Engineering driven cause-effect modeling and statistical analysis for multi-operational machining process diagnosis

Process fault identification for product quality improvement is a critical issue in both design and manufacturing, especially for multistage manufacturing processes. In this paper, an integrated approach is proposed to develop cause-effect models from engineering knowledge and to conduct associated statistical analysis of the measurement data. First, a cause-effect diagram and predicted symptom vectors (PSV) are formulated to recognize the cause-effect relationship between process variables and product qualities. Then factor analysis and factor rotating technique are employed to extract the symptoms reflected from measurement data. Finally the potential process faults are identified by comparing predicted symptoms and extracted symptoms. A case study is conducted to demonstrate the effectiveness of the proposed methodology.