No abstract

Rock burst is the result of the development and extension of micro-cracks during the loading process of large-scale rock mass in underground space engineering. Dynamic monitoring results by acoustic emission (AE) can accurately perceive the inner fracture evolution of rock mass and effectively warn about its induced disasters early. By contrastive testing the AE parameters in the whole fracture process of the intact and holey rock samples under graded loading, their spatiotemporal evolution rules were analyzed in this paper, and the damage model of rock samples based on AE localization events was established to analyze the relationship between rock damage and loads. The results show that: (1) Under the condition of grading loading, AE parameter increases with the increase of axial stress and show three states, respectively, which are slow-growth, stabilization and rapid increasing; meanwhile, the damage of the sample has a cumulative effect with time. (2) The AE counts and energy are highly correlated with the fracture of the sample that the more severe the damage of the sample, the faster the crack propagation as well as the higher the acoustic emission counts and the energy amplitude. The damage state of granite sample can be accurately judged by two parameters to character the damage evolution process and fracture mechanism. (3) Compared with the intact rock sample, due to the pressure relief effect of the hole, the rock sample containing the hole takes a long time in the compaction stage and with higher load stress level. Although the AE counts and energy were lower in the damage process, the general law of their response during damage and instability process still exists.

Unsupervised domain adaptation‐based fault diagnosis methods have been extensively studied due to their powerful knowledge transferability under different working conditions. Despite their encouraging performance, most of them cannot sufficiently account for the temporal dimension of the vibration signal, resulting in incomplete feature information used in the domain alignment procedure. To alleviate the limitation, we present a self‐supervised domain adaptation fault diagnosis network (SDAFDN), which considers two temporal dependencies to improve the transferability of the learned representations. Specifically, we first design a down‐sampling and interaction network that considers the temporal dependency among subsequences with low temporal resolution in feature space. Then, we combine domain adversarial learning with feature mapping to achieve domain alignment. Finally, we introduced a self‐supervised learning module, which considers the temporal dependency between the past and future temporal segments via classification tasks. Extensive experiments on public Paderborn University and PHM data sets demonstrate the superiority of the proposed SDAFDN and the effectiveness of considering temporal dependencies in domain alignment.

No abstract