The framework is an extension of hautlearn. NSAL learns hybrid automata from traces with linear and nonlinear dynamics as well as neural ordinary differencial equations (neural ODEs).
Run the main
function in each example to generate the automata models.
- Control System Toolbox
- Mapping Toolbox
- Robust Control Toolbox
- Simulink
- Stateflow
- Symbolic Math Toolbox
- Xiaodong Yang, Omar Ali Beg, Matthew Kenigsberg, and Taylor T. Johnson, "A Framework for Identification and Validation of Affine Hybrid Automata from Input-Output Traces". ACM Trans. Cyber-Phys. Syst. 6, 2, Article 13 (April 2022), 24 pages. [https://doi.org/10.1145/3470455]