Constrained linear state signal shaping model predictive control for harmonic compensation in power systems

For signal shaping problems, in contrast to reference following problems, a possibility is to make use of the signal's shape as a control objective, described by a difference equation. In order to do this, a shape class is defined, giving a measure of how close to a specific shape a signal is. The shape class defines the weighting matrices of the standard quadratic cost function of a model predictive controller. The optimization problem is solved with constraints using a standard quadratic program solver. An application example shows the suitability of the approach for active power filters.