摘要
空压机在超负荷运行时,其内部存在摩擦副生热及涡流损耗的强耦合作用,难以捕捉间歇性高温突变特性,使得诊断准确率较低。为此,提出红外热成像特征提取下卷烟厂空压机高温突变检测研究。利用红外热成像技术采集卷烟厂超负荷运行的空压机温度图像,并运用二维Otsu阈值法构建空压机温度图像的直方图,确定图像分割界限,引入区域生长法细化图像分割结果,得到高温区域二值图像。依据分割后的二值图像,提取图像均值、峰度、能量等基础统计特征,并进一步引入时-空双域特征组合(包括小波包能量熵与Zernike矩),以更全面地表征局部高温过热、间歇性高温过热与隐性高温过热3类典型故障的温度分布与形态瞬变特性。利用提取到的综合特征参数构建样本数据集,并输入支持向量机中,根据测试样本与超平面的位置关系诊断卷烟厂超负荷运行空压机的温度过高故障。实验结果表明,所提方法可精准辨识空压机高温区域并提取高温特征,对3类温度过高故障的诊断准确率均稳定在96%以上,相较其他方法展现出显著优势。
When the air compressor operates under overload,there is a strong coupling effect of frictional heat generation and eddy current loss inside,which makes it difficult to capture the intermittent high temperature mutation characteristics,resulting in low diagnostic accuracy.To this end,infrared thermal imaging feature extraction is studied for detecting sudden changes in high temperature of air compressors in cigarette factories.Using infrared thermal imaging technology to collect temperature images of overloaded air compressors in cigarette factories,and using two-dimensional Otsu threshold method to construct histograms of air compressor temperature images,determine image segmentation boundaries,introduce region growing method to refine image segmentation results,and obtain binary images of high-temperature areas.Based on the segmented binary image,basic statistical features such as image mean,kurtosis,and energy are extracted,and further spatiotemporal dual domain feature combinations(including wavelet packet energy entropy and Zernike moments)are introduced to more comprehensively characterize the temperature distribution and transient characteristics of three typical faults:local high temperature overheating,intermittent high temperature overheating,and hidden high temperature overheating.Construct a sample dataset using the extracted comprehensive feature parameters and input it into a support vector machine.Diagnose the high temperature fault of the overloaded air compressor in the cigarette factory based on the position relationship between the test samples and the hyperplane.The experimental results show that the proposed method can accurately identify the high-temperature area of the air compressor and extract high-temperature features.The diagnostic accuracy of three types of high temperature faults is stable at over 96%,showing significant advantages compared to other methods.
作者
刘捷
张秀静
朱登飞
何龙龙
覃柏智
LIU Jie;ZHANG Xiujing;ZHU Dengfei;HE Longlong;QIN Baizhi(China Tobacco Guangxi Industrial Co.,Ltd.,Nanning 530001,China;Beijing ENVADA Electric Power Engineering Technology Co.,Ltd.,Beijing 100080,China)
出处
《国外电子测量技术》
2025年第9期162-169,共8页
Foreign Electronic Measurement Technology
关键词
超负荷运行
空压机
高温突变
红外热成像
支持向量机
overload operation
air compressor
high temperature fault
infrared thermography
support vector machine
