Computer Engineering and Applications ›› 2020, Vol. 56 ›› Issue (10): 261-267.DOI: 10.3778/j.issn.1002-8331.1901-0253

Previous Articles     Next Articles

Ensemble Empirical Mode Decomposition of Partial Discharge Signal Based on Storm

YANG Hang, ZHU Yongli   

  1. School of Control and Computer Engineering, North China Electric Power University, Baoding, Hebei 071003, China
  • Online:2020-05-15 Published:2020-05-13



  1. 华北电力大学 控制与计算机工程学院,河北 保定 071003


The stable operation of power equipment is related to people’s life and property safety. The fault diagnosis of power equipment can be realized by installing sensors to collect time series waveform signals and then analyzing and processing the signals. Ensemble Empirical Mode Decomposition(EEMD) algorithm has its unique advantages in the field of non-linear and non-stationary signal processing. However, because of its complexity, as an operation-intensive algorithm, its operation speed can not meet the actual needs in the case of serial execution. Therefore, two EEMD decomposition methods based on Storm are proposed for parallel processing of EMD process and segmented parallel processing of signals. Experiments show that both parallel schemes are faster than traditional serial execution schemes, and the piecewise parallel method has more advantages in long signal processing because of its higher parallelism. The two parallel EEMD algorithms provide a reliable solution for the fast processing of time series signals.

Key words: Ensemble Empirical Mode Decomposition(EEMD), parallelization, Storm, wave signal processing


电力设备的稳定运行关系到人们的生命和财产安全,通过安装传感器对时序波形信号进行采集,对信号进行分析处理,可以实现对电力设备的故障诊断。集合经验模态分解(Ensemble Empirical Mode Decomposition,EEMD)算法在非线性、非平稳的信号处理领域有其独特的优势,但是由于其算法复杂度较大,作为运算密集型的算法在串行执行的情况下运算速度无法满足实际需求。因此,借助分布式计算框架Storm并行处理的特点,提出了基于Storm的EMD过程并行和信号分段并行的两种EEMD分解方式。实验表明,两种并行化方案都比传统的串行执行方式速度更快,并且基于分段并行的方式由于其更高的并行度,在长信号处理中更具优势。两种并行EEMD算法的提出为时序信号的快速处理提供了可靠的解决方案。

关键词: 集合经验模态分解(EEMD), 并行化, Storm, 波形信号处理