Industrial system working condition identification using operation-adjusted hidden Markov model

In this article, the problem of industrial system working condition identification in the context of complex operation modes is considered. The problem is challenging due to the fact that the system dynamics are significantly affected by the operation modes. Specifically, the condition monitoring signals may behave quite differently for different operation modes. To overcome this difficulty, an operation-adjusted hidden Markov model (HMM) is proposed by combining the operation information into the construction of HMM observation models. Modeling and classification methods using the formulated HMM are provided for system condition identification under variable operation conditions. Using numerical studies and real-world data, it is demonstrated that the proposed method outperforms commonly used machine learning methods by providing more accurate condition identification results.