在阐述励磁涌流信号的特征和参数化时频分析方法的基础上,运用遗传算法和粒子群算法对匹配追踪算法进行优化,对励磁涌流和故障电流进行了参数化时频分析.通过比较遗传算法和粒子群算法中的迭代次数对特征值获取的影响以及遗传算法和粒子群算法产生的结果,得出结论:(1)迭代次数增加,匹配追踪算法收敛速度增快；但进一步增加迭代次数,对算法收敛速度影响不大.(2)使用局部粒子群算法对匹配追踪算法进行优化,算法收敛速度较快.通过该分析方法得出的结论,将为使用模式识别对变压器励磁涌流和故障电流进行鉴别提供了依据.%In this paper, based on the introduction of the characteristics and recognition method of Inrush, a Time-Fre-quency analysis on Inrush and internal fault are performed. Through comparison among the effects that different iteration times of same algorithm takes and difference between different algorithm, the following conclusions are drawn: (1) Matching Pursuit algorithm' s convergence rate increases as the iteration times increases; (2) Among three optimi-zation algorithm, Local Particle-Swarm Optimization algorithm has better optimized outcomes. These analysis achieve-ments lay a foundation for future study on pattern recognition method in transformer Inrush.
展开▼
