Process monitoring and fault detection strategies: a review

Purpose – Monitoring of a process leading to the detection of faults and determination of the root causes are essential for the production of consistent good quality end products with improved yield. The history of process monitoring fault detection (PMFD) strategies can be traced back to 1930s. Thereafter various tools, techniques and approaches were developed along with their application in diversified fields. The purpose of this paper is to make a review to categorize, describe and compare the various PMFD strategies. Design/methodology/approach – Taxonomy was developed to categorize PMFD strategies. The basis for the categorization was the type of techniques being employed for devising the PMFD strategies. Further, PMFD strategies were discussed in detail along with emphasis on the areas of applications. Comparative evaluations of the PMFD strategies based on some commonly identified issues were also carried out. A general framework common to all the PMFD has been presented. And lastly a discussion into future scope of research was carried out. Findings – The techniques employed for PMFD are primarily of three types, namely data driven techniques such as statistical model based and artificial intelligent based techniques, priori knowledge based techniques, and hybrid models, with a huge dominance of the first type. The factors that should be considered in developing a PMFD strategy are ease in development, diagnostic ability, fault detection speed, robustness to noise, generalization capability, and handling of nonlinearity. The review reveals that there is no single strategy that can address all aspects related to process monitoring and fault detection efficiently and there is a need to mesh the different techniques from various PMFD strategies to devise a more efficient PMFD strategy. Research limitations/implications – The review documents the existing strategies for PMFD with an emphasis on finding out the nature of the strategies, data requirements, model building steps, applicability and scope for amalgamation. The review helps future researchers and practitioners to choose appropriate techniques for PMFD studies for a given situation. Further, future researchers will get a comprehensive but precise report on PMFD strategies available in the literature to date. Originality/value – The review starts with identifying key indicators of PMFD for review and taxonomy was proposed. An analysis was conducted to identify the pattern of published articles on PMFD followed by evolution of PMFD strategies. Finally, a general framework is given for PMFD strategies for future researchers and practitioners.