Towards a Bayesian Network methodology to improve maintenance of complex semiconductor systems

International audienceThe evolution of microelectronics is characterized by an intense competitive environment between manufacturers; so, the semiconductor product must be done with reduced cycle time, improved yield and enhanced equipment effectiveness. In this paper, a general methodology to improve maintenance of complex systems is proposed, the objective is to develop equipment behavior model. A literature review about maintenance and risk analysis methods is presented. Then, giving the system characteristics the study is oriented towards Bayesian Networks methods. The increasing use of these techniques is due to the benefits that provide. Following the proposed methodology, prototypes of data extraction and processing are designed. They allow identify the real Failure Modes which will instantiate the first Bayesian models