realistic compared to simplified boundaries previously used; and (d) the use of real deviation noise in fault class description instead of simulated errors further approaches the diagnostic conditions and results to the level expected in practice.
The present paper compares the fault recognition capabilities of two gas turbine diagnostic approaches: data-driven and physics-based (a.k.a. gas path analysis, GPA). The comparison takes into consideration two differences between the approaches, the type of diagnostic space and diagnostic decision rule. To that end, two stages are proposed. In the first one, a data-driven approach with an artificial neural network (ANN) that recognizes faults in the space of measurement deviations is compared with a hybrid GPA approach that employs the same type of ANN to recognize faults in the space of estimated fault parameter. Different case studies for both anomaly detection and fault identification are proposed to evaluate the diagnostic spaces. They are formed by varying the classification, type of diagnostic analysis, and deviation noise scheme. In the second stage, the original GPA is reconstructed replacing the ANN with a tolerance-based rule to make diagnostic decisions. Here, two aspects are under analysis: the comparison of GPA classification rules and whole approaches. The results reveal that for simple classifications both spaces are equally accurate for anomaly detection and fault identification. However, for complex scenarios, the data-driven approach provides on average slightly better results for fault identification. The use of a hybrid GPA with ANN for a full classification instead of an original GPA with tolerance-based rule causes an increase of 12.49% in recognition accuracy for fault identification and up to 54.39% for anomaly detection. As for the whole approach comparison, the application of a data-driven approach instead of the original GPA can lead to an improvement of 12.14% and 53.26% in recognition accuracy for fault identification and anomaly detection, respectively.
This work proposes a universal data-driven approach to compute and monitor gas turbine unmeasured variables. To this end, a large amount of unmeasured and measured data is first computed at steady state for both baseline and faulty engine conditions using a nonlinear thermodynamic model. On the data generated, polynomial models that relate the unmeasured quantities with the measured variables are then determined. These data-driven models allow the computation of unmeasured variables and their deviations. Accuracy analysis is conducted separately for baseline and current estimates of unmeasured variables and for deviation estimates. All the results prove that the estimates are exact enough. Thus it is possible to obtain a universal fast and accurate method for computing important unmeasured gas turbine quantities that is suitable for practical applications. The method promises a drastic increase in the diagnostic capabilities of online monitoring systems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.