基于t-SNE与CFSFDP算法的多源局部放电脉冲分类技术

为解决在进行多源局部放电脉冲分类时因等效时频特征分布重叠而导致的脉冲无法有效分离的问题,提出一种基于t-SNE与CFSFDP算法的局部放电脉冲分类技术.该技术首先通过一种相位同步装置同时采集放电脉冲信号与其对应的相位信息,以单一放电脉冲的时频谱图作为对象,通过t-SNE算法对频谱数据进行降维,再对降维结果进行CFSFD...

Multi-source Partial Discharge Pulse Classification Technology Based on t-SNE and CFSFDP Algorithms

When multi-source partial discharge pulses are classified, the pulses cannot be effectively separated due to the distribution overlapping of their equivalent time-frequency features. In order to solve this problem a partial discharge pulse classification technology is proposed based on t-SNE and CFSFDP Algorithms. Firstly, a phase synchronization device is used to collect the discharge pulse signals and their corresponding phase information, and based on the time-frequency spectra of a single discharge pulse, the t-SNE algorithm is used to reduce the dimensionality of the spectrum data. And then the CFSFDP algorithm is used to cluster the dimensionality reduction results. Finally, based on the phase information collected with the phase synchronization device, the PRPD spectra of different discharge pulses are reconstructed for further analysis. The experimental results show that the t-SNE and CFSFDP based method can effectively classify different discharge pulses, and the PRPD spectrum reconstructed with the phase synchronization device meets the characteristics of discharge.