Auto-body assembly process fault diagnosis based on a dynamic variation modeling approach

This paper aims to provide a new dynamic modeling approach for root cause detection of the auto-body assembly variation.

The dynamic characteristics, such as fixture element wear and quality of incoming parts, are considered in assembly variation modeling with the dynamic Bayesian network. Based on the network structure mapping, the parameter learning of different types of nodes is conducted by integrating process knowledge and Monte Carlo simulation. The inference was that both the measurement data and maintenance actions are evidence for the improvement of diagnosis accuracy.

The proposed assembly variation model which has incorporated dynamic manufacturing features could be used to detect multiple process faults effectively.

A dynamic variation modeling method is proposed. This method could be used to provide more accurate diagnosis results and preventive maintenance guidelines for the assembly process.