Anomaly detection in aerospace product manufacturing: Initial remarks

Manufacturing companies need to acquire, analyze and share large amounts of information and data to sustain competitive advantage in complex environments. In the context of complex manufacturing, an increasing number of devices, sensors and people are connected to internal networks dramatically changing the ability to generate, communicate, share and access data. Therefore, the data volume has become so large that it cannot be processed using conventional methods. Many companies have dramatically boosted profits and have met consumer demands more proactively, by utilizing automated data collection to feed information into a big data analytics program. In the aerospace manufacturing sector, there is a growing need to consider Big Data solutions to add value to their business services and to optimize their internal production processes. Manufacturing data are an important source of knowledge that can be recorded from different data sources such as sensors and enterprise. The majority of this data are stream processed i.e., they are produced by analytics performed on 'in-motion' data. A real-time predictive analysis can help detecting manufacturing anomalies thus improving the production processes and the quality of product. This paper aims to shortly describe the initial findings of an action research study performed in the aerospace industry pilot of the TOREADOR European project