矿用强力输送带缺陷弱磁特性与智能识别研究

项目名称： 矿用强力输送带缺陷弱磁特性与智能识别研究

项目编号： No.U1361121

项目类型： 联合基金项目

立项/批准年度： 2014

项目学科： 矿业工程

项目作者： 马宏伟

作者单位： 西安科技大学

项目金额： 60万元

中文摘要： 针对矿用强力输送带缺陷定位、定量和定性分析难题，研究基于弱磁检测的强力输送带钢丝绳芯缺陷特性与智能识别理论与算法。针对强噪声背景下的信号获取问题，提出基于小波变换的变步长LMS自适应信号去噪算法，提高缺陷信号的信噪比；研究高维模式空间钢丝绳芯缺陷特征提取问题，提出基于改进邻域粗糙集的弱磁检测信号特征约简算法，获得反映缺陷本质的特征向量；研究有限样本情况下钢丝绳芯缺陷模式识别问题，提出基于粒子群优化的模糊二叉树支持向量机多类分类的缺陷分类算法，提高识别的准确率和效率。本研究成果对于弱磁检测缺陷定位、定量和定性分析以及强力输送带缺陷检测与评价均具有重要的学术价值和应用前景，对于提高矿用强力输送带运行的可靠性、确保安全生产具有极其重要的意义。

中文关键词： 强力输送带；缺陷；弱磁检测；特征提取；智能识别

英文摘要： Wire rope defects characteristics of steel cord conveyor belt and intelligent recognition algorithms based on weak magnetic testing will be studied, which is used to solve the problem of location, quantitative and qualitative analysis of the defects. For the problem of signal acquiring under strong noise, a variable step-size LMS adaptive filter algorithm based on wavelet transform is presented to improve signal-to-noise of defect signal. For the problem of feature extraction of the defects with high-dimensional spaces, a modified neighborhood rough set is given to gain nature feature vectors of the defects. For the problemof defects recognition under the finite samples, a fuzzy binary tree SVM multi-class classification algorithm based on particle swarm optimization is put forward to improve accuracy and efficiency of defects recognition. This research have important academic value andapplication prospect on location, quantitative and qualitative analysis of weak magnetic defects, as well as defects detection and evaluation of steel cord conveyor belt. It has very important significance for improving operation reliability of conveyor belt and ensuring safety production.

英文关键词： Steel cord conveyor belt；Defects ；Weak magnetic testing；Feature extraction；Intelligent recognition