基于改进LS-SVM的雷达故障诊断技术  

Improved LS-SVM Based Radar Fault Diagnosis Technology

作　　者：王旭婧[1] 陈长兴[1] 赵学军[1] 任晓岳[1] 

机构地区：[1]空军工程大学理学院,西安710051

出　　处：《火力与指挥控制》2015年第2期63-65,69,共4页Fire Control & Command Control

基　　金：陕西省电子信息系统综合集成重点实验室基金(201107Y16);陕西省自然科学基金资助项目(2014JM8344)

摘　　要：针对传统雷达系统故障诊断流程复杂、诊断时间长且准确率仍有上升空间的问题,将改进最小二乘支持向量机算法引入到雷达系统故障诊断中,建立雷达故障诊断模型。该方法很好地解决了分类和函数估计问题,加快了诊断速度,改善了最小二乘支持向量机的鲁棒性和稀疏性,并提高了诊断准确率。通过对某型雷达故障诊断实例仿真,对比改进最小二乘支持向量机和传统神经网络在雷达故障诊断上应用的效果,仿真结果验证了该方法较传统神经网络,具有更好的雷达故障诊断效率和准确率。In order to solve the complexity of traditional radar fault diagnosis process,longer diagnosis and potential room for accuracy,a new radar fault diagnosis model has been devised in this paper which introduced Least Squares Support Vector Machine （LS-SVM）into radar fault diagnosis. This method has solved the classification and function estimate problem well,speeded up the diagnosis, improved robustness and rarefaction in LS-SVM and increased the accuracy of the fault diagnosis. Contrast the effects of improved LS-SVM and traditional neural network used in radar fault diagnosis through the actual example simulation in a certain type,and simulation examples show that this method,compared with traditional neural network,has more efficiency and accuracy in radar fault diagnosis.

关 键 词：雷达系统 准确率 最小二乘支持向量机 故障诊断 

分 类 号：V411.8[航空宇航科学与技术—航空宇航推进理论与工程]