| Summary: | 碩士 === 國立臺灣海洋大學 === 電機工程學系 === 93 === Abstract
Not only the line loss but also the transformer loss is there in the power system. The transformer loss is divided into the copper loss and the core loss. The copper loss is bigger than the core loss, but the copper loss is made in the load period. If the transformer is connected to the power system, the core loss would make in the one day. For this reason, the core loss is quite big percent in the transformer loss.
When we are calculating the transformer loss, should take into account nonlinear hysteretic phenomenon, natural unbalanced phenomenon and the voltage is little change if the load changes.
The core loss of the transformer changes vary random, because these nonlinear phenomenons affect each other. These phenomenons bring the polynomials algorithm, which calculates the core loss of the transformer, is not accurate. This paper uses artificial neural network to build the nonlinear relation between the core loss of the transformer and voltage, because artificial neural network has the parallel processing ability which can deal with the high nonlinear problem.
This paper builds some artificial neural network model whose error is smaller than the polynomials algorithm. Even the artificial neural network predicts the iron lose of the transformer in the three phase balanced system is lower than 50%.
|
|---|
