الفهرس | Only 14 pages are availabe for public view |
Abstract Power system dynamic equivalents are critical for obtaining a quality operating performance. The representation of power system components in detailed models makes the aggregated system model too complicated. This way numerous off-line simulations of the nonlinear system dynamics is conducted, and as a result, a huge cpu time and memory allocations are needed. To alleviate a such problem armed with the recently developed approximation techniques offered by the artificial neural network (ANN), a dynamic system model is devised to describe the complex dynamic performance. Synchronous machines are widely used in power generation, and their stable performance under all anticipated system disturbances is vital for sustaining overall system performance. In this thesis ANN models have been developed for the system generators for the purpose of speeding up the system simulations. A two layered, feed-forward and back-propagated learning capability ANNs was used to emulate the nonlinear models of synchronous generators in a large power system. A combination of current state variables and the system inputs were used as network inputs, while the step ahead predicted state variables were used as network outputs. The results verify that the ANN generator dynamic model can emulate generator dynamics well and should therefore be suitable as a representation method for dynamic stability analysis . |