Prognosis of wear progress based on regression analysis of condition monitoring parameters

Erkki Jantunen

    Research output: Contribution to journalArticleScientificpeer-review

    6 Citations (Scopus)

    Abstract

    For the maintenance personnel the key questions in every day life are: Is everything working properly and if not should we do something? It is especially important to know when some action should be taken i.e. will the machine in question hold until the next scheduled maintenance or does it not? Considering this from the condition monitoring point of view it is important to get a reliable indication of an upcoming failure so early that the necessary maintenance tasks can be planned well in advance and really perform them when the production machinery is stopped for scheduled maintenance. It is not an easy task to predict from measured parameters how quickly the fault will progress. The paper discusses some possible models for the progress of condition monitoring parameters i.e. how the condition monitoring parameters indicate the development of wear as a function of time. The prediction of the development/increase of these parameters is based on regression analysis techniques. The choice of these models is discussed keeping in mind that for practical purposes they should be simple and fast to use. The models are tested with some very common components which suffer from a type of wear which tends to progress with increasing speed towards the end of the life of the component. The first example is from tool wear monitoring where the life of the tool is very short and the measured values usually follow a certain trend and the second example is from a bearing test where the trend of the measured parameter is not that obvious. In both cases the suggested regression analysis technique works very well and can give prognosis of the further development of the monitored parameter.

    Original languageEnglish
    Pages (from-to)3-15
    Number of pages13
    JournalTribologia
    Volume22
    Issue number4
    Publication statusPublished - 1 Dec 2003
    MoE publication typeA1 Journal article-refereed

      Fingerprint

    Keywords

    • Bearing fault
    • Condition monitoring
    • Diagnosis
    • Monitoring parameters
    • Prognosis
    • Regression analysis
    • Rotating machinery
    • Tool wear
    • Wear progress

    Cite this