基于分段多项式截断奇异值分解法识别桥梁移动荷载

发布时间：2018-05-06 21:02

本文选题：移动荷载识别 + 识别精度　；参考：《华北水利水电大学》2017年硕士论文

【摘要】：桥梁的移动荷载识别属于结构动力学逆问题的范畴,大多数桥梁移动荷载识别方法最终都转化为线性方程组的求解,不同的求解方法得到的结果识别精度不同。为了改进时域法(TDM)识别桥面移动荷载时存在的识别精度受测量噪声、测点类型及数量影响较大等缺陷,本文基于第一识别法,第二识别法,频时域法和时域法的基本理论,结合桥梁的移动荷载识别特点,探讨了几种移动荷载识别相关的正则化方法,提出了基于分段多项式截断奇异值分解法(PPTSVD)识别桥梁移动荷载。采用简化欧拉梁模型,以两轴常力移动轴载和两种类型时变移动荷载为例,由反演车辆荷载作用下桥梁的弯矩响应和加速度响应识别桥面移动荷载,对12种不同测点布置工况进行数值模拟分析,得到了不同噪声水平下分段多项式截断奇异值分解法(PPTSVD)的识别结果,并将该识别精度与时域法(TDM)和截断奇异值分解法(TSVD)识别精度进行比较。其中PPTSVD与TDM识别精度比较结果表明:两种识别方法的识别误差均随测量噪声增大而增大,当噪声水平达到10%时,TDM对应的12种识别工况中有11种识别工况识别误差均超过100%,识别结果不能接受;PPTSVD对应的12种识别工况识别误差均低于100%且其中有10种识别工况识别误差低于30%,其抗噪性能较TDM提高明显。两种方法识别精度均受测点类型及数量影响较大,当识别工况包含较多加速度响应测点时两种方法识别精度较高。当仅由弯矩响应识别桥梁移动荷载时,TDM识别结果不能接受而PPTSVD仍能给出较好的识别结果。PPTSVD识别精度高、识别结果受测点类型及数量影响较小且具有良好的鲁棒性,更适应于桥梁移动荷载的现场识别。
[Abstract]:The identification of moving loads of bridges belongs to the inverse problem of structural dynamics. Most of the identification methods of moving loads of bridges are transformed into the solution of linear equations, and the accuracy of the results obtained by different methods is different. In order to improve the identification accuracy of bridge deck moving load by time domain method (TDM), such as measurement noise, type and number of measuring points, this paper is based on the basic theory of the first identification method, the second identification method, the frequency time domain method and the time domain method. Combined with the characteristics of bridge moving load identification, several regularization methods related to moving load identification are discussed, and a method based on piecewise polynomial truncated singular value decomposition (PPTSVD) is proposed to identify the bridge moving load. A simplified Euler beam model is used to identify the moving loads on the deck of the bridge under the action of vehicle loads by taking two axial constant moving axle loads and two types of time-varying moving loads as examples, and the moment response and acceleration response of the bridge under the action of the vehicle load are inversed to identify the moving loads on the bridge deck. Through numerical simulation and analysis of 12 different measuring point layout conditions, the recognition results of piecewise polynomial truncated singular value decomposition (PPTSVD) under different noise levels are obtained. The recognition accuracy is compared with time domain method (TDM) and truncated singular value decomposition method (TSVD). The comparison of recognition accuracy between PPTSVD and TDM shows that the recognition errors of the two methods increase with the increase of measurement noise. When the noise level reaches 10, 11 of the 12 identification conditions corresponding to TDM have more than 100 recognition errors. The recognition results can not accept that the recognition errors of the 12 identification conditions corresponding to PPTSVD are all less than 100% and there are 10 identifiers among them. When the recognition error is less than 30, the anti-noise performance of the system is better than that of TDM. The accuracy of the two methods is greatly affected by the type and number of measuring points, and the two methods have higher recognition accuracy when the identification conditions contain more acceleration response points. When only the moment response is used to identify the moving load of the bridge, the PPTSVD can still give a better recognition result. The accuracy of the identification is high, and the identification result is less affected by the type and number of the measured points and has good robustness. It is more suitable for the field identification of bridge moving load.
【学位授予单位】：华北水利水电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U441

【参考文献】