山东煤炭科技2025，Vol.43Issue(4)：91-95,5.DOI:10.3969/j.issn.1005-2801.2025.04.019
基于多尺度跨周期模型的矿用通风机故障诊断
Fault Diagnosis of Mining Ventilation Fan based on Multi-scale Cross Cycle Model
于双 1张宗伟 1卜冉冉1
作者信息
- 1. 枣庄矿业(集团)济宁七五煤业有限公司,山东 济宁 277600
- 折叠
摘要
Abstract
In order to detect bearing faults in a timely manner and ensure the stable operation of ventilation fans,a multi-scale cross period model based on time series modules and multi-scale convolutional neural networks is proposed for fault diagnosis of mining ventilation fan bearings.This model does not require manual feature extraction and can directly input vibration signals into the network for fault diagnosis.The time series modules uses discrete Fourier transform to convert time-domain signals into frequency-domain signals,learns the periodicity and cross cycle features of vibration signals,so as to capture the time-frequency characteristics of vibration signals;Convert vibration signals into two-dimensional tensors and use multi-scale convolutional neural network to learn features at different levels and scales.The multi-scale convolutional neural network contain multiple convolution kernels of different sizes,which can capture complex information in vibration signals;Verification is conducted based on actual measured bearing data.The experimental results show that the accuracy of the proposed method on the test set reaches 99.0%,which can effectively distinguish various fault modes and verify the superiority of the proposed method.关键词
通风机/故障诊断/神经网络/时序模块/深度学习Key words
ventilation fan/fault diagnosis/neural network/time series modules/deep learning分类
矿业与冶金引用本文复制引用
于双,张宗伟,卜冉冉..基于多尺度跨周期模型的矿用通风机故障诊断[J].山东煤炭科技,2025,43(4):91-95,5.
