Distributed Performance Boosting through Neural Control for Plug-and-Play Networks
Unconstrained neural controllers improve plug-and-play network performance while preserving a guaranteed gain bound.
The paper proposes distributed neural performance-boosting controllers for plug-and-play networked systems whose baseline controllers already preserve stability. Using internal model control, stability reduces to an ell-2 gain bound that free parametrizations enforce for every trainable parameter, so unconstrained gradient training needs no projections. Neighbor-based bounds update only affected subsystems after a plug-and-play change, while a network-norm bound can be less conservative but needs network-wide information. A DC-microgrid case study reports better voltage transients under topology changes and load variations without losing nominal closed-loop stability.
- Neural IMC layers keep a prescribed network gain bound at every training step.
- Neighbor-based and network-norm parametrizations trade locality against conservatism.
- A DC-microgrid study improves voltage transients under topology and load changes.
From the original article
1 min read at arxiv.orgWe address the problem of improving the performance of pre-stabilized networked systems subject to plug-and-play (PnP) operations, namely additions or removals of subsystems. While distributed baseline controllers can preserve stability after a PnP transition, augmenting them with expressive neural policies can invalidate the formal stability guarantees.
By Danilo Saccani · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.