An effective health indicator based on two dimensional hidden Markov model

Abstract

It is still a great challenge to find an effective degradation indication for health degradation assessment. For that purpose, many computational algorithms have been motivated or suggested. These algorithms commonly use the multiple statistic-based features obtained through calculating the vibration signal to build health indicator which does not consider the internal relevancy among the features. For the purpose of building an effective degradation indicator with pronounced tendency, a novel method is proposed. With the ability of keeping the local structure of data set, Locality preserving projections (LPP) is used to reduce the dimension of the general feature set with the main information remained. Since 2-Dimensional Hidden Markov model (2-D HMM) not only take the relevance between the individuals of fault features into consideration but also capture the global characteristics of the multiple features, 2-D HMM based negative log likelihood probability is built as a new degradation index. The experimental results indicate that the proposed indicator show great abilities to make degradation performance of the bearing and is sensitive to weak defects.