Michael Gallimore

Applied sensor fault detection and validation using transposed input data PCA and ANNs

The paper presents an efficient approach for applied sensor fault detection based on an integration of principal component analysis (PCA) and artificial neural networks (ANNs). Specifically, PCA-based y-indices are introduced to measure the differences between groups of sensor readings in a time rolling window, and the relative merits of three types of ANNs are compared for operation classification. Unlike previously reported PCA techniques (commonly based on squared prediction error (SPE)) which can readily detect a sensor fault wrongly when the system data is subject bias or drifting as a result of power or loading changes, here, it is shown that the proposed methodologies are capable of detecting and identifying the emergence of sensor faults during transient conditions. The efficacy and capability of the proposed approach is demonstrated through their application on measurement data taken from an industrial generator.

Sensor fault detection for industrial gas turbine system by using principal component analysis based y-distance indexes

The paper presents a readily implementable and computationally efficient method for sensor fault detection based upon an extension to principal component analysis (PCA) and y-distance indexes. The proposed extension is applied to system data from a sub-15MW industrial gas turbine, with explanations of the eigenvalue/eigenvector problem and the definition of z-scores and principal component (PC) scores. The y-distance index is introduced to measure the differences between sensor reading datasets. It is shown through use of real-time operational data that in-operation sensor faults can be detected through use of the proposed y-distance indexes. The efficacy of the approach is demonstrated through experimental trials on Siemens industrial gas turbines.

Novelty detection based on extensions of GMMs for industrial gas turbines

The paper applies the application of Gaussian mixture models (GMMs) for operational pattern discrimination and novelty/fault detection for an industrial gas turbine (IGT). Variational Bayesian GMM (VBGMM) is used to automatically cluster operational data into steady-state and transient responses, where extraction of steady-state data is an important preprocessing scenario for fault detection. Important features are extracted from steady-state data, which are then fingerprinted to show any anomalies of patterns which may be due to machine faults. Field data measurements from vibration sensors are used to show that the extensions of GMMs provide a useful tool for machine condition monitoring, fault detection and diagnostics in the field. Through the use of experimental trials on IGTs, it is shown that GMM is particularly useful for the detection of emerging faults especially where there is a lack of knowledge of machine fault patterns.

Unit operational pattern analysis and forecasting using EMD and SSA for industrial systems

This paper studies operational pattern analysis and forecasting for industrial systems. To analyze the global change pattern, a novel methodology for extracting the underlying trends of signals is proposed, which is based on the sum of chosen intrinsic mode functions (IMFs) obtained via empirical mode decomposition (EMD). An adaptive strategy for the selection of the appropriate IMFs to form the trend, is proposed. Then, to forecast the change of the trend, Singular Spectrum Analysis (SSA) is applied. Results from experiment trials on an industrial turbine system show that the proposed methodology provides a convenient and effective mechanism for forecasting the trend of the operational pattern. In so doing, it therefore has application to support flexible maintenance scheduling, rather than the traditional use of calendar based maintenance.

Hybrid Hierarchical Clustering — Piecewise Aggregate Approximation, with Applications

Piecewise Aggregate Approximation (PAA) provides a powerful yet computationally efficient tool for dimensionality reduction and Feature Extraction (FE) on large datasets compared to previously reported and well-used FE techniques, such as Principal Component Analysis (PCA). Nevertheless, performance can degrade as a result of either regional information insufficiency or over-segmentation, and because of this, additional relatively complex modifications have subsequently been reported, for instance, Adaptive Piecewise Constant Approximation (APCA). To recover some of the simplicity of the original PAA, whilst addressing the known problems, a distance-based Hierarchical Clustering (HC) technique is now proposed to adjust PAA segment frame sizes to focus segment density on information rich data regions. The efficacy of the resulting hybrid HC-PAA methodology is demonstrated using two application case studies viz. fault detection on industrial gas turbines and ultrasonic biometric face identification. Pattern recognition results show that the extracted features from the hybrid HC-PAA provide additional benefits with regard to both cluster separation and classification performance, compared to traditional PAA and APCA alternatives. The method is therefore demonstrated to provide a robust and readily implemented algorithm for rapid FE and identification for datasets.

Sensor fault detection for industrial systems using a hierarchical clustering-based graphical user interface

The paper presents an effective and efficient method for sensor fault detection and identification within a large group of sensors based upon hierarchical cluster analysis. Fingerprints of the hierarchical clustering dendrograms are found for normal operation using normalized data, and sensor faults are detected through cluster changes occurring in the dendrogram. The proposed strategy is built into a user-friendly graphical interface, which is applied to a sub-15MW industrial gas turbine. It is shown, through use of real-time operational data, that inoperation sensor faults can be detected and identified by the hierarchical clustering-based graphical user interface.

Self-organising symbolic aggregate approximation for real-time fault detection and diagnosis in transient dynamic systems

The development of accurate fault detection and diagnosis (FDD) techniques are an important aspect of monitoring system health, whether it be an industrial machine or human system. In FDD systems where real-time or mobile monitoring is required there is a need to minimise computational overhead whilst maintaining detection and diagnosis accuracy. Symbolic Aggregate Approximation (SAX) is one such method, whereby reduced representations of signals are used to create symbolic representations for similarity search. Data reduction is achieved through application of the Piecewise Aggregate Approximation (PAA) algorithm. However, this can often lead to the loss of key information characteristics resulting in misclassification of signal types and a high risk of false alarms. This paper proposes a novel methodology based on SAX for generating more accurate symbolic representations, called Self-Organising Symbolic Aggregate Approximation (SOSAX). Data reduction is achieved through the application of an optimised PAA algorithm, Self-Organising Piecewise Aggregate Approximation (SOPAA). The approach is validated through the classification of electrocardiogram (ECG) signals where it is shown to outperform standard SAX in terms of inter-class separation and intra-class distance of signal types.

Operational pattern analysis for predictive maintenance scheduling of industrial systems

The paper presents a method to identify the operational usage patterns for industrial systems. Specifically, power measurements from an industrial gas turbine generator are studied. A fast Fourier transform (FFT) and image segmentation is used to develop an intuitive representation of operation. A spectrogram is adopted to study the average usage through the use of spectral power indices, with singular spectral analysis (SSA) applied for operational trend extraction. Through use of these techniques, two fundamental inputs for predictive maintenance scheduling viz. the users behaviour with regard to long-term unit startups patterns, and the duty cycle of power requirements, can be readily identified.

Trend extraction based on Hilbert-Huang transform

Trend extraction is an important tool for the analysis of data sequences. This paper presents a new methodology for trend extraction based on Hilbert-Huang transform. Signals are initially decomposed through use of EMD into a finite number of intrinsic mode functions (IMFs). The Hilbert marginal spectrum of each IMF is then calculated and a new criterion, termed the cross energy ratio of the Hilbert marginal spectrum of consecutive IMFs, is defined. Finally, through use of the new criterion, the underlying trend is obtained by adaptively selecting appropriate IMFs obtained by EMD. Results from experimental trials are included to demonstrate the benefits of the proposed method for extracting trends in data streams.