The partial discharges that occur in failures of high voltage switchgear can produce acoustic signals, which can be used to examine equipment failures using acoustic signal analysis. The early methods based on ultrasonic detection have several shortcomings. For example, the equipment is expensive and the effective detection range is small. In light of this, an algorithm for examining partial discharge of switchgear based on speech spectrum features is proposed in this paper. The algorithm first calculates the phonogram of the acoustic signals of discharge, then projects the speech spectrum map to a high dimensional space, and presents the center distance of each block monochromatic map as features. Based on these, a mixed self-encoding deep learning network is constructed. The recognition of ability of the model is improved by integrating noise reduction self-encoding and sparse self-encoding network. In the test of the partial discharge of switchgear, the speech spectrum features proposed in this paper can help to improve the fault detection efficiency. Compared with the speech-feature-based algorithm, the detection rate of discharge examination based on speech spectrum features increase by 4.3%. Combined with the deep learning network algorithm, the algorithm recognition rate can reach 99.7%.
Partial discharge, Acoustic signal analysis, Speech spectrum, Deep learning network
 G.A. Hussain, L. Kumpulainen, J.V. Klüss, M. Lehtonen, The smart solution for the prediction of slowly developing electrical faults in mv switchgear using partial discharge measurements, 2013, IEEE Transactions on Power Delivery, vol. 28, no. 4, pp. 2309-2316.
 S. Kaneko, S. Okabe, H. Muto, M.Y.C. Nishida, Electromagnetic wave radiated from an insulating spacer in gas insulated switchgear with partial discharge detection, 2009, IEEE Transactions on Dielectrics & Electrical Insulation, vol. 16, no. 1, pp. 60-68.
 L. Calcara, M. Pompili, F. Muzi, Standard evolution of partial discharge detection in dielectric liquids, 2017, IEEE Transactions on Dielectrics and Electrical Insulation, vol. 24, no. 1, pp. 2-6.
 H. Mohammadi, F. Haghjoo, Distributed capacitive sensors for partial discharge detection and defective region identification in power transformers, 2017, IEEE Sensors Journal, vol. 17, no. 6, pp. 1626-1634.
 L. Zhang, X.T. Han, J.H. Li, Partial discharge detection and analysis of needle-plane defect in SF6 under negative oscillating lightning impulse voltage based on UHF method, 2017, IEEE Transactions on Dielectrics and Electrical Insulation, vol. 24, no. 1, pp. 296-303.
 Y.M. Yang, X.J. Chen, Partial discharge ultrasonic analysis for generator stator windings, 2014, Journal of Electrical Engineering & Technology, vol. 9, no. 2, pp. 670-676.
 B.H. Ward, A survey of new techniques in insulation monitoring of power transformers, 2001, IEEE Electrical Insulation Magazine, vol. 17, no. 3, 16-23.
 W.M. Cao, Y. Wang, On-line monitoring of unattended substation equipment based on audio recognition, 2013, Journal of Hunan University (Natural Science Edition), vol. 40, no. 9, pp. 48-55.
 S. Du, Research on fault diagnosis algorithm of electrical equipment based on audio feature, 2014, Shandong University.
 H.P. Guo, F.Z. Yang, Kansei evaluation model of tractor shape design based on GA-BP neural network, 2016, Advances in Modelling and Analysis C, vol. 71, no. 1, pp. 92-109.
 T.A. Lampert, S.E.M. O’keefe, A detailed investigation into low-level feature detection in spectrogram images, 2011, Pattern Recognition, vol. 44, no. 9, pp. 2076-2092.
 J. Dennis, H.D. Tran, E.S. Chng, Overlapping sound event recognition using local spectrogram features and the generalised hough transform, 2013, Pattern Recognition Letters, vol. 34, no. 9, pp. 1085-1093.
 J. Dennis, H.D. Tran, H.Z. Li, Spectrogram image feature for sound event classification in mismatched conditions, 2011, Signal Processing Letters, IEEE, vol. 18, no. 2, pp. 130-133.
 G.E. Hinton, R.R. Salakhutdinov, Reducing the dimensionality of data with neural networks, 2006, Science, vol. 313, no. 5786, pp. 504-507
 D.T. Grozdic, S.T. Jovicic, M. Subotic, Whispered speech recognition using deep denoising autoencoder, 2017, Engineering Applications of Artificial Intelligence, vol. 59, pp. 15-22
 D. Luo, R. Yang, B. Li, J.W. Huang, Detection of double compressed AMR audio using stacked autoencoder, 2017, IEEE Transactions on Information Forensics and Security, vol. 12, no. 2, pp. 432-444
 K. Sun, J.S. Zhang, C.X. Zhang, J.Y. Hu, Generalized extreme learning machine autoencoder and a new deep neural network, 2017, Neurocomputing, vol. 230, pp. 374-381
 P. Vincent, H. Larochelle, I. Lajoie, Y. Bengio, P.A. Manzagol, Stacked denoising autoencoders: learning useful representations in a deep network with a local denoising criterion, 2010, Journal of Machine Learning Research, vol. 11, no. 12, pp. 3371-3408.
 J. Lyons, A. Dehzangi, R. Heffernan, A. Sharma, K. Paliwal, A. Sattar, Y. Zhou, Y. Yang, Predicting backbone Cα angles and dihedrals from protein sequences by stacked sparse auto-encoder deep neural network, 2014, Journal of Computational Chemistry, vol. 35, no. 28, pp. 2040-2046.