基于粗糙集的量子神经网络在变压器故障诊断中的应用

Application of Quantum Neural Network in Transformer Fault Diagnosis Based on Rough Set

为了准确有效识别变压器故障模式,将粗糙集和量子神经网络结合进行变压器故障诊断。采用量子神经网络在宏观上收集数据信息,在微观上通过修正量子间隔将模糊交叉的数据按一定的比例合理分配到相关联的模式中,从而提高模式识别的准确性;利用粗糙集的约简去除冗余的属性、规则,提高量子神经网络的速率。与同输入下BP神经网络的诊断结果进行比较,可知本文方法在变压器故障模式识别方面具有更高的准确性。

In order to identify the transformer fault pattens accurately and efficiently, the quantum neural network combining with the rough sets is used in the fault diagnosis of transformer. The quantum neural network is adopted to collect the macroscopic information, and by correcting the micro-intervals of quantum, the cross and uncertain data are distributed to different patterns according to a certain proportion, so that the accuracy of pattern recognition is improved. Simultaneously, the rough sets are used to improve the speed of quantum neural network by reducing the redundant attributes. Through the comparison of diagnosis with BP neural networks under the same input, the method shows higher accuracy in the fault pattern recognition of transformer.