A Bayesian belief approach to quality control of resin transfer molding process

In recent years, there has been a significant global shift towards use of polymer matrix composite materials in a wide range of industries, including aerospace, automotive, marine, and sports, among others. Despite the rapid uptake and widespread adoption of this material technology, there are still technical challenges faced daily by manufactures due to the inherent complexities of different composite processing techniques. This paper aims at establishing a new scheme for better quality control management of resin transfer molding (RTM) processing for thermosetting composites, using a Bayesian belief network (BBN). The data were collected through knowledge engineering among the manufacturing experts with the help of real-life application history. A total of 13 governing factors for the RTM process, which would theoretically have a role on the final product quality, were identified. The sensitivity analysis of the BBN showed that the major contributing factors of the quality are the resin viscosity profile, part design features, resin cure peak temperature, and reinforcement permeability. The conditional probability tables were constructed using a quality index from industrial experts, and the causal relationships captured by the BBN were built using knowledge engineering. It is also shown how the basic BBN model can be further updated by integrating the interaction weights between the attributes that define the product quality.