Bearing Fault Diagnosis Method Based on Average Pooling Layer Time Convolution Network
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
万方数据
为了进一步提高滚动轴承故障检测的准确性、改善时间卷积网络模型(Temporal Convolutional Network,TCN)存在的过拟合问题,本研究提出了增加平均池化层的时间卷积网络(Temporal Convolution with Average Pooling Network,TCAPN)模型.该方法首先使用膨胀因果卷积代替传统卷积神经网络,其次在残差模块多个地方加入平均池化层改善模型过拟合问题,最后结合多个改进残差模块构建本研究提出的TCAPN模型.实验结果表明,在相同工况条件下,TCAPN模型能够更快地收敛,并且平均故障诊断准确率达到了98.73%,相较于TCN模型提高了2.87%,验证了该模型具有高准确性和鲁棒性.
To further improve the accuracy of rolling bearing fault detection and address the overfitting issue in the Temporal Convolutional Network(TCN)model,a novel approach called Temporal Convolution with Average Pooling Network(TCAPN)was proposed in this study.The proposed method utilized expansive causal convolution instead of traditional convolutional neural networks.Additionally,average pooling layers were incorporated at multiple locations within the residual modules to mitigate the overfitting problem.The TCAPN model was constructed by combining multiple improved residual modules.Experimental results demonstrated that under the same operating conditions,the TCAPN model exhibits faster convergence and achieves an average fault diagnosis accuracy of 98.73%,which is 2.87%higher than the TCN model.These findings validate the high accuracy and robustness of the proposed TCAPN model.
NETL
NSTL
万方数据
