Frequency Constrained MPC for Efficient Grid Side Operation of Wind Power Conversion Systems
Orcun Karaca, Georgios Darivianakis
TL;DR
The paper addresses grid-side control of wind power converters, where conventional MPC can be computationally burdensome and may struggle to meet grid-code constraints. It proposes an indirect MPC that restricts decision variables to a prescribed frequency content, so the horizon length does not inflate the optimization size; the problem is solved as a $QP$ with linear constraints and followed by a $CB$-$PWM$ modulator. A key contribution is achieving a substantial size reduction by enforcing frequency components on the decision variables, enabling long horizons while keeping the optimization tractable, and leveraging averaged/differenced state representations with integral action to handle grid dynamics and offsets. Case studies demonstrate fast fault response and maintenance of dc-link voltage balance and grid-code compliant current spectra, with switching frequencies around $f_c=750$ Hz producing practical, moderate ripple in the measured signals.
Abstract
Model predictive control (MPC) has proven its applicability in power conversion control with its fast dynamic response to reference changes while ensuring critical system constraints are satisfied. Even then, the computational burden still remains a challenge for many MPC variants. In this regard, this paper formulates an indirect MPC scheme for grid-side wind converters. A quadratic program with linear constraints is solved in a receding horizon fashion with a subsequent PWM modulator. To facilitate its solution within a few hundreds of microseconds, its decision variables (modulating signals) are restricted to a specific frequency content. This approach limits the increase in problem size due to horizon length. In case studies, the proposed MPC exhibits fast response in faults and operates the converter within its safety limits.
