Visual Analytics for Production Line Failure Detection

here are several situations where a company has to analyze data gathered from its production line (machines, raw materials, operations, warehouses, ...) to identify the cause of the situation and to react as soon as possible. One of these situations is the detection of production failures after a faulty product has been detected. Failure detection has always been a critical issue in the industrial sector, mainly because measures should be taken both to prevent them happening again, and to fix already finished or sold products. Access to production line data is required to know what has happened. Most companies store this data, at least by using an ERP software. However, it is complicated and human resource intensive to detect a failure just by looking to the raw data provided by an ERP.

This paper focuses on this situation, supporting employees visualizing and navigating production line data. An application transforming production line data into a graph structure, and applying Visual Analytics and Big Data approaches, has been designed and developed. This application has been integrated with an existing ERP, Izaro from Group Zucchetti. Results have been positively evaluated, as they greatly improve existing tools to analyze and navigate production line data from the ERP.