A visual analytics with evidential inference for big data: case study of chemical vapor deposition in solar company

Facing the frontier challenges of big data, the alarm analysis of sensors becomes more significant in manufacturing industries. However, many relevant and irrelevant alarms from unpredictable occurrences, unlimited sensors, unknown values, undetermined causes, and uncertain relevance impose their information useless. This paper proposes a big data analytics to solve damage cause with alarms. It constructs granules to reduce data size and explores a baseline to make inference. For illustrating the analytical technique, a case study of chemical vapor deposition within a solar company is presented. Its results provide inferential knowledge for the cause of damage and low performance. The contribution of this paper lies in integrating interdisciplinary techniques and fulfilling analytics with evidential inference.