Modelling a Manufacturing Line using Extended Object Oriented Bayesian Network

Abstract Bayesian Network (BN) is a widely used modelling tool in probabilistic reasoning; however it turns out to be difficult to use this tool to model a large scale complex system such as a manufacturing line due to the number of parameters when the system exceeds a certain amount of components. Motivated by the necessity to both reduce the complexity of the model while increasing the capacity of integrating a large number of parameters, this communication ambitions to propose a new modelling approach, called Extended Object Oriented Bayesian Network (EOOBN). The EOOBN is an underlying mathematical tool which has much more flexibility than classical Bayesian Networks. The main aim of the communication is then to present a methodology dedicated to EOOBN construction. After having introduced the main concepts and described the EOOBN building principles, an industrial application is proposed to illustrate the developments.