基于LVQ神经网络的变压器故障诊断方法

结合模糊理论,提出一种基于学习向量量化器(LVQ)的变压器故障诊断方法。它首先在无监督学习模式下,采用数据压缩技术,完成输入空间上的向量重构。接着结合监督学习机制,从输入数据选择特征赋予每个类。该方法是一种将自组织映射(SOM)和监督学习模式结合起来的自适应模式分类技术,具有结构简单,适应性强和分类精度高的特点。变压器故障诊断实例显示了该方法的有效性。

Fault Diagnosis Methods of the Transformer Based on LVQ Neural Network

Combined with fuzzy theory,a novel transformer faults diagnosis method is proposed based on learning vector quantization（LVQ） in this paper.Firstly,under non supervised learning pattern,the method applies data compression technique to accomplish vector reconstruction in input interspace.Then,combined with supervised learning mechanism,it selects the characteristics from the input data to each classification.Obviously,the method is an adaptive pattern classification technique incorporating self-organizing map（SOM） with supervised learning pattern,and owns simple structure,strong adaptability and high classification precision.A practical example in transformer fault diagnosis indicates the availability of the method.