A learning, representation and diagnostic methodology for engine fault diagnosis

Machine-learning techniques and their applications in manufacturing

“…Other applications include generating decision rules for the conceptual design of steel members under bending [152], diagnosing engine faults [153], detecting defects in disk drive manufacturing [154], diagnosing motor pumps [155], analysing nondestructive testing of spotweld quality [156], managing and controlling the procurement of raw materials [157], accelerating rotogravure printing [158,159], optimizing processes in electrochemical machining [160], selecting appropriate cutting tools in grinding [161], determining suitable cutting conditions in operation planning [162,163], re-formulating and generalizing the machining knowledge from a machining database [164], choosing sheet metal working conditions [165], discovering the laws governing metallic behaviour [166], modelling job complexity in clothing production systems [167], acquiring and refining operational knowledge in industrial processes [168,169], and identifying arbitrary geometric and manufacturing categories in CAD databases [170].…”

A Review of Rules Family of Algorithms

“…Medical applications such as lymphography, prognosis of breast cancer recurrence, location of primary tumour and thyroid problem diagnosis have been reported [5,24,25]. Other applications include investment appraisal [26], forensic classification of glass fragment evidence [27], extraction of decision rules for analysis of test data for the space shuttle main engine [28], experimental generation of decision rules for the conceptual design of steel members under bending [10], soil classification [29], stock control [30], software resource analysis [31], assessing credit card applications [32], military decision making [33], dynamic system identification [34,35], engine fault diagnosis [36] and identification of the mass-spectra of complex materials [37,38,39].…”

Intelligent Systems in aerospace

“…To enhance the capabilities of MLS, Ke and Ali employed an inductive learning technique, called Task-Oriented Clustering (TOC), which uses attribute vectors combined with symbolic conceptual descriptions to deal with the structural object domain. The modified new system is called "The Learning and Diagnosis System (LDS) (Ke & Ali, 1989). The representation combining numerical attribute vectors and symbolic descriptions makes it easier in LDS to apply various domain knowledge and to improve the time-efficiency of the diagnosis task as well as the incremental learning process.…”