齿轮箱早期故障信号分析与智能识别的数学形态学方法

项目名称： 齿轮箱早期故障信号分析与智能识别的数学形态学方法

项目编号： No.51205405

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 机械工程学科

项目作者： 李兵

作者单位： 中国人民解放军军械工程学院

项目金额： 24万元

中文摘要： 数学形态学是一种极具特色的非线性分析理论，近年来在机械故障诊断领域得到了广泛的应用，但这些研究大多集中在机械故障信号处理的层面，对数学形态学理论在机械故障信号特征参数提取和智能识别中的研究还非常少。因此，本项目拟充分利用数学形态学在非线性分析和计算效率上的优势，研究其在齿轮箱早期故障信号处理、特征参数提取和智能识别中的应用途径，建立完整、系统的基于数学形态学理论的齿轮箱早期故障诊断方法，主要内容包括：（1）信号的数学形态学处理方法，主要研究多尺度形态滤波器和自适应形态提升小波的设计与优化；（2）信号的数学形态学特征参数提取方法，主要研究数学形态谱和形态学分形维数的优化计算；（3）信号的数学形态学智能识别方法，主要研究形态学神经网络的结构设计、训练算法以及优化策略。本项目研究成功不仅可以拓展数学形态学理论的应用范围，而且可以提高齿轮箱早期故障诊断水平和效率，具有重要的理论和应用价值。

中文关键词： 齿轮箱；早期故障；数学形态学；信号分析；信号智能识别

英文摘要： The mathematical morphology (MM) theory is a very distinctive nonlinear analysis theory， which has been employed widely for mechanical fault diagnosis in recent years. However, most of the researches have justly focused on the processing phase of the mechanical fault signals. Few researches have been done on applying the MM theory to features extraction and intelligent recognition of the mechanical fault signals. Therefore, based on the advantages of nonlinear analysis and computing efficiency of MM theory, this program intends to explore the application of MM theory in the signal processing, features extraction and intelligent recognition for gearbox early fault diagnosis. The goal of this program is to establish an integrated and systematic signal analysis and intelligent recognition scheme for early fault diagnosis of gearbox based on the mathematical morphology theory. The main contents researched by this program are listed as following: (1) Mathematical morphology processing of the signal, including the design and optimization of multi-scale morphological filter and the adaptive morphological lifting wavelet; (2) Mathematical morphology features extraction of the signal, including optimization calculation of the morphological pattern spectrum and morphological fractal dimensions. (3) Mathematical morphology

英文关键词： gearbox；early defect；mathematical morphology；signal analysis；signal intelligent classification