Multi-Machine Gaussian Topic Modeling for Predictive Maintenance

In this paper, we propose a coherent framework for multi-machine analysis, using group clustering model, which can be utilized predictive maintenance (PdM). The benefits from the repetitive structure posed by multiple machines and enables assessment of health condition, degradation modeling comparison machines. It is based on hierarchical probabilistic denoted Gaussian topic model (GTM), where cluster patterns are shared over therefore it allows one to directly obtain proportions This then used as basis cross between identified similarities differences lead important insights about their behaviors. aggregation data streams predefined set features extracted time window. Moreover, contains schema takes uncertainty assignments into account specify desirable degree reliability assignments. By simulation example, highlight how in order inherent variations such Furthermore, comparative study with commonly mixture (GMM) demonstrates that GTM able identify while GMM fails. Such result consequence level being modeled absent GMM. Hence, trained view not available miss important, possibly even key patterns. Therefore, argue more advanced models, like GTM, interpreting understanding behavior across ultimately obtaining efficient reliable PdM systems.