The goal of this project is to create a data-driven optimization framework to support maintenance planning for complex high-tech systems, which are made up of many interconnected components. The interactions between these components make it difficult to predict how the system will behave. Our focus is on designing a predictive maintenance strategy that can identify patterns indicating potential equipment failures, allowing repairs to be made before a breakdown occurs. By incorporating monitoring information, we aim to predict when a maintenance action is necessary, thus preventing costly downtime. Our approach takes a dynamic view, allowing the maintenance plan to be adjusted as more information becomes available. Ultimately, we want to develop a cost-minimization strategy with robust performance, while having as few assumptions about uncertain parameters such as degradation processes and component lifetime as possible. To achieve this, we base our methodology on Robust Optimization.