Machine condition prognosis based on sequential Monte Carlo method

Machine condition prognosis is an important part of the decision-making in condition-based maintenance. By predicting the degradation of working conditions of machinery, it can organize a predictive maintenance program and prevent production loss. For complex systems, the trending data of the performance degradation is nonlinear over time known as a time series. This paper proposes a prognosis algorithm applied in a real dynamic system. Sequential Monte Carlo method, also known as a particle filter, can be used in nonlinear systems without any assumption of linearity. It is based on the sequential important sampling and resampling algorithm, which represents the posterior probability density function by a set of randomly drawn samples (called particles) and their associated weights. The prediction estimations are computed based on those samples and their weights. The real trending data of low methane compressors acquired from condition monitoring routines is employed for evaluating the proposed method. The results show that the proposed method offers a potential to predict the trending data in real systems of machine condition prognosis.