Data quality evaluation for smart multi-sensor process monitoring using data fusion and machine learning algorithms

Condition monitoring and control of manufacturing processes are among the key issues for the development of smart factories. The employment of multiple sensor systems for manufacturing process monitoring typically involves the detection and collection of a huge volume of heterogeneous data acquired by sensors of different nature. Smart monitoring implementation requires valuable sensorial data from multiple sensors to achieve high prediction performance in the decision making phase. Data fusion techniques can be employed to combine multiple data sources in order to generate more accurate and reliable information for decision making aimed at achieving the automatic identification of machine, tool, and part failures. This paper focuses on sensorial data quality evaluation using data fusion techniques and machine learning algorithms for intelligent multiple sensor monitoring of manufacturing processes. For this purpose, a sensorial data set derived from an experimental campaign of Inconel 718 machining was subjected to different fusion techniques in order to find the best data combination for accurate classification and recognition purposes.