摘要
将最小二乘支持向量机(least square support vector machine,LS-SVM)应用于航空发动机气路故障诊断。首先,分析了用于气路故障诊断的巡航偏差数据类别,建立用于进行机器学习的诊断模型训练集,构建基于LS-SVM的气路故障诊断模型;其次,采用模式搜索法优化LS-SVM建模,获取最优建模参数;最终,通过直接面向地空数据链(aircraft communication addressing and reportingsystem,ACARS)链路的报文解析组件,实时获取发动机巡航偏差数据集,远程诊断发动机气路故障。航路应用和对比实验表明:最小二乘支持向量机模型具有较高的诊断精度,适用于气路故障的远程诊断。
We apply the least squares support vector machine (LS-SVM) to the fault diagnosis of the gas path of an aeroengine. First we analyze the data types of aeroengine's cruise deviation, establish the diagnosis model training sets to be learned by machines and build the diagnosis model for the fault of a gas path based on the LS-SVM. Next, we use the pattern search method to optimize the LS-SVM modeling and obtain optimal modeling parameters. Finally, by decoding and storing the reports from an aircraft communication addressing and reporting system ( ACARS), we obtain real-time data sets of aeroengine's cruise deviation and carry out the remote diagnosis of the faults of its gas path. A comparative experiment shows that the LS-SVM model produces accurate diagnosis results and is suitable for remote diagnosis of the faults of aeroengine's gas path.
出处
《机械科学与技术》
CSCD
北大核心
2007年第5期595-599,共5页
Mechanical Science and Technology for Aerospace Engineering
基金
航空科学基金项目(98I52091)
民航基金项目(E0501-MH)资助
关键词
航空发动机
远程诊断
支持向量机
气路参数
aeroengine
remote diagnosis
support vector machine(SVM)
gas path