Fault diagnosis based on comprehensive geometric characteristic and probability neural network
Fault diagnosis based on comprehensive geometric characteristic and probability neural network
复制标题
基于综合几何特征和概率神经网络的故障诊断
DOI：
10.1016/j.amc.2013.12.122
复制
发表时间：
2014-03-01
影响因子：
4
通讯作者：
Zhang, Huifeng
中科院分区：
文献类型：
--
作者：
Chen, Xiaoyue;Zhou, Jianzhong;Zhang, Huifeng
故障诊断是保证水轮发电机组安全运行的重要环节。近年来，轴心轨迹识别已成为重油机组故障诊断的一种有效方法。本文的目的是提出一种基于综合几何特征和概率神经网络的轴心轨迹识别方法(CGC-PNN)用于重油机组故障诊断。该方法从结构、区域和边界三个方面提出了宏观欧拉数(ME)、模糊凸凹特征(FCC)和边界层特征(BL)来表示轴心轨迹。因此，由ME、FCC和BL组成的特征向量充分融合了最有效和最全面的图像信息。根据特征向量的简单性，引入概率神经网络作为分类器。最后，将该方法应用于800个样本，实验结果表明，该方法在HGU故障诊断中具有较高的准确率。(C)2013 Elsevier Inc.保留所有权利。
Fault diagnosis is very important to ensure the safe operation of hydraulic generator units (HGU). Shaft orbit identification has been highlighted as an effective method for HGU fault diagnosis in the past few years. The purpose of this paper is to propose a novel shaft orbit identification method based on comprehensive geometric characteristics and probability neural network (CGC-PNN) for HGU fault diagnosis. In this method, macroscopic Euler-number (ME), fuzzy convex-concave feature (FCC) and boundary-layer feature (BL) are proposed to represent shaft orbits from three different aspects: structure, region and boundary. Therefore, the most effective and comprehensive image information is fully integrated by the feature vector composed of ME, FCC and BL. Furthermore, probability neural network (PNN) has been introduced as the classifier according to the simplicity of the feature vector. Finally, we apply the proposed method to 800 samples and the experimental results indicate that the proposed method can achieve an efficient accuracy in HGU fault diagnosis. (C) 2013 Elsevier Inc. All rights reserved.