Least Squares Support Vector Machines for performance degradation modeling of, CNC equipments

Abstract

Due to the applications of high technology, a catastrophic failure of Computerized numerical control (CNC) machine tool rarely happens at normal operation conditions. Traditional reliability assessment methods based on time-to-failure distributions have the difficulty to estimate the reliability precisely. This paper presents a novel reliability assessment methodology that estimates the reliability of machine tool with machining performance degradation while only small samples are required. To analyze a performance degradation process on the machine tool, Least Squares Support Vector Machines (LS-SVM) is introduced. Parameter optimization and a two-stage searching method of the parameter scope are proposed to improve the LS-SVM regression performance. A performance degradation model based on LS-SVM is built and a machining performance degradation experiment on OTM650 machine tool is used to study the feasibility of the developed technique. The results show that the proposed reliability assessment methodology is effective.