In order to improve mission-perform capability and reduce maintenance costs of equipment, the research of avionics prognostics is carried out. Support Vector Machine (SVM) is a kind of machine learning methods developed from statistics learning theory, which can well resolve practical problems of many previous learning methods such as small samples, nonlinear, over learning, high dimension, local minimum points, and thus plays an important role in avionics prognostics. The traditional method is difficult to achieve good forecasting results for the unequal interval time series. This paper carried out the research of small samples and unequal interval time series avionics prognostics using the SVM regression (SVMR) model, and gave out the results of comparing the forecasting results with the regression analysis based on least squares (LS) and artificial neural network (ANN), which indicated that the method of SVM has a higher forecasting accuracy than the other two ways based on the avionics and can satisfy the requirements of avionics prognostics. The SVMR model has some theoretical value and practical significance for the avionics prognostics.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Research of small samples avionics prognostics based on Support Vector Machine


    Contributors:


    Publication date :

    2011-05-01


    Size :

    870710 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Why prognostics for avionics?

    Hecht, H. | IEEE | 2006



    Prognostics for Electronics Components of Avionics Systems

    Celaya, Jose R. / Saha, Bhaskar / Wysocki, Philip F. et al. | NTRS | 2009


    Avionics health management: searching for the prognostics grail

    Kirkland, L.V. / Pombo, T. / Nelson, K. et al. | IEEE | 2004


    11.0203 Avionics Health Management: Searching for the Prognostics Grail

    IEEE | British Library Conference Proceedings | 2004