arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2009.05686math.OCcs.LGcs.SYeess.SY
收藏

QRnet:基于LQR增强神经网络的最优调节器设计

QRnet: optimal regulator design with LQR-augmented neural networks

Naval Postgraduate School(美国海军研究生院)

Tenavi Nakamura-Zimmerer, Qi Gong, Wei Kang

摘要:

本文提出了一种新的计算方法来设计高维非线性系统的最优调节器。所提出的方法利用物理信息机器学习来解决最优反馈控制中出现的高维Hamilton-Jacobi-Bellman方程。具体地说,我们增加线性二次调节器与神经网络来处理非线性。我们训练增强模型的数据生成而不离散的状态空间,使应用程序的高维问题。我们使用所提出的方法来设计一个不稳定的Burgers方程的候选最优调节器,并通过这个例子,证明改进的鲁棒性和准确性相比,现有的神经网络配方。

英文摘要:

In this paper we propose a new computational method for designing optimal regulators for high-dimensional nonlinear systems. The proposed approach leverages physics-informed machine learning to solve high-dimensional Hamilton-Jacobi-Bellman equations arising in optimal feedback control. Concretely, we augment linear quadratic regulators with neural networks to handle nonlinearities. We train the augmented models on data generated without discretizing the state space, enabling application to high-dimensional problems. We use the proposed method to design a candidate optimal regulator for an unstable Burgers' equation, and through this example, demonstrate improved robustness and accuracy compared to existing neural network formulations.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