Multi-objective optimal control of vibratory energy harvesting systems

Abstract

This article examines the use of actively controlled electronics to maximize the energy harvested from a stationary stochastic disturbance. In prior work by the authors, it has been shown that when the harvester dynamics are linear and the transmission losses in the electronics are resistive, the optimal feedback controller is the solution to a nonstandard linear-quadratic-Gaussian optimal control problem. This article augments the theory in the following three distinct ways: (a) It illustrates how to use linear matrix inequalities to balance the objective of energy harvesting against other response control objectives (such as minimum requirements on closed-loop damping and maximum levels of voltage response), in the synthesis of the optimal feedback law; (b) it establishes a more realistic characterization of the transmission losses in the actively controlled power electronics used to regulate the extraction of power; and (c) it illustrates how the optimal control theory for resistive loss models can be extended to accommodate the more realistic loss models. The theory is illustrated in the context of a piezoelectric energy harvesting model, and an example is used to illustrate that the theory can be used to simultaneously optimize the feedback law, together with the switching frequency and storage bus voltage of the power electronics.