.png){kind=icon}
.png){kind=icon}
Return
Optimization-Driven Intelligent Adaptive Robust Control of Nonlinear Systems with Input Saturation
.png){kind=icon}
PRE
AI
M
H
O
DOI：10.1007/s40313-026-01251-3.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This study addresses the design of a robust intelligent-adaptive controller for nonlinear systems affected by parametric uncertainties, external disturbances, and input saturation constraints. The intelligent-adaptive component utilizes a hybrid neural network that integrates Radial Basis Function Neural Network (RBFNN) and Multilayer Perceptron (MLP) elements, which are used to cancel uncertain nonlinearities. This structure leverages the complementary advantages of both architectures to ensure that the combined output remains within actuation constraints. A projection algorithm is employed to guarantee bounded estimation of the RBFNN weights. The robust component synergistically combines H-infinity and guaranteed cost control to systematically address system parameter uncertainties and external disturbances. Additionally, the proposed controller takes advantage of the model reference technique to achieve the desired transient behavior by generating a reference signal that prescribes the target system response. The High-gain Observer (HGO) is employed for state estimation, while the Artificial Protozoa Optimizer (APO) algorithm is used to compute the optimal gains of the guaranteed-cost H-infinity controller that minimizes the ultimate error bound. The synthesis is carried out via Linear Matrix Inequalities (LMIs). This formulation explicitly incorporates input saturation, addresses parameter uncertainties, and ensures H-infinity disturbance attenuation as well as ultimate boundedness of tracking error trajectories within a specified region of stability. The APO algorithm is further utilized to optimize the hybrid neural network's hyperparameters, the estimated RBFNN weight norm bound, and the adaptation rate matrix. This optimization is achieved by minimizing an objective function defined by the Integral of Time-weighted Norm of Estimated Error (ITNE). Finally, a numerical example and a practical example based on a magnetic levitation system (MLS) are presented to validate the proposed method. The results demonstrate that the controller attains accurate trajectory tracking of the reference model while accommodating system nonlinearities and parametric uncertainties. It effectively rejects external disturbances and maintains stability under input saturation.
Keywords:
Robust control
H-infinity control
Adaptive control
Guaranteed-cost control
Intelligent control
Artificial protozoa optimizer
Input saturation
Linear matrix inequality
Journal
J
IF:
1.3
Papers:
92
Citations:
1.1K
Organization
Cited Papers