基于任意传感器排布的叶尖定时信号压缩感知辨识方法

基于叶片振动和叶尖计时系统的特点,推导了任意传感器角度分布下叶片振动响应的分布规律,并基于叶尖定时信号频域的稀疏性,探究了压缩感知方法在叶尖定时信号的重构和倍频辨识上的应用。通过同步信号和非同步信号的大量数值实验,分析了包括信号重构误差、传感器数量、旋转周期数、频率分辨率和信噪比等因素对辨识效果的影响,提出了压缩感知重构叶尖定时信号的基本方法和步骤。进一步将压缩感知方法应用于某型号涡扇发动机钛合金宽弦风扇叶片有限元仿真获得的振动响应数据重构,验证了该方法的有效性。结果表明:在加入了叶尖测点位置误差和30 dB随机噪声的前提下,在80%、90%和100%三种转速工况下均清晰辨识出幅值较大的激励倍频。基于任意传感器角度分布的压缩感知方法对叶尖定时信号辨识效果较好。     

转子振动信号的二阶非平稳源盲分离

复杂的转子系统的多振动混叠信号分离是振动信号处理领域的一个难题。在介绍盲源分离基本原理的基础上,首次选用了二阶非平稳源盲分离算法对带有较强噪声的实际转子系统的振动信号进行盲分离,满意地分离出了各个振动源信号。采用SONS算法得到的振动源频谱分析清晰地反映出了各个振动源的频谱特性,且较好地抑制了噪声的影响。研究中提出的测量方法,为实际复杂结构的振动源盲识别提供了支持。     

高可靠小条件数压缩感知叶尖定时信号辨识

为增强基于任意角度压缩感知(CS)叶尖定时信号(BTT)重构的稳定性,采用小生境微种群遗传算法提出了基于小条件数的高稳定性传感器安装布置方法。研究发现:任意角度CS的恢复矩阵条件数越小,不同信噪比下同步和非同步振动信号重构误差越趋于稳定在较低范围,优化后发现条件数为1的排布方案下,相邻传感器角度间隔趋向于均布2m取K的形式。利用条件数为1的矩阵冗余特性,设计了高可靠性的传感器排布方案,数值试验和有限元仿真验证表明叶尖定时系统在失效一个传感器的情况下仍可稳定得到准确的重构结果。在信噪比为5 dB时,冗余排布相比参考排布的主倍频幅值重构误差降低4.4%以上,验证了采用最小条件数排布的任意角度CS信号辨识方法在噪声及传感器失效情况下仍可保证BTT测量结果的有效性和可靠性。      

Multi-probe linear fitting and time of arrival linear correction method to analyze blade vibration based on blade tip timing without once-per-revolution

Blade vibration monitoring can ensure the safe operation of aero-engine rotor blades. Among the methods of blade vibration monitoring, Blade Tip Timing (BTT) method has attracted more and more attention because of its advantages of non-contact measurement. However, it is difficult to install the Once-Per-Revolution (OPR) probe in the confined space of aero-engine, and the failure and instability of the OPR signal will reduce the reliability of the blade vibration analysis results, which directly affects the accuracy of the blade vibration parameters identification. The Multi-Probe linear fitting and Time of Arrival (ToA) Linear Correction method based on the BTT (MP-LC-BTT) without OPR is proposed to reduce the errors of single probe linear fitting method for blade vibration displacement analysis. The proposed method can also correct the calculation error of blade vibration displacement due to the nonlinear change of rotation speed, which can improve the analysis accuracy of the blade vibration displacement. A new blade vibration model conforming to the actual vibration characteristics is established, and the effectiveness of the proposed method is verified by numerical simulation. Finally, the reliability and accuracy of the MP-LC-BTT method have been verified by the experiments which include two high-speed blade test-benches and an industrial axial fan. This method can be used in the actual aero-engine monitoring instead of the BTT method with OPR.