基于振动和声频信号HHT特征提取的高速列车轨道伤损探测方法研究

项目名称： 基于振动和声频信号HHT特征提取的高速列车轨道伤损探测方法研究

项目编号： No.61071182

项目类型： 面上项目

立项/批准年度： 2011

项目学科： 金属学与金属工艺

项目作者： 王艳

作者单位： 哈尔滨工业大学

项目金额： 10万元

中文摘要： 以检测高速轮轨列车的轮轨伤损为背景，针对现有的轨道探伤设备和技术无法满足对轮轨进行实时检测的情况，提出一种利用无线传感器网络实现轨道振动和声频信号测量，并对伤损信号进行处理与分类的新型高速列车伤损探测方法。本年度主要研究了轨道模型的建立、轨道信号的HHT分解及轮轨固有模态的提取、轨道固有模态与伤损关系的分析、伤损信号的特征提取与分类以及无线传感器网络的建立。利用有限元软件建立了轨道模型，通过仿真得出相应的轮轨伤损信号，利用HHT(Hilbert-Huang Transform)对信号进行了分解，并结合NTF(Non-negative Tensor Factorization)和RVM(Relevance Vector Machine)方法对伤损信号进行特征提取与分类。初步建立了伤损识别规则，能够较好的识别出轨道伤损的存在及其种类。

中文关键词： 轮轨伤损检测；HHT；伤损关系分析；无线传感器网络

英文摘要： The project focuses on the flaw detection of the wheel/rail in High-speed railway. Based on the existing detection equipments and technologies, the wheel/rail flaw can not be detected detected in real time. Therefore, a novel flaw detection method is proposed which is based on the extraction and classification of vibration and acoustic signals measured by a wireless sensor network.The research points include the rail modeling, the Hilbert-Huang transform (HHT) decomposition of the signals, the analysis of the relation -ship between rail inherent modes and flaws, the extraction and classification of the flaw signals and the distribution of the wireless sensor network. A rail model is constructed by the Finite Element Software. Using the rail model, the corresponding flaw signals are acquired by simulations. Then the signals are decomposed by HHT, and the signal feautres are extracted by Non-negative Tensor Factorization(NTF). At last, the features are classified by Relevance Vector Machine(RVM) and the flaw identification rules are set up. Based on the rules, the existence and type of the rail flaws can be identified effectively.

英文关键词： flaw detection in highspeed railway; Hilbert Huang transform; flaw relationship analysis; wireless sensor network