摘要
矿井提升机健康状态预测是保障煤矿安全稳定开采的重要方法。传统利用接触式传感器采集状态信号存在安装不便、误差大等弊端。针对上述问题,提出基于声纹深度时序的矿井提升机健康状态预测方法。利用非接触式传感器采集提升机声纹信号,通过自适应滤波降噪、预加重和分帧加窗等方法预处理;采用MFCC算法提取13维信号特征;构建DRSN-GRU-Attention神经网络预测模型实现健康状态预测,与DRSN,GRU,DRSN-GRU模型对比,验证该模型对提升机健康状态预测性能更优。
The prediction of the health state of the mine hoist is an important method to ensure the safe and stable mining of coal mines.The traditional use of contact sensors to collect status signals has disadvantages such as inconvenient installation and large errors.Aiming at the above problems,a health state prediction method of mine hoist based on voiceprint depth time series is proposed.Using non-contact sensors to collect hoist voiceprint signals,preprocessing through adaptive filtering,noise reduction,pre-emphasis,and framed windowing;using MFCC algorithm to extract 13-dimensional signal features;constructing DRSN-GRU-Attention neural network prediction model to achieve health state prediction,compared with the DRSN,GRU,DRSN-GRU models,it is verified that the model has better performance in predicting the health state of the hoist.
作者
王锋
李敬兆
王国锋
郑昌陆
WANG Feng;LI Jingzhao;WANG Guofeng;ZHENG Changlu(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan 232001,China;Huaihe Energy Group Co.,Ltd.,Huainan 232001,China;Shanghai SH-Driver Electric Co.,Ltd.,Shanghai 201800,China)
出处
《煤炭技术》
CAS
北大核心
2023年第2期179-182,共4页
Coal Technology
基金
国家自然科学基金项目(51874010
61170060)
北京理工大学高精尖机器人开放性研究项目(2018IRS16)
物联网关键技术研究创新团队(201950ZX003)。
关键词
故障预测
声纹识别
神经网络
矿井提升机
failure prediction
voiceprint recognition
neural network
mine hoist
