摘要
准确检测绝缘油黏度对油浸式电气设备的安全运行至关重要,但传统检测方法存在检测时间长、操作复杂等问题。本文首先搭建绝缘油黏度多频超声检测平台,从频域信号中提取多频超声幅值、相位等频域特征;其次研究不同黏度绝缘油超声信号的幅值、相位频率响应变化规律;最后,利用核主成分分析(kernel principal component analysis,KPCA)对输入声学参量降维,基于多频超声检测技术提出蜣螂优化算法(dung beetle optimizer,DBO)优化支持向量机(support vector machine,SVM)的黏度预测模型。结果表明:DBO优化后的绝缘油黏度预测模型测试集平均相对误差为1.55%,预测准确率达到98.45%,相比于SVM黏度预测模型的准确率提升了12.48%,验证了方法的有效性和准确性。
Accurate detection of the viscosity of insulating oil is crucial for the safe operation of oil-immersed electrical equipment.However,traditional detection methods are plagued by lengthy detection times and complex procedures.In this study,we first built a multi-frequency ultrasonic detection platform for insulating oil viscosity and extracted frequencydomain features such as multi-frequency ultrasonic amplitude and phase from the ultrasonic signals.Secondly,the change laws of amplitude and phase frequency response of ultrasonic signal for insulating oil with different viscosity were studied.Finally,the kernel principal component analysis(KPCA)was used to reduce the dimensionality of the input acoustic features.Based on multi-frequency ultrasonic detection technology,the dung beetle optimizer(DBO)algorithm was proposed to optimize the viscosity prediction model of the support vector machine(SVM).The results show that the average relative error of the test set of the insulating oil viscosity prediction model optimized by DBO is 1.55%,and the prediction accuracy reaches 98.45%,which is 12.48% higher than the accuracy of the SVM viscosity prediction model,verifying the effectiveness and accuracy of the proposed method.
作者
姚远
杨林凡
黄子文
杨红梅
刘立
周渠
YAO Yuan;YANG Linfan;HUANG Ziwen;YANG Hongmei;LIU Li;ZHOU Qu(State Grid Changshou Electric Power Supply Branch,Chongqing 401220,China;College of Engineering and Technology,Southwest University,Chongqing 400715,China)
出处
《绝缘材料》
北大核心
2025年第11期104-111,共8页
Insulating Materials
基金
国网重庆市电力公司科技项目(522008240007)。
关键词
绝缘油
黏度
多频超声
DBO-SVM
insulating oil
viscosity
multi-frequency ultrasonic
DBO-SVM
