Application of the Bayesian Network to Machine breakdowns using Witness Simulation

 Abstract—This paper explores the use of Bayesian network modeling of machine breakdowns within a cement manufacturing plant. The Bayesian network modeling is introduced using Hugin software and then implemented into a Witness Simulation model using historical data, expert knowledge and opinions. The models simulate 3 parameters of the machine based on life consumption and usage of each parameter, the developed Witness model produces a probability failure rate based on these parameter usages. The failure probability developed by Witness is implemented using the Chain Rule; this is compared with the probability for failure based on the Bayesian network model. This enables the user to see the probability of failure developed by the implementation of the chain rule and Witness based on the parameter usage, on a live streaming fashion as the model is running. This can be used as a decision making tool for management to consider machine maintenance affectively based on parameter usage.