prognostics-health-management

This paper proposes a framework for remaining useful life prediction using multimodal large language models grounded by time-series retrieval, demonstrating that retrieval-based methods improve accuracy and stability over non-retrieval baselines.