ACCURATE models of power system loads are essential for analysis and simulation of the dynamic behavior of electric power systems [1]. Having accurate models of the loads that are able to reliably reflect underlying phenomena of the physical loads is important for the purposes of designing automatic control systems and optimization of their configuration. More importantly, the dynamic properties of power system loads have a major impact on system stability [1]–[3]. Inparticular, previous work on the subject of voltage stability reported in the literature indicates that the parameters of both static and dynamic loads have significant impact on voltage stability of the power systems [3],[4].On the other hand, the impact of power system load models on inter-area oscillations is discussed in[1],demonstrating the influence that load parameters have on the dominant system eigenvalues . This dependence reveals the link between the effectiveness of power system damping controllers [e.g., power system stabilizers (PSSs)] and the correctness of the eigenstructure of the system, which is dependent on the load model.
To be able to predict the behavior of a system, reliable models of system components are needed that faithfully reflect the dynamical behavior of the actual physical components of the system. Most of the power system components can be satisfactorily modeled by considering the physical laws which govern the respective components. There are, however, some cases when power system modeling is quite a complicated exercise. Modeling power system loads is one of them. It is known that at high voltage levels, the power system loads have to be aggregated in order to obtain manageable models suitable for analysis and simulations[1].Depending on the load type(e.g. lighting, motor load, heating, etc.),the parameters of the aggregate load model may vary in a wide range. When the parameters of all load components are well known, the parameters of the aggregate load models can be readily determined. If the parameters of separate loads are not known or the load structure is known, but the proportion of various load components is not, deriving an aggregate load becomes more difficult.
It can be argued that in the absence of precise information about a power system load, one of the most reliable ways to obtain an accurate model of the load is to apply an identification technique. That is, if field measurements of load quantities (e.g., the voltage and current/power) adequately describing its behavior are available, then a dynamic and/or static equivalent of the load can be obtained by analyzing functional relationships between these quantities.