摘要
Abstract
In order to ensure the safe and stable operation of the mine power supply system,in view of the characteristics of the mine power supply system,such as numerous fault types,large noise and small number of data samples,this paper proposes a method which uses the Markov distance limited loss function and constructs the corresponding fault identification model to im-prove the stacked denoising sparse autoencoder.The method can realize accurate identification and judgment of fault types and fault line selection.The model is applied to the identification of fault data of the mine power supply system.The results show that after only 50 iterations,the classification and identification accuracy of the model can reach more than 90％.When the number of iterations reaches 90,the classification and identification accuracy of the model can reach 100％,which greatly re-duces the complexity of calculation and avoids excessive fitting.The average classification and identification accuracy of the model for fault type and fault line selection are 99.46％and 99.32％,respectively,with high classification and identification accuracy,which can be reasonably applied in fault classification of mine power supply system.关键词
矿井供电系统/马氏距离限制损失函数/栈式降噪稀疏自编码器/故障识别模型Key words
mine power supply system/Markov distance limited loss function/stacked denoising sparse autoencoder/fault identification model分类
信息技术与安全科学
