基于Mel-CNN模型的磨煤机声纹识别研究

朱斌1; 李锲1; 曹宏2; 常达3

基于Mel-CNN模型的磨煤机声纹识别研究

引用本文：朱斌1，李锲1，曹宏2，常达3.基于Mel-CNN模型的磨煤机声纹识别研究[J].计算技术与自动化,2025,(4):179-183

摘要点击次数: 10

全文下载次数: 4

作者	单位
朱斌1，李锲1，曹宏2，常达3	（1.国家能源集团永州发电有限公司，湖南永州 425000；2.北京华控智加科技有限公司，北京 100080 3.国网淄博供电公司，山东淄博 255022）

中文摘要:磨煤机是燃煤电站的重要辅助设备之一，它的运行状况直接关系到机组的安全稳定运行。因此，提出了一种基于Mel-CNN模型的磨煤机轴承声纹识别方法，以有效检测磨煤机的运行状态。首先利用噪声采集平台模拟了各种轴承故障作用下的噪声信号，提取Mel谱特征，然后将磨煤机轴承各状态下产生的Mel谱图输入到CNN模型中，该模型对各状态下生成的Mel谱图进行训练、学习和特征提取。最终实验结果表明，Mel-CNN模型在训练时间、模型大小指标等方面优于其他模型，在轴承故障声纹识别方面具有较大优势。

中文关键词:磨煤机 Mel谱图声纹识别 CNN 轴承故障

Research on Voiceprint Recognition of Coal Mill Based on Mel-CNN Model

Abstract:Coal mill is one of the important auxiliary equipment of coal-fired power station, and its running condition is directly related to the safe and stable operation of the unit. A method based on Mel-CNN model is proposed to identify the bearing vowels of coal mill, so as to detect the running state of coal mill effectively. Firstly, the noise signal under the action of various bearing failure is simulated by the noise acquisition platform, and the Mel spectrum features are extracted. Then, the Mel spectrum generated under each state of the mill bearing is input into the CNN model, which trains, learns and extracts the Mel spectrum generated under each state. The final experimental results show that the Mel-CNN model is superior to other models in terms of training time and model size, and has great advantages in bearing fault voicing recognition.

keywords:coal mill Mel spectrum voiceprint recognition CNN bearing failure

查看全文 查看/发表评论 下载pdf阅读器