Minimax Linear Regulator Problems for Positive Systems (2411.04809v1)

Abstract: Exceptional are the instances where explicit solutions to optimal control problems are obtainable. Of particular interest are the explicit solutions derived for minimax problems, which provide a framework for tackling challenges characterized by adversarial conditions and uncertainties. This work builds on recent discrete-time research, extending it to a multi-disturbance minimax linear framework for linear time-invariant systems in continuous time. Disturbances are considered to be bounded by elementwise linear constraints, along with unconstrained positive disturbances. Dynamic programming theory is applied to derive explicit solutions to the Hamilton-Jacobi-BeLLMan (HJB) equation for both finite and infinite horizons. For the infinite horizon a fixed-point method is proposed to compute the solution of the HJB equation. Moreover, the Linear Regulator (LR) problem is introduced, which, analogous to the Linear-Quadratic Regulator (LQR) problem, can be utilized for the stabilization of positive systems. A linear program formulation for the LR problem is proposed which computes the associated stabilizing controller, it it exists. Additionally necessary and sufficient conditions for minimizing the $l_1$-induced gain of the system are derived and characterized through the disturbance penalty of the cost function of the minimax problem class. We motivate the prospective scalability properties of our framework with a large-scale water management network.