Investigation on the analysis software for switching characteristics of large capacity GTO

Owing to the lack of sufficient power loss data of different working conditions for GTO in the technical manuals of many companies, the manufactures and the consumers often encounter difficulties in selecting suitable devices for the design of practicable application systems. The authors have constructed a large capacity test circuit to carry out experimental investigation on the loss characteristics of GTO under various arrangements of circuit parameters. Then the powerful ANN (artificial neural network) and the BP (backpropagation) algorithm are introduced to build up the quantitative relations between the varied circuit parameters and the recorded loss data so that power loss tables can be created. In addition, the paper poses a revised steepest algorithm to fit the high-order heat transfer model of GTO according to its transient thermal behavior. All the algorithms are integrated in a software package based on which to carry through power loss analysis and output the transient and average junction temperature data or curves