Analyzing and forecasting the reliability for repairable systems using the time series decomposition method

Purpose – The purpose of this study is to propose the time series decomposition approach to analyze and predict the failure data of the repairable systems. Design/methodology/approach – This study employs NHPP to model the failure data. Initially, Nelson's graph method is employed to estimate the mean number of repairs and the MCRF value for the repairable system. Second, the time series decomposition approach is employed to predict the mean number of repairs and MCRF values. Findings – The proposed method can analyze and predict the reliability for repairable systems. It can analyze the combined effect of trend‐cycle components and the seasonal component of the failure data. Research limitations/implications – This study only adopts simulated data to verify the proposed method. Future research may use other real products' failure data to verify the proposed method. The proposed method is superior to ARIMA and neural network model prediction techniques in the reliability of repairable systems. Practical implications – Results in this study can provide a valuable reference for engineers when constructing quality feedback systems for assessing current quality conditions, providing logistical support, correcting product design, facilitating optimal component‐replacement and maintenance strategies, and ensuring that products meet quality requirements. Originality/value – The time series decomposition approach was used to model and analyze software aging and software failure in 2007. However, the time series decomposition approach was rarely used for modeling and analyzing the failure data for repairable systems. This study proposes the time series decomposition approach to analyze and predict the failure data of the repairable systems and the proposed method is better than the ARIMA model and neural networks in predictive accuracy.