基于隐层优化算法的模糊BP网变压器故障诊断

Transformers Fault Diagnosis Based on Fuzzy BP Network Using Optimization Algorithm for Hidden Layer Nodes

针对变压器故障诊断BP网络隐层节点数过少或过多造成的学习过程不收敛、网络范化能力弱等问题,文章采用隐层优化算法,给出了合理的网络结构,并在此基础上训练网络,给出实例验证了该算法的有效性。

The problem that the hidden layer nodes of BP network to fault diagnosis in power transformer are insufficient or redundant can result in many problems such as study unconvergence,weak ability of generalization.In order to solve these problems,optimization algorithm for hidden layer nodes is presented to decide rational numbers of hidden layer nodes,and to train the network on that base.The example study in this paper indicates that the algorithm is effective.