中文摘要：

在IGBT寿命预测研究中,应用雨流法进行器件所承受应力周次统计可取得较好的效果;但其必须基于完整的应力谱,故存在数据量大、无法满足实时应用需求等弊端。文章提出一种改进的雨流算法,其针对应力谱测试数据流,先经数据清洗处理转换成极值数据,再进行应力循环辨识和计数,最终得到实时的动态应力幅谱和均值结果输出。仿真和工程化应用结果表明,该方法的研究可为实现IGBT健康状态及应用寿命的实时在线测算提供技术支撑,具备一定的应用价值。

英文摘要：

Rainflow algorithm is an effective method for the experienced stress-cycle statistics in IGBT module lifetime prediction, but it has obvious disadvantages of huge data and is unacceptable to real-time application field as it must be on the base of complete stress spectrum, In this paper, an improved rainflow algorithm was presented. Firstly, the measured stress spectrum data is filtered and transformed into extreme data, and then the stress cycles are identified and counted, finally the real-time stress amplitude spectrum and its mean value array can be draw out. Simulation and field application results verified the effectiveness of this method. This improved rain flow algorithm can significantly support the implementation of online IGBT operational status assessing and lifetime prediction for power devices.