Adaptive Neural Sliding-Mode Control for Fuzzy Singularly Perturbed Systems: Sojourn-Probability-Based Stochastic...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• System Dynamics

Background:

• Fuzzy singularly perturbed systems are susceptible to deception attacks and stochastic communication protocols (SCP).
• Traditional SCP characterizations using transition probability may not fully capture system stochasticity.
• Imperfect premise matching in fuzzy controllers can degrade performance.

Purpose of the Study:

• To develop an adaptive neural network (NN) sliding-mode control (SMC) strategy for fuzzy singularly perturbed systems.
• To enhance robustness against unrestricted deception attacks and stochastic communication protocols.
• To improve the characterization of stochastic behavior using sojourn-probability-based SCP.

Main Methods:

• An adaptive NN-SMC strategy is proposed, integrating fuzzy rules and singular perturbation parameters.
• A sojourn-probability-based SCP is established for precise stochastic characterization.
• An NN-based technique is employed to estimate and mitigate deception attack impacts.

Main Results:

• The controller design addresses imperfect premise matching challenges.
• Exponential ultimate boundedness in the mean square sense is guaranteed for the closed-loop system.
• The specified sliding surface reachability is ensured.

Conclusions:

• The proposed adaptive NN-SMC framework offers a robust solution for fuzzy singularly perturbed systems.
• The strategy effectively handles deception attacks and complex stochastic communication protocols.
• Validation through examples confirms the practical applicability and resilience of the control approach.