风力机叶片疲劳裂纹特征提取方法研究
发布时间：2018-11-25 10:22
【摘要】：风力机叶片的故障已经成为现有风场中的隐患，本文旨在研究风力机叶片承受随机载荷和交变载荷共同作用时的裂纹萌生、生长和扩展信号特征，分析不同的初始裂纹的辨识方法，了解裂纹动态生存状态与叶片疲劳损伤程度之间的因果关系，识别风力机叶片的损伤程度和类型，由此研制具有自主知识产权的大型风力机叶片疲劳损伤辨识系统，从而解决难以实时监测大型风力机叶片的问题，在故障尚轻微时尽早地准确识别其位置和程度，提前对叶片故障预警，保障风力机高效安全地运行，大大降低风力机后期维修成本。 本文建立了初始裂纹及裂纹生长扩展的诊断模型，通过试验设计，搭建试验平台采集不同裂纹类型、不同裂纹阶段对应的故障信号，为有效地监测叶片状态优化声发射传感器安装位置，同时确定叶片裂纹故障信号的采样频率、采样长度、滤波频率等信号采集和检测的技术参数。分析风力机叶片承受循环载荷作用下的裂纹变化特征，明确瞬时声发射导波传递对裂纹生存状态关联机制，研究局部集中应力导致裂纹生长的叶片疲劳损伤特征，确定裂纹的形变特点、增长率以及叶片疲劳损坏程度之间的因果关系，由此及时准确地评估风力机叶片疲劳状态。 本文结合试验模拟的方法进一步分析风力机叶片裂纹萌生和扩展机理，了解动态应力对叶片疲劳破坏的影响，根据裂纹类型和状态判定风力机叶片的疲劳损伤程度，构建一个以声发射信号为监测参量、基于自适应小波分析提取微细裂纹故障特征的机制。首先结合Shannon熵方法实现小波基函数的自适应选取，实现消除背景噪声、分离有用信息，提取裂纹故障信号中的微细特征，在此算法基础上将采集到的风力机叶片裂纹声发射信号进行特征提取，再使用到小波尺度谱及重分配尺度谱中，通过对比得到不同类型裂纹的特征信号，，完成对初始裂纹的萌生扩展状态的特征提取。风力机的玻璃钢叶片材料不存在明确的疲劳极限，当叶片出现裂纹导致叶片固有频率下降时，不同部位裂纹对固有频率的影响不同，裂纹深度扩展后振型将发生变化，而且产生裂纹的原因多样，由此引发的裂纹生存状态不同，因此提取叶片疲劳裂纹特征的分析机制是非常复杂的。从多分辨率角度入手来提取裂纹特征，分析采集数据中的敏感参数，挖掘叶片微细裂纹故障的特征参数，建立初始裂纹的诊断形式，展现叶片疲劳裂纹在不同频率段的特征，成为解决此问题的关键。本文使用多分辨率的奇异值分解并重构信号从而得到噪声干扰更少的信号；再进行重分配尺度谱的多分辨率计算使得在每一分辨率上的信号更加准确且更具有实际操作性，同时结合能量表达方法，得到可以指导实践的特征向量。 针对风力机叶片疲劳短裂纹从萌生，生长及扩展，到多裂纹以及长裂纹的出现直到叶片断裂的这一裂纹群体性行为。使用实时的声发射信号采集裂纹的特征就会出现时间跨度长，外界因素不好控制且损伤状态不好界定的问题。故采用分形理论来分析经过疲劳加速试验得到的叶片缩尺模型的不同裂纹阶段的裂纹几何特征。极大的排除了外界因素的限制和干扰，从而展现了从裂纹萌生到叶片断裂的全过程，并用分形维数这一参量表达了损伤变化程度。
[Abstract]:The fault of the wind turbine blade has become a hidden danger in the existing wind field. The purpose of this paper is to study the crack initiation, growth and extension signal characteristics of the wind turbine blade under the mutual action of the random load and the alternating load, and to analyze the identification methods of different initial cracks. To understand the causal relationship between the dynamic life state of the crack and the fatigue damage degree of the blade, the damage degree and the type of the wind turbine blade are identified, and the fatigue damage identification system of the large wind turbine blade with the independent intellectual property is developed. so as to solve the problem that the large-scale wind turbine blade is difficult to be monitored in real time, the position and the degree of the blade can be accurately identified as early as the fault is still slight, the early warning of the blade fault is advanced, the high-efficiency and safe operation of the wind turbine is guaranteed, and the later maintenance cost of the wind turbine is greatly reduced. In this paper, the diagnosis model of initial crack and crack growth expansion is established, and the test platform is designed to collect the fault signals corresponding to different crack types and different crack stages, so as to effectively monitor the blade state and optimize the installation of the acoustic emission sensor. the technology of signal acquisition and detection of the sampling frequency, the sampling length, the filtering frequency and the like of the blade crack fault signal at the same time In this paper, the characteristics of crack change under the action of the wind turbine blade under the action of cyclic loading are analyzed, and the related mechanism of the transient acoustic emission guided wave to the crack survival state is determined, and the fatigue damage characteristics of the blade caused by the local concentrated stress are studied, and the deformation of the crack is determined. The causal relationship between the characteristics, the growth rate and the degree of fatigue