The paper introduces an approach for system identification using regularization, demonstrating enhanced stability and performance compared to traditional methods. It covers linear and nonlinear models, emphasizing the importance of balancing model complexity and fit quality.
The study explores the application of recurrent neural networks in system identification, highlighting the benefits of quasi-Newton methods over gradient descent approaches. It discusses the use of ℓ1-regularization to induce sparsity patterns in models, enhancing interpretability and control design.
Additionally, the paper presents results from experiments on synthetic and industrial robot datasets, showcasing the effectiveness of the proposed method. The comparison with existing tools like N4SID demonstrates superior performance in terms of stability and accuracy.
Overall, the research provides valuable insights into efficient system identification techniques that can benefit various control applications.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Alberto Bemp... at arxiv.org 03-07-2024
https://arxiv.org/pdf/2403.03827.pdfDeeper Inquiries