基于L-M算法的雷达板级电路快速故障诊断  

Board-Grade Circuit Fault Quick Diagnosis in Radar Based on L-M Algorithm

作　　者：李合平[1] 邹明虎[1] 王志云[1] 黄允华[1] 

机构地区：[1]军械工程学院光学与电子工程系,河北石家庄050003

出　　处：《测试技术学报》2004年第4期364-368,共5页Journal of Test and Measurement Technology

基　　金：国防科技预研项目(413270303)

摘　　要：　新型雷达可更换的板级电路很多,为了快速诊断板级电路故障,将一种基于L-M算法的神经网络用于故障诊断;分析了L-M算法BP网络进行雷达板级电路故障诊断的原理;建立了故障诊断样本训练和测试平台;并以实例在该平台上对L-M算法BP网络进行训练和实际诊断.结果表明,该方法诊断准确性高,比其它方法更为快速有效,较好地解决了雷达可更换板级电路的故障诊断问题.There are many removable board-grade circuits in new type radar. In order to recognize circuit faults quickly, an artificial neural network based on L-M algorithm is adopted; the circuit fault diagnosis principle is analyzed; the platform of fault sample train and fault diagnosis is structured, and testing samples are trained with the platform, and then circuit faults are recognized. The diagnosis result shows that the fault diagnosis method is superior to others, can accurately locate fault area, and resolve the problem in new type radar removable board-grade circuit diagnosis.

关 键 词：L-M算法 BP网络 故障诊断 神经网络 快速诊断 测试平台 电路 诊断准确性 训练 方法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]