基于应变模态法智能识别海洋导管架平台的构件裂纹

发布时间：2019-01-11 15:38

【摘要】：海洋平台结构庞大,受风、浪、流和冰等环境因素长期作用,同时还受到地震、台风、海啸和船舶碰撞等意外作用的威胁。在载荷作用下,导管架海洋平台会出现裂纹。由于平台部分结构位于海面以下,裂纹不易被发现,难以直接进行人工检测。当重要部件发生裂纹并在极端海况下产生扩展时,会导致整个结构失效,危及工作人员的生命安全,产生重大的经济损失和海洋环境污染。所以,及时并尽早地发现结构的裂纹具有重要意义。本文以某一含四分之一跨处单裂纹的简支梁为例,计算了该梁的非贯穿单边裂纹损伤侧上、下表面、贯穿裂纹的上表面和内部非贯穿裂纹表面裂纹和内部裂纹等情况的不同损伤程度的位移模态和应变模态。根据已有的基于应变模态差分原理的损伤位置直接指标法ISMSD,利用等间距差分格式计算该简支梁非贯穿单边裂纹应变模态差分曲线,经Matlab编程计算将曲线进行光滑,计算得到直接指标值。由直接指标值的最大值找到对应的两有效极值点,这两个有效极值点间即是损伤位置。实例计算简支梁非贯穿单边裂纹损伤应变模态差分曲线,这些应变模态差分曲线在损伤处发生剧烈变化。差分曲线非峰值点损伤在损伤处不出现极值,因而损伤处的差分值不为零。损伤量不同,差分曲线损伤处突变程度略有不同,其规律相似。运用带有Grubbs的支持向量机法和带有Grubbs的BP神经网络法对该非贯穿裂纹简支梁进行损伤程度智能识别,识别并评估了四分之一跨处单裂纹的损伤程度,并从性能及准确度方面对两种方法进行了比较。若选取应变模态差作为网络输入指标,本文采用的两种方法都可以得到比较高的识别精度,而且有良好适应性。支持向量机方法相对误差更小。采用有限元软件ANSYS计算了某导管架海洋平台模型的一水平管件在完整状态、含单裂纹、含双裂纹三种情形时不同位置和不同损伤程度的频率和应变模态。验证了损伤会引起结构的频率降低和应变模态突变,频率降低的幅度随损伤程度的增加而增大。此外还发现,微小裂纹损伤时引起的频率变化很小：水平管端点损伤和中点损伤的频率下降幅度基本一致；双裂纹情形时的频率下降幅度均高于单裂纹情形时；损伤处应变模态曲线发生了显著改变,随损伤量的增加,应变模态曲线突变增大。采用SCE-UA算法和粗粒度并行遗传算法对平台模型的10处单裂纹进行了损伤程度的逐一智能识别。将应变模态差作为SCE-UA算法和遗传算法的输入数据,这两种方法均能取得较高的识别精度,具有良好的适应性。其中SCE-UA算法损伤识别结果误差更小,更精确。振动诊断中的应变模态法具有相对简单,成本较低,具有实时性、在线性、提取信号方便性和遥测性、可控性等诸多优点。本文的研究为工程实际应用提供了一定的参考价值,在结构损伤诊断识别中具有推广价值。
[Abstract]:The offshore platform has a large structure and is subject to environmental factors such as wind, waves, currents and ice for a long time, and is also threatened by unexpected effects such as earthquakes, typhoons, tsunamis and ship collisions. Under the load, cracks will appear in the jacket offshore platform. Because part of the platform structure is located below the sea level, the cracks are not easy to be found, so it is difficult to detect directly by hand. When an important component is cracked and propagated under extreme sea conditions, it will lead to the failure of the whole structure, endanger the life of the staff, and cause serious economic losses and marine environmental pollution. Therefore, it is of great significance to discover the cracks of the structure in time and as early as possible. In this paper, a simple supported beam with a single crack of 1/4 span is taken as an example to calculate the upper and lower surface of the non-penetrating unilateral crack damage side. The displacement modes and strain modes of different damage degree of the upper surface and inner non-penetrating crack surface crack and internal crack of penetrating crack. According to the existing damage location direct index method based on strain mode difference principle, ISMSD, calculates the strain mode difference curve of the simple supported beam by equal distance difference scheme. The curve is smoothed by Matlab programming. The direct index value is calculated. The corresponding two effective extremum points are found from the maximum value of the direct index value, and the damage position is between the two effective extremum points. An example is given to calculate the strain mode difference curves of non-perforating unilateral crack damage of simply supported beams, which change dramatically at the damage site. There is no extreme value at the damage point, so the difference value of the damage is not zero. The degree of mutation at the damage point of the difference curve is slightly different with different damage amount, and the law is similar. The support vector machine with Grubbs and the BP neural network with Grubbs are used to identify the damage degree of the non-penetrating crack beam, and the damage degree of 1/4 span single crack is identified and evaluated. The performance and accuracy of the two methods are compared. If the strain modal difference is selected as the input index of the network, the two methods used in this paper can obtain high recognition accuracy and good adaptability. The relative error of support vector machine is smaller. The finite element software ANSYS is used to calculate the frequencies and strain modes of a horizontal pipe fitting in a jacket offshore platform model under three conditions: complete state, single crack and double crack. It is proved that the damage can cause the frequency decrease and the strain mode mutation of the structure, and the amplitude of the frequency decrease increases with the increase of the damage degree. It is also found that the frequency variation caused by micro-crack damage is very small: the frequency decrease range of horizontal tube end damage and midpoint damage is basically the same, and the frequency decrease amplitude of double crack case is higher than that of single crack case. The strain mode curve of the damage site changed obviously, and the sudden change of the strain mode curve increased with the increase of the damage amount. SCE-UA algorithm and coarse-grained parallel genetic algorithm are used to identify the damage degree of 10 single cracks in the platform model one by one. The strain modal difference is used as the input data of SCE-UA algorithm and genetic algorithm. These two methods can achieve high recognition accuracy and have good adaptability. The result of SCE-UA damage detection is less and more accurate. The strain mode method in vibration diagnosis has many advantages, such as relatively simple, low cost, real-time, linear, easy to extract signal and telemetry, controllability and so on. The research in this paper provides a certain reference value for the practical application of engineering, and it is worth popularizing in the structural damage diagnosis and identification.
【学位授予单位】：大连理工大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TU317;P742

【参考文献】