 The proposed approach to monitoring wind turbine systems features a built-in visualization tool that provides greater interpretability and trust in the model outcomes. This enables faster detection of abnormal conditions and improved understanding of the system's behavior. The approach was tested using two case studies, demonstrating its effectiveness in detecting abnormal conditions up to 11 days ahead of time compared to other approaches. This article was authored by a Dayton-Olivera Filio, Raiadza-Muri, Philippe Cameron, and others.