Adaptive Neural Network Control for Pneumatic Artificial Muscles

The present article is based on the widely accepted three-element model of pneumatic artificial muscles. By leveraging the input-hidden-output structure of neural networks and employing unknown function estimation techniques within the model, it facilitates parameter estimation. Subsequently, a sliding manifold control-based controller is devised to achieve precise tracking control, ensuring asymptotic convergence of tracking error to zero. Furthermore, stability analysis and proof are conducted using Lyapunov methods, ultimately validating its feasibility and effectiveness through simulation experiments.