Temporal Deep Unfolding for Nonlinear Maximum Hands-off Control

TL;DR

This paper introduces a deep unfolding method to compute sparse control inputs for nonlinear stochastic systems, aiming to stabilize the system while minimizing control effort, demonstrated through numerical experiments.

Contribution

It presents a novel deep unfolding approach for maximum hands-off control in nonlinear stochastic systems, combining neural networks with control optimization.

Findings

Effective stabilization with sparse control inputs

Successful numerical demonstration of the method

Potential for real-time control applications

Abstract

This paper proposes a computational technique based on "deep unfolding" to solving the finite-time maximum hands-off control problem for discrete-time nonlinear stochastic systems. In particular, we seek a sparse control input sequence that stabilizes the system such that the expected value of the square of the final states is small by training a deep neural network. The proposed technique is demonstrated by a numerical experiment.