摘要
声发射技术可以实现阀门内漏的在线无损检测,但需要建立声发射信号特征量与泄漏量之间的关系。为满足实际应用的需要,同时避免精确建模的困难,采用支持向量机(SVM)分类的方法,根据声发射信号的特征量,确定阀门泄漏量等级,从而判断阀门是否需要维修。在传统网格搜索法的基础上,根据SVM特征参数的特点,对网格搜索法进行改进,以提高搜索效率,缩短参数寻优的时间。在此基础上,采用SVM对液体阀门泄漏量进行分类,对阀门完好状态、小泄漏状态和大泄漏状态进行判断,预测的准确率为95.12%。
Acoustic emission technology enables online non-destructive detection of valve internal leakage.However,it necessitates the establishment of a relationship between the characteristic parameters of acoustic emission signals and the leakage magnitude.To meet practical application requirements while circumventing the challenges associated with precise modeling,a Support Vector Machine(SVM)classification approach is employed.Based on the characteristic parameters of acoustic emission signals,the leakage magnitude level of the valve is determined,thereby facilitating the assessment of whether valve maintenance is required.Building upon the traditional grid search method,modifications are introduced in accordance with the unique characteristics of SVM feature parameters to enhance search efficiency and reduce the time required for parameter optimization.Subsequently,SVM is utilized to classify the leakage magnitude in liquid valves,distinguishing between intact conditions,minor leakage,and major leakage states.The prediction accuracy achieved through this approach is 95.12%.
作者
叶国阳
徐玉全
YE Guoyang;XU Yuquan(Jianghuai Advanced Technology Center,Hefei,Anhui 230031,China)
出处
《自动化应用》
2025年第14期213-217,221,共6页
Automation Application
关键词
声发射技术
阀门
内漏
支持向量机
网格搜索法
acoustic emission technology
valve
internal leakage
SVM
grid search method
