Ensemble of unsupervised fuzzy C-means classifiers for clustering health status of oil sand pumps

Detection of anomalies and faults in slurry pumps is an important task with implications for their safe, economical, and efficient operation. Wear, caused by abrasive and erosive solid particles, is one of the main causes of failure. Condition monitoring and on-line assessment of the wear status of wetted components in slurry pumps are expected to improve maintenance management and generate significant cost savings for pump operators. In this context, the objective of the present work is to present a framework for the assessment and measurement of the wear status of slurry pumps when available data is extremely limited. Four sequential steps are performed: data collection, feature extraction, feature selection and classification. The main idea is to combine the predictions of multiple unsupervised classifiers fed with different inputs taken from different signals, based on fuzzy C-means clustering, to reduce the variance of the results so that they are less dependent on the specifics of a single classifier. This will also reduce the variance of the bias, because a combination of multiple classifiers may learn a more expressive concept class than a single classifier. © 2012 Taylor & Francis Group.