强噪声下结构振动特征提取与损伤检测研究

发布时间：2019-05-26 23:20

【摘要】：在各类基于振动响应的工程结构损伤检测过程当中,由于外在自然环境或者人为因素的影响,所测得的振动响应信号不仅包含了结构的损伤信息,同时还存在着大量的噪声干扰。噪声的存在严重影响了结构有效振动响应数据的提取,特别是对于结构初期小损伤状态,此时结构损伤响应特征相对微弱,如果不采用合理的降噪办法就难以得到较好损伤识别结果。因此,本文主要研究的是强噪声背景下基于振动响应信号的特征提取及损伤定位。论文针对振动响应信号的特征提取及损伤定位主要完成了以下工作:针对旋转机械转子碰摩故障检测中存在的噪声干扰问题,提出了基于二次采样的大参数自适应随机共振算法。该算法很好地解决了传统随机共振算法仅适用于小参数检测的局限性,仿真和试验数据分析表明该方法能够显著提高测试数据信噪比指标。本文基于Duffing混沌振子的结构损伤检测,提出了一种新的基于Duffing混沌振子和响应灵敏度结合的结构损伤检测方法。通过一组特定的参数选择对响应信号进行提取,从而避免了繁琐的传统Duffing振子参数的选择方法。把该方法应用于三个维度的梁模型和桥梁模型,结果表明该方法可以很好的强噪声背景下实现结构的损伤定位,这可以为结构损伤识别提供一个更好的思路。为了实现对试验参数的精确识别,在气动噪声条件下进行了超声速飞行试验数据的相空间重构和奇异谱分析相结合的方法。首先,通过数值模拟的方法证明了该方法的可行性。然后,通过对某型超声无人机的声振动试验,对试验数据进行相空间重构,实现了对信号子空间和噪声子空间的奇异值分解。通过定义奇异值差分谱来确定真实信号子空间维数,并对现有的最大差分谱理论,提出了一种优化差分谱峰值的方法。重构结果表明,该方法适用于飞机在超音速飞行条件下的声振动试验的数据处理。针对薄板结构(铝板)损伤检测过程中存在的噪声干扰问题,提出了建立在奇异谱分析上的最大似然性原理分析的损伤检测方法。将Lamb波激励响应信号进行奇异谱分析,通过优选差分谱理论选择最优重构信号进行重构,基于最大相似性原则,通过采用遗传算法(Genetic Algorithms,GA)对重构信号参数进行优化以实现对于测量信号构成部分的分析。铝板试验结果有效证明了该方法的实用性和有效性。考虑强噪声背景环境下的非线性结构损伤检测问题,针对具体的非线性质量-弹簧系统设计了对应的物理等效模型,通过增加结构自由度的方式将原有非线性系统等效为增强的线性系统进行分析,建立了系统非线性项的动力平衡方程。通过奇异谱分析、直接参数识别方法以及矩阵最小秩扰动理论实现了对于非线性系统强噪声背景下的损伤定位及损伤程度判定。
[Abstract]:In the process of damage detection of various engineering structures based on vibration response, due to the influence of external natural environment or human factors, the measured vibration response signal not only contains the damage information of the structure. At the same time, there is also a large number of noise interference. The existence of noise seriously affects the extraction of effective vibration response data of the structure, especially for the initial small damage state of the structure, the damage response characteristics of the structure are relatively weak. If reasonable noise reduction method is not adopted, it is difficult to get better damage identification results. Therefore, this paper mainly studies the feature extraction and damage location based on vibration response signal in strong noise background. Aiming at the feature extraction and damage location of vibration response signal, the main work of this paper is as follows: aiming at the problem of noise interference in rotor rub-impact fault detection of rotating machinery, A large parameter adaptive stochastic resonance algorithm based on quadratic sampling is proposed. The algorithm solves the limitation that the traditional stochastic resonance algorithm is only suitable for small parameter detection. Simulation and experimental data analysis show that the method can significantly improve the signal-to-noise ratio (SNR) index of the test data. Based on the structural damage detection of Duffing chaotic oscillator, a new structural damage detection method based on Duffing chaotic oscillator and response sensitivity is proposed in this paper. The response signal is extracted by a set of specific parameter selection, thus avoiding the tedious traditional Duffing oscillator parameter selection method. The method is applied to the beam model and bridge model of three dimensions. The results show that the method can realize the damage location of the structure under the background of strong noise, which can provide a better idea for structural damage identification. In order to identify the test parameters accurately, the phase space reconstruction and singular spectrum analysis of supersonic flight test data are carried out under the condition of pneumatic noise. Firstly, the feasibility of the method is proved by numerical simulation. Then, through the acoustic vibration test of a certain ultrasonic UAV, the phase space reconstruction of the test data is carried out, and the singular value decomposition of signal subspace and noise subspace is realized. The dimension of real signal subspace is determined by defining singular value difference spectrum, and a method to optimize the peak value of difference spectrum is proposed for the existing maximum difference spectrum theory. The reconstruction results show that the method is suitable for data processing of aircraft acoustic vibration test under supersonic flight conditions. In order to solve the problem of noise interference in the process of damage detection of thin plate structure (aluminum plate), a damage detection method based on the principle of maximum likelihood based on singular spectrum analysis is proposed. The Lamb wave excitation response signal is analyzed by singular spectrum analysis, and the optimal reconstructed signal is selected by optimal selection difference spectrum theory. based on the principle of maximum similarity, the genetic algorithm (Genetic Algorithms, is adopted. GA) optimize the reconstructed signal parameters to realize the analysis of the measured signal components. The experimental results of aluminum plate show the practicability and effectiveness of the method. In this paper, the problem of nonlinear structural damage detection in strong noise background is considered, and the corresponding physical equivalent model is designed for the nonlinear mass-spring system. By increasing the degree of freedom of the structure, the original nonlinear system is equivalent to the enhanced linear system, and the dynamic equilibrium equation of the nonlinear term of the system is established. Through singular spectrum analysis, direct parameter identification method and matrix minimum rank disturbance theory, the damage location and damage degree of nonlinear system under strong noise background are realized.
【学位授予单位】：西北工业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：O327

【相似文献】