DEFECT ROOT CAUSE ANALYSIS IN LARGE SCALE MANUFACTURING SYSTEM

Due to increasing level of complexity and huge number of operational aspects in manufacturing systems, quality control is of crucial importance. In order to prevent production losses due to defective products, some measures must be taken. In this paper a probabilistic approach based on large scale baysian network (LSBN) is presented that enables an effective defect root cause analysis (DRCA) for quality improvement in product of large scale manufacturing systems. The proposed approach is capable in finding the most probable root cause in defective product (several defects detected simultaneously). The proposed approach is based on Baysian inference for reasoning under uncertainties and wide spectrum of system sizes. It is model based and accumulates the system knowledge within the problem domain, which data is gathered from experience, various manufacturing data source and defect related knowledge. The system learning can be supervised by user feedback on the actual root cause. The general DRCA is applied to the defective vehicle body surfaces in paint shop of an automotive assembly plant.