Accelerating Optimal Experimental Design for Robust Synchronization of Uncertain Kuramoto Oscillator Model Using Machine Learning

Recent advances in objective-based uncertainty quantification (objective-UQ) have shown that such a goal-driven approach for quantifying model is extremely useful real-world problems aim at achieving specific objectives based on complex uncertain systems. Central to this objective-UQ the concept of mean objective cost (MOCU), which provides effective means impact operational goals hand. MOCU especially optimal experimental design (OED) as potential efficacy an (or data acquisition) campaign can be quantified by estimating expected remain after campaign. However, MOCU-based OED tends computationally expensive, limits its practical applicability. In paper, we propose novel machine learning (ML) scheme significantly accelerate computation and expedite design. The main idea use ML efficiently search robust operator under uncertainty, necessary step computing MOCU. We apply proposed ML-based acceleration experiments aimed optimally enhancing control performance Kuramoto oscillator models. Our results show up 154-fold speed improvement without any degradation performance.