Modelling of a power network compensated by hybrid power filters with different distorted loads

A power system, constituting a distribution system, nonlinear consumer loads and hybrid power compensators, is a complex system characterised by changing line impedances and load and generator conditions. The use of sophisticated control techniques in these systems is becoming very important, necessitating the need for accurate system modelling, that is software compatible with design programs for artificial intelligence based controllers. This paper focuses on the development of a comprehensive state space model with 64 states to serve as a time domain analysis platform for hybrid power filter control and related problems. A typical scenario comprising 2 generators, 4 transformer substations and 3 consumers with interconnecting lines, is modelled. Each consumer has a hybrid power compensator constituting a static VAr compensator and a PWM-converter based compensator associated with it. The stability and accuracy of the discrete state space model is also analysed in the paper. Results from the model are compared to actual measurements on distorted conditions in industrial networks, and shown to correspond extremely well, also for hybrid compensator operation. The compatibility with artificial neural network software is illustrated. The model is unique because of its scale and the level of diversity in terms of power system problem solving.