论文部分内容阅读
介绍了基于模糊理论的自适应共振网络在变压器故障诊断中的应用。根据对实际运行变压器进行统计得出的故障特征气体分布规律，确定分段隶属函数的分界值，并给出隶属函数的经验参数。针对变压器故障诊断的特点，在ＡＲＴ—２网络的基础上，构造了具有输入隐层的ＦＡＲＴ（ＦｕｚｙＡｄａｐｔｉｖｅＲｅｓｏｎａｎｃｅＴｈｅｏｒｙ）网络，对各特征气体采用不同的隶属函数处理，以增强网络对主要气体特征的灵敏度。并通过实例进行了检验，证实了该方法更为有效。
The application of adaptive resonance network based on fuzzy theory in transformer fault diagnosis is introduced. According to the distribution law of fault characteristic gas which is obtained from the statistics of actual operation transformer, the cut-off value of sub-membership function is determined and the empirical parameters of membership function are given. Aiming at the characteristics of transformer fault diagnosis, based on ART-2 network, FART (Fuzy Adaptive Resonance Theory) network with input hidden layer is constructed, and different membership functions are applied to each characteristic gas to enhance the sensitivity of the network to the main gas characteristics. An example is used to verify the method is more effective.