Data mining for industrial system identification: A turning process

Abstract

The modeling of an industrial process is always a challenging issue and has a significant effect on the performance of the industry. In this study, one of the most important industrial processes, a turning process, is considered as a black box system. Since it is also a dynamic system, i.e., its characteristics changing over time, the system identification method has been applied on the measurement data in order to obtain an empirical model for explaining a system output, surface roughness. The inputs of the system are feed rate, cutting speed and tool nose radius. According to the study, three non-parametric models, Box-Jenkins, autoregressive moving average with exogenous inputs (ARMAX) and output error (OE), are recommended to be used to construct mathematical models based on data mining available from the manufacturing process. These system identification models are appropriate to model the dynamic turning process since they have the capability to construct both dynamic and noise parameters separately.