摘要
作为水泥工业的重要生产设备,回转窑的运行状态决定了企业的生产效益。主要针对回转窑托轮振动信号为媒介对其故障识别做出研究,将奇异谱分解等信号分解方法应用于回转窑托轮振动信号的分解和重构,使用分解得到的托轮、筒体谐波幅值和重构信号的时域特征构建特征向量输入到分类模型中。针对特征维数过高可能造成的分类准确性差异,使用主成分分析方法对特征集进行降维处理,最后采用降维后的特征集建立故障分类模型对样本数据进行故障识别分类,结果表明,奇异谱分解能更好地实现信号分解与重构,降维特征集训练出的分类器能更准确地识别出回转窑的故障类型。
As an important production equipment in the cement industry,the operating condition of rotary kiln determines the production efficiercy of the enterprise.In this paper,we study the fault identification of rotary kiln by applying signal decomposition methods,such as singular spectrum decomposition,to the decomposition and reconstruction of rotary kiln wheel vibration signals,and using the harmonic amplitudes of the decomposed wheel and cylinder and the time domain features of the reconstructed signals to construct feature vectors for input into the classification model.The results show that the singular spectrum decomposition can better achieve the signal decomposition and reconstruction,and the classifier trained by the reduced Witt’s feature set can more accurately identify the fault types of rotary kilns.
作者
杨佳林
张云
YANG Jia-lin;ZHANG Yun(School of Mechanical and Electronic Engineering,Wuhan University of Technology,Wuhan 430070,China)
出处
《武汉理工大学学报》
CAS
2022年第5期98-104,共7页
Journal of Wuhan University of Technology
关键词
回转窑
托轮振动信号
奇异谱分解
故障识别
rotary kiln
roller vibration signal
singular spectrum decomposition
fault identification
