Machine performance degradation assessment and remaining useful life prediction using proportional hazard model and support vector machine

Abstract

Machine performance degradation assessment and remaining useful life (RUL) prediction are of crucial importance in condition-based maintenance to reduce the maintenance cost and improve the reliability. They provide a potent tool for operators in decision-making by specifying the present machine state and estimating the remaining time. For this ultimate purpose, a threestage method for assessing the machine health degradation and forecasting the RUL is proposed.In the first stage, only the normal operating condition of machine is used to create identification model for recognizing the dynamic system behavior. Degradation index which is used for indicating the machine degradation is subsequently created based on the root mean square of residual errors. These errors are the difference between identification model and behavior of system. In the second stage, the Cox's proportional hazard model is generated to estimate the survival function of the system. In the last stage, support vector machine, which is one of the remarkable machine learning techniques, in association with time-series techniques is utilized to forecast the RUL. The data of low methane compressor acquired from condition monitoring routine is used for validating the proposed method. The result shows that the proposed method could be used as a reliable tool to machine prognostics.

Bearing Remaining Useful Life Prediction Based on a Nonlinear Wiener Process Model

“…Besides, support vector machine (SVM) is also a widely selected technique for prognostics. Tran et al [10] forecasted the RUL through the SVM and time-series methods. However, datadriven methods do not use the useful information among the degradation process, and they are restricted to the quantity and quality of the collected data.…”

Remaining useful life prediction of the ball screw system based on weighted Mahalanobis distance and an exponential model

“…Generally, RUL prediction approaches can be classified into two categories: data-driven techniques, and model-based techniques [16]. For data-driven approaches, the RUL estimation models are established from historical data through machine learning techniques, e.g., artificial neural network [17], support vector machine [18], and neuro-fuzzy systems [19]. These methods need no prior knowledge about the concerned system but require a large amount of historical failure data.…”

The Application of LCS and Information Entropy as a Novel Fusion Algorithm for Degradation Feature Extraction

“…However, features extracted by traditional methods [6] are normally based on the single monitoring signal. Tran et al extracted features by the analysis of the monitoring signal in the time domain [7]. Zhao used the empirical mode decomposition (EMD) in vibration signal analysis and extracted the approximate entropy as the degradation feature [8].…”