论文部分内容阅读
为便于器件的设计选型 ,根据已有实验数据 ,采用人工神经网络 ( ANN )建立了外围环境参数和大功率门极自关断晶闸管 ( GTO)开关特性 (包括损耗特性、开通和关断时间、电压和电流的变化率等 )之间的非线性关系 ,训练误差范围为± 5 %。基于最速下降原理 ,建立了 GTO的等效传热模型—— Foster网络 ,与实验测得的热响应曲线相比 ,此模型的误差范围为± 3 %。最后将这些算法集成在开关特性综合分析软件中 ,为 GTO的动态开关特性研究和应用系统设计提供了分析平台。
In order to facilitate the design and selection of the device, according to the existing experimental data, the artificial neural network (ANN) was used to establish the parameters of the peripheral environment and the switching characteristics of high-power gate self-turn-off thyristor (GTO) including loss characteristics, , The rate of change of voltage and current, etc.), the training error range is ± 5%. Based on the steepest descent principle, an equivalent heat transfer model of GTO, the Foster network, is established. The error range of this model is ± 3% compared with the experimentally measured thermal response curve. Finally, these algorithms are integrated in the switch characteristic analysis software, which provides an analysis platform for the research and application of GTO’s dynamic switching characteristics.