摘要
针对矿井提升机在井下恶劣环境中工作易发生故障,且故障主要依赖于人工检测的问题,提出了一种基于WPCE-CNN的音频感知提升机健康状况研究。首先,通过小波包分解获取音频信号的二维系数特征矩阵,然后输入具有批量归一化层(BN)和Relu非线性激活层的Block层卷积神经网络(CNN)进行分类识别。最后,实验结果表明,本文提出的WPCE-CNN音频感知提升机健康状况的故障分类研究,准确率高于传统卷积神经网络,达到97.7%,能有效地进行提升机音频信号提取及故障诊断分类任务,此研究在降低人力和物力成本的同时,提高了工作人员的安全性和生产效率。
Aiming at the problem that mine hoist is prone to failure when working in harsh underground environment,and the failure mainly depends on manual detection,an audio perception hoist health condition research based on WPCE-CNN was proposed.Firstly,two dimensional coefficient characteristic matrix of audio signal is obtained by wavelet packet decomposition,and then the Block layer convolutional neural network(CNN)with batch normalization layer(BN)and Relu nonlinear activation layer is input for classification and recognition.Finally,the experimental results show that the proposed WPCE-CNN fault classification research on the health status of the audio perception elevator has a higher accuracy than the traditional convolutional neural network(97.7%),and can effectively extract the audio signal and fault diagnosis classification tasks of the elevator.This research reduces the cost of manpower and material resources while improving the safety and productivity of the staff.
作者
李敬兆
邢梦垚
LI Jingzhao;XING Mengyao(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan Anhui 232001,China;School of Computer Science and Engineering,Anhui University of Science and Technology,Huainan Anhui 232001,China)
出处
《佳木斯大学学报（自然科学版）》
CAS
2022年第1期111-115,共5页
Journal of Jiamusi University：Natural Science Edition
基金
国家自然科学基金项目(51874010)
物联网关键技术研究创新团队(201950ZX003)。
关键词
提升机音频信息
批量归一化
小波包分解
卷积神经网络
elevator audio information
batch normalization
wavelet packet decomposition
convolutional neural network
