Port-metriplectic Neural Networks for Coupled Physical Phenomena
Please login to view abstract download link
We generalize the port-Hamiltonian formalism for machine learning of coupled physical phenomena. The developed method satisfies by construction the principles of thermodynamics in the learned physics (conservation of energy, non-negative entropy production). The port-Hamiltonian formalism is modified so as to achieve a port-metriplectic framework. We show that the constructed networks are able to learn the physics of systems made of distinct constituents. This alleviates the burden associated to the experimental characterization and posterior learning process of this kind of systems. Predictions can be done, however, at the scale of the complete system. Examples are shown on the performance of the proposed technique.
