基于多模态融合的步进式加热炉电气故障辨识研究

Research on Electrical Fault Identification of Walking Beam Heating Furnace Based on Multimodal Fusion

为降低加热炉运行过程中的事故发生率,提出了基于多模态融合的步进式加热炉电气故障辨识研究,首先,利用基于非线性扩散的导数谱增强模型展开信号增强处理。然后,对增强后的电气信号展开EMD分解,获取信号中的所有IMF分量。利用小波包变换提取电气故障的细节特征信息,并电气故障的细节特征信息进行多模态融合。最后,利用随机森林算法对多模态融合的特征展开故障辨识。实验结果表明,所提方法的损失函数值下降较快并且可以降到0.05以下,特征提取能力强、故障辨识效果好,能够为步进式加热炉电气故障辨识提供帮助。

To reduce the occurrence rate of accidents during the operation of the heating furnace,a study on electrical fault identification of walking beam heating furnace based on multimodal fusion is proposed.Firstly,a derivative spectrum enhancement model based on nonlinear diffusion is used for signal enhancement processing.Then,perform EMD decomposition on the enhanced electrical signal to obtain all IMF components in the signal.Extract the detailed feature information of electrical faults using wavelet packet transform,and perform multimodal fusion of the detailed feature information of electrical faults.Finally,the Random forest algorithm is used to identify the fault of multimodal fusion features.The experimental results show that the loss function value of the proposed method decreases rapidly and can be reduced to below 0.05.The feature extraction ability is strong and the fault identification effect is good,which can help to improve the electrical fault identification of the walking beam furnace.