基于数据驱动的海洋管道缺陷故障诊断与三轴向重构方法研究

项目名称： 基于数据驱动的海洋管道缺陷故障诊断与三轴向重构方法研究

项目编号： No.61203086

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 自动化学科

项目作者： 马大中

作者单位： 东北大学

项目金额： 25万元

中文摘要： 本申请以实现海洋管道缺陷的故障诊断与重构为目标，研究海洋管道无损检测的故障诊断问题。拟采用数据驱动的方法，按照金属损失缺陷与无损检测数据特征关系建模-不确定及耦合数据的解耦与还原-金属损失缺陷类型分类-金属损失缺陷三轴向重构的主线进行研究，解决采用单轴向传感器进行无损检测识别金属损失缺陷与重构的理论基础问题。主要研究内容如下：1）通过有限元仿真分析与实测数据相结合的方法，建立金属缺陷损失与无损检测数据特征的关系模型2）提出无损检测数据盲源解耦算法，实现数据的有效解耦与信息还原；3）提出基于数据重心的模糊最小-最大神经网络分类算法，提高故障分类的准确度；4）提出基于数据特征融合的支持向量机缺陷重构算法，实现在单轴传感器下对金属损失缺陷的三轴向重构。最后应用上述方法，实现对海洋管道金属损失缺陷的故障诊断与重构，为我国海洋管道安全运行与寿命预测提供支撑。

中文关键词： 无损检测；故障诊断；三轴向重构；数据驱动；数据采集

英文摘要： The goal of this application is to achieve the detection and reconstruction of defects of sea pipeline. The problem of fault diagnosis is researched for sea pipeline non-destructive testing. The main line of the research based on data driven is as follows: building the relationship model for metal loss defects and characteristic of non-destructive testing data - decoupling and restoring the uncertainty and coupling data - pattern classification of metal loss defects - three-axial reconstruction of metal loss defects. The basic theory of non-destructive testing that using one axial sensor to detect and reconstruct the metal loss defect is solved. The main contents are as follows: 1) According to combine finite element analysis and measured data, the relationship model of metal loss defects and characteristic of non-destructive testing data is built. 2) A blind source decoupling algorithm based on non-destructive testing data is proposed for decoupling and restoring the noise data. 3) The fuzzy min-max neural network based the gravity of data is proposed for improving the accuracy of pattern classification. 4) A defect reconstruction algorithm is proposed, which is based on data characteristic fusion and support vector machine, to achieve three-axial reconstruction of metal loss defects based on one axial sensor.

英文关键词： Non-destructive examination；Fault diagnosis；3-axial reconstraction；Data driven；Data acquisiton