首页> 外文会议>A critical link: Diagnosis to prognosis >Fault severity estimation using a neural network fault tracking approach
【24h】

Fault severity estimation using a neural network fault tracking approach

机译:使用神经网络故障跟踪方法进行故障严重性估计

获取原文
获取原文并翻译 | 示例

摘要

A much desired milestone for the machinery monitoring community is the ability to move beyond static diagnostics and develop a capacity to track an evolving failute, provide estimates on the severity of the impending fault, and develop a prognostic capability. This paper will provide some insight as to how one develops such a capability using wavelet-based fault monitoring and diagnostic kernels. Essentially, the underpinnings of such a capability result from the ability to 1) develop robust wavelet-based parameters related to an underlying assessment of the operational machine model, 2) rapid extraction of wavelet-based vectors containing the desired individual parameters and 3) post-processing of the raw outputs of a globally trained, nonlinear classifier (e.g. backpropagation neural network) for tracking of the evolving failure through feature space. This paper explores this technique by applying it to a digitized vibration record of an evolving gas turbine blade tip failure obtained during cycle testing of an General Electric F100-GE-129 turbofan gas turbine.
机译:机械监控界一个非常需要的里程碑是超越静态诊断的能力,发展跟踪不断发展的故障的能力,提供对即将发生的故障的严重性的估计,以及发展预测能力。本文将提供一些有关如何使用基于小波的故障监视和诊断内核开发这种功能的见解。本质上,此功能的基础来自以下能力:1)开发与操作机模型的基础评估有关的健壮的基于小波的参数; 2)快速提取包含所需单个参数的基于小波的向量; 3)后期处理-对经过全局训练的非线性分类器(例如,反向传播神经网络)的原始输出进行处理,以跟踪通过特征空间演化的故障。本文通过将其应用于通用电气F100-GE-129涡轮风扇燃气轮机的循环测试期间获得的演变中的燃气轮机叶片尖端故障的数字化振动记录中,探索了该技术。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号