Abstract
Identifying and selecting optimal prognostic health indicators in the context of predictive maintenance is essential to obtain a good model and make accurate predictions. Several metrics have been proposed in the past decade to quantify the relevance of those prognostic parameters. Other works have used the well-known minimum redundancy maximum relevance (mRMR) algorithm to select features that are both relevant and non-redundant. However, the relevance criterion is based on labelled machine malfunctions which are not always available in real life scenarios. In this paper, we develop a prognostic mRMR feature selection, an adaptation of the conventional mRMR algorithm, to a situation where class labels are a priori unknown, which we call unsupervised feature selection. In addition, this paper proposes new metrics for computing the relevance and compares different methods to estimate redundancy between features. We show that using unsupervised feature selection as well as adapting relevance metrics with the dynamic time warping algorithm help increase the effectiveness of the selection of health indicators for a rotating machine case study.
Highlights
Predictive maintenance, or condition-based maintenance, consists of recommending maintenance decisions based on the information collected through condition monitoring, usually in the form of time series
In a recent survey (Lei et al, 2018), the authors outline a systematic framework for predictive maintenance based on four steps: (1) Data acquisition: capturing and storing the data coming from the different sensors, (2) Health indicators’ construction: finding features that represent the health’s evolution of the monitored machine, (3) Health states’ division: dividing the machine’s lifetime into several health states and (4) RUL prediction: estimating the time remaining before the machine needs to be replaced
We need to compare the maximum point on each curve while for the former option, we can either compare the sum of differences (SD) between scores corresponding to the same number of features, or the number of times one curve is above the other (N D), based on the 10 scores computed for the increasing feature set size
Summary
Predictive maintenance, or condition-based maintenance, consists of recommending maintenance decisions based on the information collected through condition monitoring, usually in the form of time series. One can use the minimum redundancy maximum relevance (mRMR) algorithm (Peng et al, 2005) that takes into account both the relevance and the redundancy between features via the mutual information criteria Several authors applied this mRMR approach with known class labels to select features in the predictive maintenance context, namely e.g. Y. While supervised feature selection is the way to go if labelled machine malfunctions are available, it is usually not the case for most real-life applications Another possibility would be to use the time before failure as class labels for a regression. The main purpose of this paper is to propose a feature selection approach for predictive maintenance that considers both the relevance and redundancy between features without the need for class labels.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call