This method integrates optimal control within physics-informed neural networks, indicating enhanced efficiency and adaptability in systems governed by PDEs.
Key Points
The proposed framework improves results in optimal control problems governed by partial differential equations, achieving faster computation.
Control physics-informed neural networks learn system states and optimal control signals simultaneously, streamlining the process.
This method embeds optimality conditions into the network architecture, enhancing both learning and performance metrics.
The effectiveness of this approach is demonstrated across various open-loop optimal control problems involving one- and two-dimensional PDEs.
Cite This Study
Barry-Straume et al. (2025) studied this question.