摘要
BP神经网络在故障诊断中发挥着越来越重要的作用。基于航空设备的参数匹配问题,提出了一种对输入值进行相关性处理的方法,然后利用BPNN有自适应Lr的梯度下降算法的进行训练。仿真结果证明该模型训练速度快且有效。
BP neural networks play important role in fault diagnosis. Based on the parameter excursion of air facility, a method of taking the relativity on the input is presented, then trained with grads descend on adoptions Lr of BPNN. The simulation experiments show that this model is quick and efficient.
出处
《舰船电子工程》
2009年第10期163-164,187,共3页
Ship Electronic Engineering
