Direct Current Series Arc Fault Detection Based on Improved Empirical Mode Decomposition
Aiming at the problem that accuracy of series arc fault detection was low when there was strong noise inter-ference in a direct current system,a direct current series arc fault detection method based on improved complete ensemble empirical mode decomposition with adaptive noise and fuzzy k-means clustering was proposed.Firstly,the improved com-plete ensemble empirical mode decomposition with adaptive noise method was used to decompose the loop current signal,and several intrinsic modal functions were obtained.Then,Hurst exponent value of each intrinsic modal function was calculated to distinguish the noise component from the useful component,and the useful components with Hurst exponent value greater than 0.5 were reconstructed.Finally,peak to peak value and fuzzy entropy of the reconstructed signal were calculated to construct the feature vector as the input of the fuzzy k-means clustering.Normal and fault states were recog-nized by using different locations of cluster centers.The simulation and test results show that the accuracy of the proposed method to distinguish the normal and fault states is 100％,and the accuracy to distinguish the interference and fault states is 93％,which can effectively identify direct current series arc faults.
series arcfault detectionempirical mode decompositionHurst exponentfuzzy k-means clustering
NETL
NSTL
万方数据
维普
