The Study Of Arcing Fault Diagnostic Techniques

Arcing fault is one of the most important electric causes of disasters such as fire in distribution system. Especially when arcing fault in series occurs in electric devices or circuits, it will not be protected effectively because the circuit current is lower than the action threshold level of traditional protection devices. Also as power electronics technology is becoming increasingly widely used up in low voltage areas, the characteristic of the normal operation current in some of the electric devices is similar to that of the arcing fault current, which will make it more difficult to detect arcing fault exactly. Therefore it is of practical value in engineering to carry out comprehensive experiment study in arcing fault with typical non-linear load and provide fast and reliable detection criterion.With the purpose of further research about the diagnosis algorithm of arcing fault in series by doing typical arcing fault in series simulation experiments under kinds of loads, the software and hardware design of comprehensive experiment platform for arcing fault simulation was done in this paper, which includes arcing fault deviser, data acquisition unit, experiment platform control system and so on. Abundance of tests with kinds of typical load was launched and fingerprints of arcing fault with typical load was built based on plenty of experiment data, which laid an important foundation for data analysis.With the processing analysis of experiment data, this paper presents a method of analyzing and extracting the characteristic frequency bands of the arcing fault current on the basis of the fast wavelet transform with multi-resolution analysis. The signal of arcing fault current is denoised using threshold denoising method combining with Birge-Massart threshold value strategy which is based on non-parametric adaptive estimation theory. Wavelet entropy was introduced to reflect the power distribution of the arcing fault current and effectively extract the transient signal with low energy in the arcing fault current thus it can be used to provide the basis for detecting the arcing fault in series.