Automata-Learning

This framework contains functions as follows:

• FnClusterSegs: clusters trace segments
• FnEstODE: estimates ordinary differential equations from the clustered trace segments
• FnLI: estimates linear inequalities for changepoints in the traces
• FnPTA: generates state transitions with guard conditions
• FnGenerateHyst: generates a hybrid automation in Hyst data structure
• FnHyst2State: converts Hyst data structure to Stateflow data structure

Run examples

Run the main function in each example to generate the "automata_learning.mdl" which will be in the "output_slsf_models" folder

Requirements (MATLAB)

• Control System Toolbox
• Mapping Toolbox
• Simulink
• Stateflow
• Symbolic Math Toolbox

References

• 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]