A methodology for developing local smart diagnostic models using expert knowledge

摘要

This paper describes an innovative modular component-based modelling approach for diagnostics and condition-monitoring of manufacturing equipment. The approach is based on the use of object-oriented Bayesian networks, which supports a natural decomposition of a large and complex system into a set of less complex components. The methodology consists of six steps supporting the development process: Begin, Design, Implement, Test, Analyse, and Deploy. The process is iterative and the steps should be repeated until a satisfactory model has been achieved. The paper describes the details of the methodology as well as illustrates the use of the component-based modelling approach on a linear axis used in manufacturing. This application demonstrates the power and flexibility of the approach for diagnostics and condition-monitoring and shows a significant potential of the approach for modular component-based modelling in manufacturing and other domains.