基于改进BP神经网络的故障诊断方法  

A fault diagnosis method based on an improved BP neural network

作　　者：吴静[1] 柳世考[2] 邓堃[1] 

机构地区：[1]空军工程大学导弹学院,陕西三原713800 [2]空军装备研究院防空所,北京100076

出　　处：《工业仪表与自动化装置》2007年第3期45-48,共4页Industrial Instrumentation & Automation

摘　　要：对于常规BP神经网络,其收敛过程存在两个很大的缺陷:收敛速度慢;存在所谓“局部最小值”问题。文章采用了一种自适应学习速率动量梯度下降反向传播算法对BP神经网络进行训练,建立了智能诊断模型,并应用于某型坦克发动机的故障诊断,测试结果表明该方法相比常规BP算法能够更有效诊断发动机油路和气路的故障,从而为故障诊断及判定的自动化提供了一个新思路。There are two major defects in the convergence process of a conventional BP nerve network, namely the slow convergence rate and the problem of local minimum. This paper adopts a reverse transmission calculation of self-adapting learning rate with a momentum gradient reduction, which sets up a new fault diagnosis model and then applies the reverse transmission algorithm to the fault diagnosis of a tank engine. The test result with improved BP algorithm is more efficient than that with the conventional one, thus providing a new approach to the fault diagnosis.

关 键 词：改进BP算法 神经网络 发动机 故障诊断 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置] TP183[自动化与计算机技术—控制科学与工程]