A novel feature selection algorithm for high-dimensional condition monitoring data

Kui Zhang, Andrew Ball, Yuhua Li, Fengshou Gu

    Research output: Contribution to journalArticle

    Abstract

    The technique of machinery condition monitoring has been greatly enhanced over recent years with the application of many effective classifiers. However, these classification methods suffer from the ‘curse of dimensionality’ when applied to high-dimensional condition monitoring data. Actually, many classification algorithms are simply intractable when the number of features in the data is sufficiently large. In order to solve the problem, engineers have to resort to complicated feature extraction methods and other statistical theories to reduce the data dimensionality. However, features extracted using these methods lose their original engineering meaning and become obscure for engineers. In this study, a novel feature selection algorithm is presented to help to identify machinery condition quickly, based only on frequency spectrum data and without considering any complicated feature extraction methods. This brings many significant benefits: it can not only help engineers out of difficulties with complicated feature extraction methods, but it also gives a clear insight into the real condition of machinery without any loss of engineering meaning from the features used. The algorithm forms a functional extension of a relevance vector machine to feature selection, with a fast step-variable sequential backward search algorithm to search relevant scale parameters within a kernel function. Two case studies on different levels of condition monitoring are conducted to demonstrate the potential of applying the algorithm to high-dimensional engineering data.
    LanguageEnglish
    Pages33-43
    JournalInternational Journal of Condition Monitoring
    Volume1
    Issue number1
    Publication statusPublished - 1 Jun 2011

    Fingerprint

    Condition monitoring
    Feature extraction
    Machinery
    Engineers
    Classifiers

    Cite this

    Zhang, Kui ; Ball, Andrew ; Li, Yuhua ; Gu, Fengshou. / A novel feature selection algorithm for high-dimensional condition monitoring data. In: International Journal of Condition Monitoring. 2011 ; Vol. 1, No. 1. pp. 33-43.
    @article{2e3cb879ccb2403f8355b8367d48ccdb,
    title = "A novel feature selection algorithm for high-dimensional condition monitoring data",
    abstract = "The technique of machinery condition monitoring has been greatly enhanced over recent years with the application of many effective classifiers. However, these classification methods suffer from the ‘curse of dimensionality’ when applied to high-dimensional condition monitoring data. Actually, many classification algorithms are simply intractable when the number of features in the data is sufficiently large. In order to solve the problem, engineers have to resort to complicated feature extraction methods and other statistical theories to reduce the data dimensionality. However, features extracted using these methods lose their original engineering meaning and become obscure for engineers. In this study, a novel feature selection algorithm is presented to help to identify machinery condition quickly, based only on frequency spectrum data and without considering any complicated feature extraction methods. This brings many significant benefits: it can not only help engineers out of difficulties with complicated feature extraction methods, but it also gives a clear insight into the real condition of machinery without any loss of engineering meaning from the features used. The algorithm forms a functional extension of a relevance vector machine to feature selection, with a fast step-variable sequential backward search algorithm to search relevant scale parameters within a kernel function. Two case studies on different levels of condition monitoring are conducted to demonstrate the potential of applying the algorithm to high-dimensional engineering data.",
    author = "Kui Zhang and Andrew Ball and Yuhua Li and Fengshou Gu",
    year = "2011",
    month = "6",
    day = "1",
    language = "English",
    volume = "1",
    pages = "33--43",
    journal = "International Journal of Condition Monitoring",
    issn = "2047-6426",
    number = "1",

    }

    A novel feature selection algorithm for high-dimensional condition monitoring data. / Zhang, Kui; Ball, Andrew; Li, Yuhua; Gu, Fengshou.

    In: International Journal of Condition Monitoring, Vol. 1, No. 1, 01.06.2011, p. 33-43.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - A novel feature selection algorithm for high-dimensional condition monitoring data

    AU - Zhang, Kui

    AU - Ball, Andrew

    AU - Li, Yuhua

    AU - Gu, Fengshou

    PY - 2011/6/1

    Y1 - 2011/6/1

    N2 - The technique of machinery condition monitoring has been greatly enhanced over recent years with the application of many effective classifiers. However, these classification methods suffer from the ‘curse of dimensionality’ when applied to high-dimensional condition monitoring data. Actually, many classification algorithms are simply intractable when the number of features in the data is sufficiently large. In order to solve the problem, engineers have to resort to complicated feature extraction methods and other statistical theories to reduce the data dimensionality. However, features extracted using these methods lose their original engineering meaning and become obscure for engineers. In this study, a novel feature selection algorithm is presented to help to identify machinery condition quickly, based only on frequency spectrum data and without considering any complicated feature extraction methods. This brings many significant benefits: it can not only help engineers out of difficulties with complicated feature extraction methods, but it also gives a clear insight into the real condition of machinery without any loss of engineering meaning from the features used. The algorithm forms a functional extension of a relevance vector machine to feature selection, with a fast step-variable sequential backward search algorithm to search relevant scale parameters within a kernel function. Two case studies on different levels of condition monitoring are conducted to demonstrate the potential of applying the algorithm to high-dimensional engineering data.

    AB - The technique of machinery condition monitoring has been greatly enhanced over recent years with the application of many effective classifiers. However, these classification methods suffer from the ‘curse of dimensionality’ when applied to high-dimensional condition monitoring data. Actually, many classification algorithms are simply intractable when the number of features in the data is sufficiently large. In order to solve the problem, engineers have to resort to complicated feature extraction methods and other statistical theories to reduce the data dimensionality. However, features extracted using these methods lose their original engineering meaning and become obscure for engineers. In this study, a novel feature selection algorithm is presented to help to identify machinery condition quickly, based only on frequency spectrum data and without considering any complicated feature extraction methods. This brings many significant benefits: it can not only help engineers out of difficulties with complicated feature extraction methods, but it also gives a clear insight into the real condition of machinery without any loss of engineering meaning from the features used. The algorithm forms a functional extension of a relevance vector machine to feature selection, with a fast step-variable sequential backward search algorithm to search relevant scale parameters within a kernel function. Two case studies on different levels of condition monitoring are conducted to demonstrate the potential of applying the algorithm to high-dimensional engineering data.

    M3 - Article

    VL - 1

    SP - 33

    EP - 43

    JO - International Journal of Condition Monitoring

    T2 - International Journal of Condition Monitoring

    JF - International Journal of Condition Monitoring

    SN - 2047-6426

    IS - 1

    ER -