damage of the blade is assessed and the fatigue of the wind turbine blade is assessed in a timely and accurate manner. In this paper, the crack initiation and expansion mechanism of the wind turbine blade is further analyzed by the method of test simulation, and the effect of dynamic stress on the fatigue damage of the blade is studied. The fatigue damage degree of the wind turbine blade is determined according to the crack type and the state, and an acoustic emission signal is constructed. In order to monitor the parameters, the micro-crack fault is extracted based on the adaptive wavelet analysis The self-adaptive selection of the wavelet base function is first realized by combining the Shannon entropy method, the background noise is eliminated, the useful information is separated, the fine features in the crack fault signal are extracted, and the collected wind turbine blade crack sound emission signal is input on the basis of the algorithm. the characteristic signals of different types of cracks are obtained through comparison to obtain the characteristic signals of different types of cracks, and the initiation and expansion state of the initial crack is completed, The characteristic feature extraction is that the material of the glass fiber reinforced plastic blade of the wind turbine does not have a definite fatigue limit, and when the natural frequency of the blade is reduced due to the crack of the blade, the influence of the crack on the natural frequency of the different parts is different, the vibration mode of the crack is changed after the crack depth is expanded, and the crack is generated The causes of the fatigue crack of the extraction blade are different, and the analysis mechanism of the fatigue crack characteristics of the extraction blade is It is very complicated to extract the characteristics of the crack from the multi-resolution angle, to analyze the sensitive parameters in the data, to find the characteristic parameters of the fault of the micro-crack of the blade, to establish the diagnosis form of the initial crack, to show the characteristics of the fatigue crack of the blade in different frequency segments, and to solve the problem. The key of the problem is to use the singular value of the multi-resolution to decompose and reconstruct the signal so as to obtain the signal with less noise interference. The multi-resolution calculation of the rescaling scale spectrum makes the signal on each resolution more accurate and practical, and at the same time the binding energy The expression method can be used to guide the practice. A feature vector for the fatigue of a wind turbine blade from the initiation, the growth and the expansion, to the multi-crack and the occurrence of a long crack, until the blade is broken A crack group's sexual behavior. The characteristics of using the real-time acoustic emission signal to collect the crack will have a long time span, and the external factors are not well controlled and damaged. In this paper, the fractal theory is used to analyze the different crack steps of the blade scale model obtained by the fatigue acceleration test. The crack geometry of the segment is greatly eliminated. The limitation and interference of the external factors are greatly eliminated, and the whole process of the crack initiation to the failure of the blade is shown, and the parameter table of the fractal dimension number is used.
