A Multi-Model Approach to Model-Free Adaptive Predictive Control for Discrete Nonlinear Systems Subject to Measurement Noise

GP-based Offset-free MPC using Disturbance Models and MHE

Data-Driven Hybrid Framework for Dynamic System Control

Nonlinear electrical and electromechanical systems pose significant challenges for observer-based control design. Conventional observer approaches require accurate mathematical models, which often fail under physical irregularities such as sensor noise, measurement delays, and parameter variations. These changes decrease estimation accuracy and degrade control action. This paper proposes a Hybrid ARX– Observer framework to overcome these limitations. It combines the stability of a traditional model-based observer with the flexibility of a data-driven ARX estimator. To guarantee input-to-state stability (ISS), observer gains are computed using a matrix-multiplier technique based on linear matrix inequalities (LMI). The ARX component employs Recursive Least Squares (RLS) adaptation to attenuate measurement noise and capture residual nonlinearities. Both estimates are then fused together using a fusion gain α. This framework is tested on a robotic arm and the results show that the hybrid framework improved the robustness and reduced estimation error up to 4% compared to the conventional observer based estimation, while remaining computationally light compared to fully data-driven alternatives.

Model-free predictive control for PMSM based on a nonlinear autoregressive exogenous model and an adaptive recursive least squares algorithm

This paper proposes a novel nonlinear autoregressive with exogenous input-based adaptive recursive least squares model-free predictive control (NARX-ARLS-MFPC) strategy for permanent magnet synchronous motor (PMSM) drives. The core challenge addressed is the performance degradation of conventional model predictive control (MPC) under inevitable motor parameter mismatches. The proposed method integrates the NARX model with an ARLS algorithm featuring a variable forgetting factor to dynamically track system changes. Comprehensive simulation studies validate the superior robustness of the strategy. Under significant inductance and flux linkage mismatches, the proposed method reduces current total harmonic distortion (THD) by 28.3% and 12.3% compared to conventional finite-control-set model predictive control (FCS-MPC) and a baseline control method, respectively. It maintains stable performance under moderate sensor noise with appropriate tuning. During load transients combined with resistance and inductance mismatches, it achieves THD reductions of 24.1% and 15.7% versus the two benchmark methods, respectively. Statistical analysis under parameter perturbations confirms its overall superior performance across key dynamic and steady-state metrics. The results demonstrate that the synergistic integration of the nonlinear model and adaptive identification effectively suppresses current harmonics caused by model inaccuracies while enhancing dynamic performance.

Data-Driven Koopman-Based Model Predictive Control for a Three-Tank Hydraulic System

This paper presents an applied study of data-driven Model Predictive Control (MPC) based on the Koopman operator framework for a three-tank hydraulic benchmark. The central contribution is a compact, physics-informed lifting strategy: observable functions are chosen directly from Torricelli’s law governing turbulent orifice flow, yielding a seven-dimensional Extended Dynamic Mode Decomposition (EDMD) model that captures the dominant nonlinearities with fewer basis functions than generic dictionaries. The resulting Koopman-MPC replaces the nonconvex optimization of nonlinear MPC (NMPC) with a convex quadratic program, achieving comparable tracking accuracy (0.65 cm vs. 0.64 cm mean error) while reducing the average per-step computation time by 36.7× (50 ms vs. 1807 ms in Python/SciPy). A Moving Horizon Estimator (MHE) operating on the full nonlinear model reconstructs the unmeasured tank level with accuracy comparable to the 2 mm measurement noise floor. These results provide a quantitative benchmark for Koopman-based predictive control on a nonlinear hydraulic system with bidirectional inter-tank coupling and regime-dependent flow transitions.

Koopman-based model predictive control for nonlinear systems with bounded model uncertainty

Tube-based model predictive control for wind power system using adaptive just-in-time learning modeling methodology

The wind power generation process exhibits strong nonlinearity and multiple constraints, making it difficult to establish an accurate global model for model predictive control. In practical applications, model mismatch often leads to a decline in control performance. To address this, a tube-based model predictive control for wind power system using adaptive just-in-time learning modeling methodology is proposed. There are two core innovations: (1) A pre-clustering adaptive just-in-time learning method is adopted to construct a local dynamic model online as the nominal system, which ensures modeling accuracy while significantly reducing computational burden and (2) without explicitly distinguishing between the maximum power point tracking region and the pitch control region, a tube-based model predictive control strategy is developed so that the power tracking error is constrained within a Tube invariant set centered on the nominal system, effectively suppressing the effects of wind speed randomness and model mismatch. Simulation results on a 5-MW wind turbine demonstrate that the proposed strategy can smooth power fluctuations, improve tracking accuracy, and achieve superior robustness and computational efficiency.