<p>Transient direction protection has the performance of ultra-high-speed operation, but it is challenging to locate the wave head under weak fault conditions, the sensitivity is not enough, and the reliability is low. To solve this problem, a fault direction relay that does not need to identify the traveling wave head is proposed, and the fault direction is judged by the multi-dimension of fault direction characteristics and working condition attributes. Firstly, the fault direction characteristics are constructed using the wavelet energy entropy difference of the fault currents of the two lines. The fault direction characteristics are combined with the start fault phase angle and transition resistance. The fault direction feature is promoted from two-dimensional space to four-dimensional space with multiple information. The multi-dimensional fault feature vector, composed of the fault direction feature and the fault condition attribute, is used as support vector machine (SVM) input. SVM fuses fault condition attributes into fault direction characteristics to overcome the influence of fault condition attributes on protection and improve protection reliability. A great number of EMTP-RV simulation experiments verify the accuracy and reliability of the proposed algorithm in fault diagnosis, and solve the problem of insufficient fault protection sensitivity of 10%.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Line Direction Protection Based on Wavelet Energy Entropy Difference and SVM

  • Zhenwei Guo,
  • Haojie Li,
  • Hong Wang

摘要

Transient direction protection has the performance of ultra-high-speed operation, but it is challenging to locate the wave head under weak fault conditions, the sensitivity is not enough, and the reliability is low. To solve this problem, a fault direction relay that does not need to identify the traveling wave head is proposed, and the fault direction is judged by the multi-dimension of fault direction characteristics and working condition attributes. Firstly, the fault direction characteristics are constructed using the wavelet energy entropy difference of the fault currents of the two lines. The fault direction characteristics are combined with the start fault phase angle and transition resistance. The fault direction feature is promoted from two-dimensional space to four-dimensional space with multiple information. The multi-dimensional fault feature vector, composed of the fault direction feature and the fault condition attribute, is used as support vector machine (SVM) input. SVM fuses fault condition attributes into fault direction characteristics to overcome the influence of fault condition attributes on protection and improve protection reliability. A great number of EMTP-RV simulation experiments verify the accuracy and reliability of the proposed algorithm in fault diagnosis, and solve the problem of insufficient fault protection sensitivity of 10%.