Degradation modeling and remaining life prediction of multi-state long-life systems under random environmental influences

Abstract

Abstract For large-scale systems such as bridges, which have long operating lifetimes, the operating states are usually categorized into multiple levels, and they are also subjected to various random environmental influences during operation. However, due to the significant granularity in the categorization of system states, it is difficult to assess the system state transitions influenced by random environmental factors, which compromises the accuracy of remaining life predictions. In this study, we focus on long-life systems with multiple states and investigate the degradation modeling and remaining life prediction considering the impact of random environmental factors. The system degradation process, based on the semi-Markov process and multi-state modeling, was decomposed into states using the sub-exponential approximation method. A state transition probability model considering exponential environmental influences was constructed. Furthermore, based on the developed model for calculating the distribution of sojourn times in multiple states, a reliability and remaining life prediction model for the system was derived. By taking the bridge deck as a case study, the verification and analysis of remaining life prediction for the bridge deck were conducted under the influences of average daily traffic volume and bridge age. The results indicate that both the average daily traffic volume and bridge age have a significant impact on the degradation of the bridge deck. The relative error of the predicted results considering the above effects falls within the range of 1.77%–12.18%.