Abstract: This study proposes a data-driven framework for dynamic performance prediction and system optimization, leveraging multi-source historical data and advanced machine learning techniques. The ...