Minimum-Entropy Optimal Control of Electromechanical Linkages for Energy Harvesting

This work considers optimal mechanical-electrical power conversion across rigid linkages equipped with current-controlled actuators. A novel cost function derived from a generalization of the Second Law of Thermodynamics is adopted from our previous work, where cycle-averaged energies are interpreted as generalized temperatures. A cost function based on generalized entropy generation is used to formulate an optimal control problem yielding a decoupled velocity feedback controller. Suboptimal gains are found, which are independent of both the excitation characteristics and the mechanical subsystem dynamics, and yield closed-loop stability. The effectiveness and simplicity of the resulting controller is demonstrated by a Monte Carlo simulation study, where random episodes of unknown, periodic forcing are applied under the proposed controller and compared with a maximum-efficiency controller. Results show that the proposed controller offers a higher statistical expectation for the average harvested power.