Diagnosis of machine tools using Knowledge Extraction and data analysis

Due to the continuous advancement in sensors and information technologies, machining processes are easily monitored and important information are gathered from the sensors’ readings. Valuable information can thus be... [ view full abstract ]

Due to the continuous advancement in sensors and information technologies, machining processes are easily monitored and important information are gathered from the sensors’ readings. Valuable information can thus be extracted and exploited in order to gain more understanding of the process’ physics. Although the mathematical representation of these physics may be complex due to the interactions of many variables, new knowledge extraction techniques offer a promising way of understanding these physics. This paper shows how to manipulate and exploit the available sensors’ data in order to develop practical solutions to the problems of machine tool diagnosis and replacement by mitigating their negative effects on the productivity. We develop conditional based maintenance strategies based on the exploitation of sensors’ data. The results show that condition based techniques predict the machine tools failures and faults correctly and with high accuracy.