Application of the Systematic Sensor Selection Strategy for Turbofan Engine Diagnostics

Abstract

The data acquired from available system sensors forms the foundation upon which any health management system is based, and the available sensor suite directly impacts the overall diagnostic performance that can be achieved. While additional sensors may provide improved fault diagnostic performance there are other factors that also need to be considered such as instrumentation cost, weight, and reliability. A systematic sensor selection approach is desired to perform sensor selection from a holistic system-level perspective as opposed to performing decisions in an ad hoc or heuristic fashion. The Systematic Sensor Selection Strategy is a methodology that optimally selects a sensor suite from a pool of sensors based on the system fault diagnostic approach, with the ability of taking cost, weight and reliability into consideration. This procedure was applied to a large commercial turbofan engine simulation. In this initial study, sensor suites tailored for improved diagnostic performance are constructed from a prescribed collection of candidate sensors. The diagnostic performance of the best performing sensor suites in terms of fault detection and identification are demonstrated, with a discussion of the results and implications for future research.

Application of the Sensor Selection Approach in Polymer Electrolyte Membrane Fuel Cell Prognostics and Health Management

“…Considering the fact that sensors may express different sensitivities to fuel cell performance change, and environment/measurement noise is usually contained in the collected measurements, more sensor measurements may not provide better performance, thus it is necessary to find the optimal sensors providing reliable results with the minimum computation cost. From a literature review, several sensor selection approaches have been applied in different systems [19][20][21][22], but their applications in fuel cell system PHM have not been fully investigated.…”

Measurement Selections for Multicomponent Gas Path Diagnostics Using Analytical Approach and Measurement Subset Concept

“…The normalized matdx is then multiplied by its transpose to produce correlation square matdx Q which consists of correlation coefficients c as the matrix elements, where Q = PP^ (9) Each correlation coefficient is actually the cosine of the angle between a pair of measurements, which indicates the level of vector parallelism in component change space [11]. Very high correlations can be identified if the value is in between 0,9 and 1.0 or in between -1.0 and -0,9.…”

“…Sowers et al [9] used optimal selection to optimize the figure of merit (FOM) as the objective function in their selection of measurement sets for turbofan engine diagnostics. However, these optimal selection techniques require complex iterative scheme and still computationally demanding to seek for the optimum solution.…”

Sensor selection in neuro-fuzzy modelling for fault diagnosis

“…Interesting work has been presented by Sowers et al in their paper [10] but from the point of view of selecting sensors for diagnostic purposes on health monitoring of an aerospace system. Their study concentrates on the application of model-based technique on statistical evaluation techniques some being application dependent and some non-application dependent.…”