A case-based reasoning system for fault detection and isolation: a case study on complex gearboxes. Issue 2 (11th January 2019)

• Record Type: Journal Article
• Authors: Boral, Soumava; Chaturvedi, Sanjay Kumar; Naikan, V.N.A.
• Is Part Of: Journal of quality in maintenance engineering. Volume 25:Issue 2(2019)
• Page Start: 213
• Page End: 235
• Publication Date: 2019-01-11
• Subjects: Artificial intelligence -- Case-based reasoning -- Intelligent maintenance -- Fault detection and isolation -- Maintenance decision support system; Plant maintenance -- Quality control -- Periodicals; Total quality management -- Periodicals; Total productive maintenance -- Periodicals; 658.202
• DOI: 10.1108/JQME-05-2018-0039
• Languages: English
• ISSNs: 1355-2511

Abstract

Abstract : Purpose: Usually, the machinery in process plants is exposed to harsh and uncontrolled environmental conditions. Even after taking different types of preventive measures to detect and isolate the faults at the earliest possible opportunity becomes a complex decision-making process that often requires experts' opinions and judicious decisions. The purpose of this paper is to propose a framework to detect, isolate and to suggest appropriate maintenance tasks for large-scale complex machinery (i.e. gearboxes of steel processing plant) in a simplified and structured manner by utilizing the prior fault histories available with the organization in conjunction with case-based reasoning (CBR) approach. It is also demonstrated that the proposed framework can easily be implemented by using today's graphical user interface enabled tools such as Microsoft Visual Basic and similar. Design/methodology/approach: CBR, an amalgamated domain of artificial intelligence and human cognitive process, has been applied to carry out the task of fault detection and isolation (FDI). Findings: The equipment failure history and actions taken along with the pertinent health indicators are sufficient to detect and isolate the existing fault(s) and to suggest proper maintenance actions to minimize associated losses. The complex decision-making process of maintaining such equipment can exploit the principle of CBR and overcome the limitations of the techniques such as artificial neural networks and expert systems. The proposed CBR-based framework is able to provide inference with minimum or even with some missing information to take appropriate actions. This proposed framework would alleviate from the frequent requirement of expert's interventions and in-depth knowledge of various analysis techniques expected to be known to process engineers. Originality/value: The CBR approach has demonstrated its usefulness in many areas of practical applications. The authors perceive its application potentiality to FDI with suggested maintenance actions to alleviate an end-user from the frequent requirement of an expert for diagnosis or inference. The proposed framework can serve as a useful tool/aid to the process engineers to detect and isolate the fault of large-scale complex machinery with suggested actions in a simplified way.