基于自适应最稀疏窄带分解的航空发动机双半内圈轴承故障诊断方法研究

Two-piece inner ring bearing is the critical load-carrying component of aeroengine. Spalling, wear and crack faults are easily occurred in the ball, inner ring and cage of two-piece inner ring under the high speed running and bi-directional axial direction heavy loading condition in aeroengine. For the fault vibration signals, the frequency components are complicated, the fault characteristic bands are aliasing and have local narrow-band characteristics, which lead to the difficulty of extracting fault characteristic information. This project is about to take the advantage of ASNBD including can adaptively adjust the center frequency, the bandwidth and the attenuation ratio, and propose the research on the aeroengine two-piece inner ring bearing fault diagnosis method. Firstly, the finite element models of aeroengine two-piece inner ring bearing faults are established, and the vibration property of the bearing is analyzed. Secondly, the ASNBD differential operators and their iteration algorithms are studied, the parameter optimization models of the filter parameters in ASNBD are established by using the differential operator as the object, and the optimization algorithm of the filter parameters are studied, and then the adaptive decomposition method which is suitable for the vibration signals of two-piece inner ring bearing is proposed. Finally, the constructing and dimension reduction method of eigenvalue matrix, as well as the recognition method of two-piece inner ring bearing fault are researched. This project can offer effective method for aeroengine two-piece inner ring bearing fault diagnosis, and is significative in science and valuable in engineering.

双半内圈轴承是航空发动机关键承载部件，高速运行和双向轴向重载条件下双半内圈轴承滚子、内圈和保持架易产生剥落、磨损和裂纹等故障，故障振动信号频率成分复杂、特征频带混叠且具有“局部窄带”特点，故障特征信息难以准确提取。项目利用自适应最稀疏窄带分解方法(ASNBD)具备的自适应调节中心频率、频带和衰减率的局部窄带截取能力，提出基于ASNBD的航空发动机双半内圈轴承故障诊断方法研究。首先，建立航空发动机双半内圈轴承故障有限元模型，分析轴承故障振动特性。其次，研究ASNBD微分算子及其迭代算法，构建以微分算子为目标的ASNBD滤波器参数优化模型，研究滤波器参数优化算法，形成适于双半内圈轴承故障振动信号的自适应分解方法。最后，研究特征值矩阵构建及降维方法，研究双半内圈轴承故障识别方法。项目为航空发动机双半内圈轴承故障诊断提供有效方法，对于实现航空发动机双半内圈轴承故障诊断具有重要的科学意义与工程价值。

双半内圈轴承是旋转机械关键承载部件之一，在发动机高压高温、高速运行和双向轴向重载条件下易产生剥落、磨损和裂纹等故障。项目通过分析双半内圈轴承故障特性，设计了典型故障实验台。针对双半内圈轴承故障振动信号频率成分复杂、特征频带混叠且受强振动噪声影响的特点，利用自适应最稀疏窄带分解方法 (Adaptive Sparsest Narrow-Band Decomposition，ASNBD)具备的局部窄带宽截取能力，提出了基于互补集合自适应最稀疏窄带分解方法 (Complementary Ensemble Adaptive Sparsest Narrow-Band Decomposition，CE-ASNBD)，有效提取了故障特征信息。为实现双半内圈轴承故障智能诊断，使用复合多尺度模糊熵(Composite Multi-scale Fuzzy Entropy，CMFE)对振动信号特征信息进行表征，结合深度置信网络(Deep Belief Nets，DBN)方法完成了双半内圈轴承的智能识别。研究的主要内容如下：.(1) 振动特性分析。双半内圈轴承常用于均分发动机运转时的双向轴向力，因此使用双向螺纹轴、施压夹具和联轴器构建加压模块，模拟双半内圈轴承的工作环境，在此基础上构建了双半内圈轴承故障实验台，并分析了故障振动信号的时域和频域特性。.(2) 故障特征提取方法。由于双半内圈轴承故障信号频率成分复杂、特征频带混叠，采用ASNBD方法提取具有局部窄带特征的故障信息。为提升ASNDB方法在面对强噪声时的分解能力，结合互补集合方法，提出了CE-ASNBD方法，有效实现了双半内圈轴承的故障特征信息提取。.(3) 故障智能识别方法。采用CMFE表征振动数据 的非线性程度，结合DBN构建了双半内圈轴承故障智能诊断方法。对双半内圈轴承各类典型故障实验数据的分析结果表明，论文方法能准确识别双半内圈轴承不同类型故障。