A novel online 4-point rainflow counting algorithm for power electronics

Abstract

The design of reliable products requires accurate load profiles as input for lifetime calculations. Different types of load profiles are required for different failure modes of a system. Lifetime models for IGBT power modules, as used in electric powertrains, are based on load cycles in the temperature time-history. These load cycles can be identified by rainflow counting algorithms with some adaptions to consider the power-on-time. At an early stage of development, load data are often estimated based on former experience with similar products. In order to acquire load data at large scale from field applications, it is required to implement the cycle counting algorithm in an electronic control unit (ECU) on the vehicle. This imposes constraints on the memory usage and computational complexity of the algorithm. Additionally, for an ECU implementation the algorithm has to be real-time capable and it has to work on a sample-per-sample basis. Herein, a novel and efficient online rainflow counting algorithm is presented, which accounts for power-on-time. The algorithm can also be applied to other load signals where rainflow counting is appropriate. Our algorithm is based on the well established 4-point rainflow algorithm which is considered superior to other algorithms.