Identification For Control Based on Neural Networks: Approximately Linearizable Models
This work addresses control design challenges for engineers dealing with nonlinear systems, offering a method that simplifies stability analysis and controller implementation, though it appears incremental by combining existing neural network identification with linearization techniques.
The paper tackles the problem of designing efficient controllers for nonlinear systems by using neural networks to identify approximately linearizable models, enabling straightforward application of linear control theory for robust controller design and stability analysis, with effectiveness demonstrated on popular benchmarks.
This work presents a control-oriented identification scheme for efficient control design and stability analysis of nonlinear systems. Neural networks are used to identify a discrete-time nonlinear state-space model to approximate time-domain input-output behavior of a nonlinear system. The network is constructed such that the identified model is approximately linearizable by feedback, ensuring that the control law trivially follows from the learning stage. After the identification and quasi-linearization procedures, linear control theory comes at hand to design robust controllers and study stability of the closed-loop system. The effectiveness and interest of the methodology are illustrated throughout the paper on popular benchmarks for system identification.