Study on insulation defect discharge features of dry-type reactor based on audible acoustic

The audible signal (20 Hz to 20 kHz) generated during the operation of a dry-type reactor is one of the key features to evaluate its operation state. However, at present, the time/frequency domain features of the audible signal under typical insulation defects are not clear, which makes it difficult to effectively carry out the fault diagnosis of the dry-type reactor based on audible signals. Based on this, the audible signal features under typical insulation defect discharge of the dry-type reactor are studied in this paper. Three typical insulation defect discharge models (metal protrusion defect, pollution defect, and metal foreign bodies defect) are simulated; then, the discharge acoustic signals are collected under the voltage of 8 kV. Analysis of acoustic signal features shows that the acoustic signal of metal protrusion defect discharge occurs in the positive half cycle of power frequency with the cycle of 20 ms. Both pollution defect discharge and metal foreign bodies defect discharge show intermittent discharge; during the discharge, the discharge acoustic signal of pollution defect has features similar to that of metal protrusion defect, while the discharge of metal foreign bodies defect shows randomness with a larger amplitude of discharge acoustic signal (up to 110 dB). Based on the frequency domain feature vectors’ extraction, it is found that the odd harmonic component of the metal protrusion defect discharge acoustic signal accounts for more than twice the even-order harmonic, and the high-frequency energy (12.5–25 kHz) is the main component. The discharge acoustic signal of the metal foreign bodies defect is mainly low-frequency energy, and the high-frequency energy proportion of pollution defect discharge acoustic signal is similar to that of low-frequency energy. In addition, the variance of high-frequency energy ratio (Rhf) of the metal foreign bodies defect discharge acoustic signal is 5.25 and 21 times that of the metal protrusion defect and pollution defect. This study provides some technical support for further detection and assessment of the operation state of the dry-type reactor based on audible signals.