Stabilization method for a numerical model of a shunt active power filter

The presented paper gives several practical control methods improving the parallel active power filter operation. This structure uses three main parameters: the measured current (observed current), the reference current (desired supplying current) and the filter current (injected by the shunt filter circuit in the junction point). If the observed current is measured on the supply site, a negative feedback loop may be used, relative to the reference current. Because of the circuit non-linearity, we have no explicit control for error margin and switching frequency limit. The switching frequency becomes higher as the error becomes lower. For the signal and power circuit this creates bandwidth and loses problems. From the numerical simulation point of view, this gives convergence problems due to the procedures used for the time step computation. Certain measures are necessary to ensure the convergence stability and the limit of rounding error. Another problem is the control for the limits of the storage capacitor voltage. This must be enough high for sustain the injected current, but enough low to prevent the breakdown of the electronic devices. The goal is to limit the switching frequency, keeping the desired limits for capacitor voltage and approximation error. The paper describes an analog control scheme, the auxiliary elements required to stabilize the current evolution and the control methods, taking into account their effects on the numerical process of convergence. Graphic results are obtained by numerical simulation on a circuit model, using the SPICE program.