基于声发射信号的低速旋转机械故障检测理论与实践研究

Lacking of effective fault diagnosis technology always was the hot and difficult problems which were considered by academy and engineering to evaluate safety and identify loss, as low-speed main bearing continuous working under complex operating conditions. After scientific research and practical verification, the fault diagnosis of the conventional vibration signal was hard to have obvious effect and the core was lacking of fault character information. Fundamental research on fault diagnosis for low-speed bearing based on acoustic emission(AE) was conducted, having a focus on dynamics model and problem-solving for main bearing system, accurate simulating loss model of combined failure for inner ring, outer ring and cage. The new way to research on loss feature extraction method under weak loss and much loss coupling was studied and the fault mechanism was illustrated. Based on research above, the integrated and systemic diagnostic framework combining pattern recognition and state prediction method was proposed. It is expected to have innovation and breakthrough in respect as high frequency response analysis for nonlinear contact dynamics, dynamics analysis model of loss, loss failure mechanism and fault development and transfer, and to get academic significance and practical value for basic research.

低速（超低速）主轴轴承在复杂工况条件下连续工作，其安全性评估与损伤识别由于缺乏有效的故障诊断技术，一直是学术和工程界研究的热点和难点。经科研和实践验证，基于常规振动信号的故障诊断难以起到明显效果，其核心在于故障诊断过程中故障特征信息缺乏的瓶颈问题。本项目开展基于声发射（AE）信号的超低速轴承故障诊断基础研究，重点解决高保真主轴系统动力学模型与求解问题，准确模拟轴承外圈、内圈、滚动体破损等复合故障特征的损伤模型、非线性接触动力学小波有限元模型。研究微弱损伤、多损伤耦合下的损伤特征提取仿真分析新方法，阐明故障机理。在上述研究基础上，融合模式识别技术和状态预测方法，提出完整的、系统的诊断框架体系。通过本项目研究，可望在非线性接触动力学高频响应分析、损伤动力学分析模型、损伤故障发生机理和故障发展和转移途径等方面有所创新和突破，获得有学术意义和实用价值的基础性研究成果。

低速、重载轴承的故障诊断一直是轴承故障诊断的重点，而声发射（AE）检测方法对低频信号诊断的优异表现受到业内的广泛关注。本课题在国家自然科学基金资助下开展如下工作，（1）深入研究低速状态下重载轴承的声发射信号特点和动态性能数值模拟的最新进展，汲取相关的研究成果。（2）构造结构非线性、温度场、耦合故障分析的声发射信号特征提取，拓展其应用领域。（3）深入研究并拓展声发射信号至低速、超低速滚柱轴承动态性能分析领域，建立采用时域、频域或者时频域分析方法进行特征提取的基本框架。（4）系统研究从高维、高频的信号向低维、多尺度单元转换矩阵理论，构造一系列基于三维多尺度故障辨识单元算法。（5）将多尺度辨识算法单元融入传统自适应信号分析体系中，提出基于声发射自适应辨识、预测核心算法框架，通过经典考题分析验证其数值逼近能力。（6）设计、制造低速、重载轴承实验平台。