【学位授予单位】：沈阳工业大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TM315
本文编号：2355751
[Abstract]:The fault of the wind turbine blade has become a hidden danger in the existing wind field. The purpose of this paper is to study the crack initiation, growth and extension signal characteristics of the wind turbine blade under the mutual action of the random load and the alternating load, and to analyze the identification methods of different initial cracks. To understand the causal relationship between the dynamic life state of the crack and the fatigue damage degree of the blade, the damage degree and the type of the wind turbine blade are identified, and the fatigue damage identification system of the large wind turbine blade with the independent intellectual property is developed. so as to solve the problem that the large-scale wind turbine blade is difficult to be monitored in real time, the position and the degree of the blade can be accurately identified as early as the fault is still slight, the early warning of the blade fault is advanced, the high-efficiency and safe operation of the wind turbine is guaranteed, and the later maintenance cost of the wind turbine is greatly reduced. In this paper, the diagnosis model of initial crack and crack growth expansion is established, and the test platform is designed to collect the fault signals corresponding to different crack types and different crack stages, so as to effectively monitor the blade state and optimize the installation of the acoustic emission sensor. the technology of signal acquisition and detection of the sampling frequency, the sampling length, the filtering frequency and the like of the blade crack fault signal at the same time In this paper, the characteristics of crack change under the action of the wind turbine blade under the action of cyclic loading are analyzed, and the related mechanism of the transient acoustic emission guided wave to the crack survival state is determined, and the fatigue damage characteristics of the blade caused by the local concentrated stress are studied, and the deformation of the crack is determined. The causal relationship between the characteristics, the growth rate and the degree of fatigue damage of the blade is assessed and the fatigue of the wind turbine blade is assessed in a timely and accurate manner. In this paper, the crack initiation and expansion mechanism of the wind turbine blade is further analyzed by the method of test simulation, and the effect of dynamic stress on the fatigue damage of the blade is studied. The fatigue damage degree of the wind turbine blade is determined according to the crack type and the state, and an acoustic emission signal is constructed. In order to monitor the parameters, the micro-crack fault is extracted based on the adaptive wavelet analysis The self-adaptive selection of the wavelet base function is first realized by combining the Shannon entropy method, the background noise is eliminated, the useful information is separated, the fine features in the crack fault signal are extracted, and the collected wind turbine blade crack sound emission signal is input on the basis of the algorithm. the characteristic signals of different types of cracks are obtained through comparison to obtain the characteristic signals of different types of cracks, and the initiation and expansion state of the initial crack is completed, The characteristic feature extraction is that the material of the glass fiber reinforced plastic blade of the wind turbine does not have a definite fatigue limit, and when the natural frequency of the blade is reduced due to the crack of the blade, the influence of the crack on the natural frequency of the different parts is different, the vibration mode of the crack is changed after the crack depth is expanded, and the crack is generated The causes of the fatigue crack of the extraction blade are different, and the analysis mechanism of the fatigue crack characteristics of the extraction blade is It is very complicated to extract the characteristics of the crack from the multi-resolution angle, to analyze the sensitive parameters in the data, to find the characteristic parameters of the fault of the micro-crack of the blade, to establish the diagnosis form of the initial crack, to show the characteristics of the fatigue crack of the blade in different frequency segments, and to solve the problem. The key of the problem is to use the singular value of the multi-resolution to decompose and reconstruct the signal so as to obtain the signal with less noise interference. The multi-resolution calculation of the rescaling scale spectrum makes the signal on each resolution more accurate and practical, and at the same time the binding energy The expression method can be used to guide the practice. A feature vector for the fatigue of a wind turbine blade from the initiation, the growth and the expansion, to the multi-crack and the occurrence of a long crack, until the blade is broken A crack group's sexual behavior. The characteristics of using the real-time acoustic emission signal to collect the crack will have a long time span, and the external factors are not well controlled and damaged. In this paper, the fractal theory is used to analyze the different crack steps of the blade scale model obtained by the fatigue acceleration test. The crack geometry of the segment is greatly eliminated. The limitation and interference of the external factors are greatly eliminated, and the whole process of the crack initiation to the failure of the blade is shown, and the parameter table of the fractal dimension number is used.
【学位授予单位】：沈阳工业大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TM315
【参考文献】
相关期刊论文 前1条
1 陈雪峰;李继猛;程航;李兵;何正嘉;;风力发电机状态监测和故障诊断技术的研究与进展[J];机械工程学报;2011年09期
本文编号：2355751
本文链接：https://www.wllwen.com/falvlunwen/zhishichanquanfa/2355751.html
