基于改进BP神经网络和D-S证据理论的装备故障诊断  

Equipment Fault Diagnosis Based on Improved BP Neural Network and D-S Evidence Theory

作　　者：丁文飞[1] 黄子俊[1] 曾剑新[1] 陈国银[1] 

机构地区：[1]空军雷达学院,武汉430019

出　　处：《舰船电子对抗》2010年第4期106-111,共6页Shipboard Electronic Countermeasure

摘　　要：针对雷达情报信息系统结构复杂,设备种类繁多的特点,研究了基于BP神经网络的故障诊断方法,给出一种变学习率和附加动量相结合的改进算法,建立了基于改进BP神经网络和D-S证据理论相结合的故障综合诊断模型。计算实例验证了这种方法的可行性和有效性。Aiming at the characteristics of radar intelligence system such as the complicated structure and various equipments,this paper studies the fault diagnosis means based on BP neural network, gives an improved algorithm combining the alter-learn rate with appending momentum, establishes a fault synthesis diagnosis model based on improved BP neural network and D - S evi- dence theory inference. By the calculation example, the feasibility and availability of this diagnostic method are verified.

关 键 词：雷达情报信息系统 故障诊断 神经网络 D-S证据理论 

分 类 号：TN95[电子电信—信号与信息处理] N945.16[电子电信—信息与通信工程]