Hybrid Zonotope-Based Backward Reachability Analysis for Neural Feedback Systems With Nonlinear Plant Models
- Resource Type
- Working Paper
- Authors
- Zhang, Hang; Zhang, Yuhao; Xu, Xiangru
- Source
- Subject
- Mathematics - Optimization and Control
Electrical Engineering and Systems Science - Systems and Control
- Language
The increasing prevalence of neural networks in safety-critical control systems underscores the imperative need for rigorous methods to ensure the reliability and safety of these systems. This work introduces a novel approach employing hybrid zonotopes to compute the over-approximation of backward reachable sets for neural feedback systems with nonlinear plant models and general activation functions. Closed-form expressions as hybrid zonotopes are provided for the over-approximated backward reachable sets, and a refinement procedure is proposed to alleviate the potential conservatism of the approximation. Two numerical examples are provided to illustrate the effectiveness of the proposed approach.
Comment: Accepted by IEEE American Control Conference 2024, 9 pages, 4 figures