Optimal Control Strategies for Linear Systems

Optimal control of linear systems has evolved into a mature discipline combining rigorous mathematical foundations with powerful computational techniques. At its core lies the formulation of a control law that minimises a given performance index—typically quadratic costs on state deviations and control effort—while ensuring internal stability. Classical methods such as the Linear Quadratic Regulator (LQR) and H∞ control address nominal and worst-case disturbance scenarios respectively, whereas Model Predictive Control (MPC) introduces explicit handling of constraints through on-line optimisation. Recent advances have shifted attention from controller coefficients to the entire closed-loop map, yielding scalable design frameworks that accommodate distributed architectures, robustness to uncertainty and sparsity in actuation. This holistic perspective has facilitated novel parameterisations that render non-convex feasibility regions amenable to convex optimisation, and emerging online schemes that adapt to unmodelled disturbances with provable performance bounds. Together, these strategies underpin applications ranging from networked infrastructure and vehicle dynamics to energy-efficient automation and large-scale process control.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

System Level Synthesis (SLS) has redefined robust and distributed optimal control by lifting the synthesis problem to the closed-loop response level. This framework yields convex characterisations of constrained robust controllers, enables locality in distributed deployments and quantifies performance degradation under model uncertainty, thereby bridging classical robust control with modern statistical inference. A novel convex-parameterisation study has identified new equivalence classes of stabilising controllers beyond the Youla and SLS forms. By exploring four distinct closed-loop map characterisations, this work reveals trade-offs under finite impulse response approximations and establishes numerical robustness criteria for each parameterisation in practical implementations. In robust regret optimal control, a discrete-time uncertain system is regulated against an undisturbed, non-causal benchmark. The resulting controller synthesis enforces a robust performance condition, solvable via DK-iteration, and is shown to outperform both classical designs and non-regret counterparts in examples including single-input single-output plants and vehicle suspension models. This regret framework unifies worst-case and competitive objectives under uncertainty.

Optimal Control Strategies for Linear Systems publication trend

The graph below shows the total number of articles in optimal control strategies for linear systems across all publications each year (not limited to Nature Index journals).

Technical terms

Linear time-invariant (LTI) system: A system whose dynamics are governed by linear equations that do not vary over time.

Linear Quadratic Regulator (LQR): A control law that minimises a quadratic cost function of states and control inputs.

H∞ control: A robust design method that minimises the worst-case energy gain from disturbances to regulated outputs.

Model Predictive Control (MPC): An on-line optimisation technique that computes control moves over a receding horizon while enforcing constraints.

System Level Synthesis (SLS): A design framework parameterising closed-loop responses directly to achieve scalable robust and distributed control.

Youla parameterisation: A classical representation of all internally stabilising controllers via a free transfer-function parameter.

Regret optimal control: A synthesis approach that minimises the performance gap relative to an ideal non-causal controller.

Maximum hands-off control: A sparse control strategy that maximises time intervals over which the control input is exactly zero.

References

1. System level synthesis. Annual Reviews in Control (2019).
2. System-level, input–output and new parameterizations of stabilizing controllers, and their numerical computation. Automatica (2022).
3. Robust regret optimal control. International Journal of Robust and Nonlinear Control (2024).