GPS结构监测数据失真分析与异常识别方法研究

发布时间：2019-05-03 20:02

【摘要】：GPS监测技术以其全天候、全自动、高效快速和高精度等特点,已经越来越多地应用于多种大型结构的安全性监测中。GPS信号作为一种功率很小的微弱信号,在其传播过程中易受到各种外界因素的干扰如多路径效应、整周跳变现象等。这些干扰源会引起相当比例数据的异常和失真,致使结构性态识别与故障诊断结果频频出现误/警现象,因此,无论是在信息获取阶段还是在安全评价环节,为了获取更为准确的监测数据,GPS监测过程中的数据失真与变异问题都应引起高度重视。针对GPS异常失真数据的检验和识别的问题,本论文主要进行了以下几个方面的理论和试验的研究。 (1)系统总结了GPS定位技术的相关理论,详细介绍了GPS定位中的误差分类。针对不同的误差项给出相应地削弱方法,以期减少在GPS监测中由于测量误差而造成的数据失真和异常。 (2)完成了基于GPS技术的大连北大桥动态监测。基于ANSYS软件建立的大桥的有限元模型,获取大桥自振频率和振型,为GPS流动站的布置提供参考。最后利用GrafNav/Net软件完成了流动站坐标的动态解算,获得了大桥的实时动态位移。 (3)提出了基于控制图的GPS异常监测数据检验方法。构建了休哈特和累积和(Cumlative Sum,简称CUSUM)两种控制图,给出了对应的异常预警模型。针对GPS监测数据不服从正态分布的问题,提出了利用累积分布函数的核密度估计将其转化为Q统计量,并以此为基础构建了基于Q统计量的控制图应用于GPS异常数据的检验中将两种控制图应用于GPS仿真和实测数据中,结果表明休哈特控制图能够检测出3倍标准差以上的大偏移,但是缺少小偏移检验能力。CUSUM控制能够检测出3倍标准差以下的连续小偏移,最小可达0.5倍的标准差,但是随着偏移的增大CUSUM控制图的误判范围在增加。实际应用中可以根据需要,选择不同的控制图进行异常数据的判别。 (4)提出了一种基于关联负选择的积分步识别算法用GPS监测数据的异常检验。首先构建GPS监测数据的关联模型,通过设定预警控制限对异常数据快速初步判定然后再利用自适应半径负选择算法对异常数据发生范围进行精确的二次识别。对于负选择算法中固定半径检测器和固定半径自体表示方法的不足文中给出了相应改进。分别用仿真的正弦数据和GPS实测数据验证了该方法的有效性,结果表明本文提出的方法能够准确识别GPS监测数据中的异常,关联负选择的双重机制保证了异常检验的效果,能够用于GPS异常数据识别的实际工程中。
[Abstract]:Because of its all-weather, automatic, high efficiency and high precision, GPS monitoring technology has been more and more used in the security monitoring of many large-scale structures. As a weak signal with very little power, the GPS signal has been used in the security monitoring of many large-scale structures more and more. In the process of propagation, it is vulnerable to the interference of various external factors, such as multipath effect, whole cycle jump phenomenon and so on. These interference sources will cause anomalies and distortions of a considerable proportion of data, resulting in frequent errors / alarm phenomena in structural behavior identification and fault diagnosis results. Therefore, both in the information acquisition phase and in the safety evaluation process, In order to obtain more accurate monitoring data, the problem of data distortion and variation in the process of GPS monitoring should be paid more attention. In order to test and identify GPS abnormal distortion data, the following theoretical and experimental studies are carried out in this paper. The main contents are as follows: (1) the related theories of GPS positioning technology are summarized systematically, and the error classification in GPS positioning is introduced in detail. In order to reduce the data distortion and anomaly caused by measurement errors in GPS monitoring, the corresponding weakening methods for different error terms are given in order to reduce the data distortion and anomaly caused by measurement errors in GPS monitoring. (2) the dynamic monitoring of Dalian North Bridge based on GPS technology is completed. Based on the finite element model of the bridge established by ANSYS software, the natural vibration frequency and mode shape of the bridge are obtained, which provides a reference for the arrangement of the GPS flow station. Finally, the real-time dynamic displacement of the bridge is obtained by using the GrafNav/Net software to calculate the coordinate of the flow station. (3) the GPS anomaly monitoring data checking method based on control chart is proposed. Two kinds of control charts, Hewhart and cumulative and (Cumlative Sum, (CUSUM), are constructed, and the corresponding anomaly early warning models are given. In order to solve the problem that GPS monitoring data do not obey normal distribution, the kernel density estimation of cumulative distribution function is used to transform it into Q statistics. Based on this, the control chart based on Q statistics is constructed to test the GPS anomaly data. Two kinds of control charts are applied to the GPS simulation and measurement data. The results show that the Hewhart control chart can detect more than 3 times the large deviation of the standard deviation, and the results show that the control chart can detect the large deviation more than 3 times the standard deviation. CUSUM control can detect continuous small offsets below 3 times standard deviation, with a minimum of 0.5 times standard deviation. However, with the increase of offset, the error range of CUSUM control chart is increasing. In practical application, different control charts can be selected to distinguish abnormal data according to needs. (4) an integration step recognition algorithm based on correlation negative selection is proposed, which uses GPS to detect abnormal data. Firstly, the correlation model of GPS monitoring data is constructed, and the early warning control limit is set to determine the anomaly data quickly and preliminarily, then the adaptive radius negative selection algorithm is used to identify the occurrence range of abnormal data accurately. For the deficiency of the fixed-radius detector and the fixed-radius self-representation method in the negative selection algorithm, the corresponding improvements are given in this paper. The validity of the proposed method is verified by the simulated sine data and the GPS measured data. The results show that the proposed method can accurately identify the anomalies in the GPS monitoring data, and the dual mechanism of negative correlation selection ensures the effectiveness of the anomaly test. It can be used in practical engineering of GPS anomaly data recognition.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P228.4

【参考文献】