|dc.description.abstract||The advancement in power electronics technology and in particular the use of nonlinear loads in microgrids where such devices are connected into distributional power grid can cause instabilities. Stability assessment of such grids is done by modeling the grid as two subsystems namely the source and load and use Nyquist stability criterion for the said source and load impedance. Stability analysis of such a grid is a difficult task where the state of the art consists of injecting a perturbation current or voltage to each subsystem of the microgrid structure which is time-consuming or may not be practically feasible.
In a microgrid, there are components which are time-varying where the properties may change with time due to temperature rise because of high power consumption, faults or aging of equipment. These effects call for a method where the system parameters are detected in real-time and from the parameters of the grid the system stability can be calculated without the need of signal injection of a perturbation current or voltage into the microgrid.
In this master's thesis, the problem is dealt with by measuring current and voltage harmonics distortions on the bus lines in an AC system caused by nonlinear loads by using a Kalman filter. An online recursive least squares estimator is used to estimate the impedance parameters based on the harmonic components at different frequencies in the frequency domain. The method for measuring harmonic components has a fast response and can be used for time-varying signals. The method is also verified in systems with dynamics and concluded that this is a steady state method. The impedance of the subsystems of the microgrid is found in real-time which leads to straightforward stability analysis based on the grid impedance, also in real-time.
The methods were verified by simulations with the use of Matlab Simulink with the aid of the SimPowerSystems library for modeling the grid. The impedance estimated from the estimated parameters of the grid were verified by numerical simulations by injecting a small-signal perturbation into relevant subsystems of the grid. A good accordance between the measured impedance by signal injection and the parameterized model was found.||