| dc.description.abstract |
<p>The primary focus of this thesis is on the modeling and control of an electromechanical
transducer to harvest energy from large structures (e.g. buildings and bridges).
The transducer consists of a back-driven ballscrew coupled to a permanent-magnet synchronous
machine. Developing control algorithms to take full advantage of the unique features
of this type of transducer requires a mechanical model that can</p><p>adequately characterize
the device's intrinsic nonlinear behavior. A new model is proposed that can effectively
capture this behavior. Comparison with experimental results verifies that the model
is accurate over a wide range of operating conditions and that it can be used to correctly
design controllers to maximize power generation. </p><p>In most vibratory energy
harvesting systems the disturbance is most appropriately modeled as a broadband stochastic
process. Optimization of the average power generated from such disturbances is a
feedback control problem, and the controller can be determined by solving a nonstandard
Riccati equation. In this thesis we show that appropriate tuning of passive parameters
in the harvesting system results in a decoupled solution to the Riccati equation and
a corresponding controller that only requires half of the states for feedback. However,
even when the optimal controller requires all of the states for feedback, it is possible
to determine the states that contribute the most to the power generation and optimize
those partial-state feedback gains using a gradient descent method.</p><p>To demonstrate
the energy harvesting capability of the transducer, impedance matching theory is used
to optimize power from a small, base-excited single-degree-of-freedom (SDOF) oscillator.
For this system, both theoretical and experimental investigations are compared and
results are shown to match closely. Finally, statistical linearization is used to
determine the optimal full-state controller and the optimal static admittance for
the experimental SDOF oscillator when it is excited by a stochastic disturbance.</p>
|
|
