基于联邦扩展卡尔曼滤波的结构损伤识别方法
发布时间：2018-07-24 14:31
【摘要】：在桥梁的健康监测系统中,桥梁结构的损伤识别工作至关重要。目前,基于时域振动信号的结构损伤识别方法发展十分迅速。基于时域法的损伤识别方法包括扩展卡尔曼滤波和最小二乘法等。扩展卡尔曼滤波方法(Extended Kalman Filtering Method,EKF)是一种能适用于非线性系统结构并对结构物理参数进行识别的实时递推算法,该方法能够准确的识别出结构参数变化的位置和变化程度。然而传统的扩展卡尔曼滤波方法存在两个局限:(1)增广的状态向量包括结构的物理参数,当结构复杂时,状态向量较大的维数将导致计算效率低下;(2)EKF属于一种集中式的滤波结构,其容错性差,当振动信号存在故障将影响识别结果。为了改进EKF方法的局限性,本文应用联邦扩展卡尔曼滤波方法(Federal Extended Kalman Filtering Method,FEKF)对结构进行损伤识别。FEKF结构灵活,计算量小且具有很好的容错性,能够准确的识别出故障信号,避免故障信号导致的识别结果发生错误。通过模态坐标变换,以模态坐标代替结构位移响应来扩展状态向量,并采用模态截断方法,起到对状态向量降维的作用,减少计算量,提高识别结果的稳定性。本文的主要研究内容如下:(1)把导航中应用广泛的FEKF用于桥梁结构的损伤识别。推导基于结构自由振动响应的FEKF损伤识别递推公式,构造简支梁结构自由振动的算例,在不同的噪声情况下采用FEKF对不同的损伤工况进行损伤识别。结果表明,采用FEKF对自由振动的桥梁结构损伤识别有良好的适用性和有效性,并且损伤参数的识别结果精度高,稳定性好。(2)目前,采用扩展卡尔曼滤波这类递推方法对移动荷载下的桥梁结构进行损伤识别工作非常少见。然而实际工程中,桥梁结构都处于服役阶段,很难测出其自由振动信号。为了满足桥梁在服役阶段进行在线损伤识别的要求,本文提出用FEKF对移动荷载下的桥梁结构进行损伤识别的方法。首先推导基于移动荷载下的桥梁结构振动响应的FEKF损伤识别递推公式。再构造移动荷载下的简支梁算例,在不同噪声以及不同工况的情况下采用FEKF对结构损伤进行识别。识别结果表明,FEKF能利用移动荷载作用下的桥梁结构的强迫振动响应信号,有效识别结构的损伤位置和损伤程度。(3)目前,桥梁结构的损伤识别工作中通常基于传感器测量信号完全正确的假设。然而当传感器发生故障时,通过传感器得到的振动信号也会因传感器的故障发生错误,从而导致结构损伤识别结果出现错误。本文结合FEKF提出采用残差2?检验法对信号是否存在故障进行检测。采用移动荷载下的简支梁算例进行验证并对振动信号施加不同的故障类型。结果表明,残差2?检验法能够很准确的识别出故障信号,并且可以对故障类型进行判别。识别出故障信号后,通过对故障信号进行排除,从而保证识别结果的正确性和稳定性。通过故障检测功能的应用,即使故障存在信号,FEKF依然能保证识别结果的精度不受到故障信号的影响,体现其良好的容错性。
[Abstract]:In the health monitoring system of the bridge, the damage identification of bridge structure is very important. At present, the method of structural damage identification based on the time domain vibration signal is developing very rapidly. The method of damage recognition based on time domain method includes extended Calman filter and least square method. The extended Calman filter method (Extended Kalman Filtering Met) Hod, EKF) is a real-time recursive algorithm which can be applied to the structure of nonlinear systems and identify the physical parameters of the structure. This method can accurately identify the position and degree of change of structural parameters. However, there are two limitations in the traditional extended Calman filtering method: (1) the augmented state vector includes the physical parameters of the structure, When the structure is complex, the larger dimension of the state vector will lead to low computational efficiency; (2) EKF belongs to a centralized filter structure, its fault tolerance is poor and the fault of the vibration signal will affect the recognition results. In order to improve the limitation of the EKF method, the federal extended Calman filter (Federal Extended Kalman Filtering Met) is applied in this paper. Hod, FEKF) the structure damage identification.FEKF structure is flexible, small calculation and good fault tolerance, can accurately identify the fault signal, avoid the fault signal caused by the error of recognition. Through modal coordinate transformation, the modal coordinate instead of structural displacement response to expand the state vector, and use modal truncation method, The main research contents of this paper are as follows: (1) FEKF is used in the damage identification of bridge structures widely used in navigation. The recurrence formula of FEKF damage identification based on structural free vibration response is derived, and an example of free vibration of simple beam structure is constructed. The FEKF is used to identify the damage of different damage conditions under different noise conditions. The results show that FEKF has good applicability and effectiveness for the damage identification of bridge structures with free vibration, and the identification results of the damage parameters are of high accuracy and good stability. (2) at present, the extended Calman filter is used to move to the movement of the bridge. The bridge structure under load is very rare for damage identification. However, in practical engineering, the bridge structure is in the service stage, it is difficult to detect its free vibration signal. In order to meet the requirements of on-line damage identification of bridge in the service stage, this paper proposes a method to identify the damage of bridge structure under the moving load with FEKF. The recursive formula of FEKF damage identification based on the vibration response of bridge structures under moving loads is derived, and a simple supported beam under moving load is constructed to identify the structural damage with FEKF under different noise and different conditions. The results show that the FEKF can use the forced vibration of the bridge structure under the action of moving load. The dynamic response signal can effectively identify the damage position and damage degree of the structure. (3) at present, the damage identification of the bridge structure is usually based on the completely correct hypothesis of the sensor measurement signal. However, when the sensor fails, the vibration signals obtained through the sensor will also be wrong because of the sensor failure, resulting in structural damage. There is a mistake in the result of the injury identification. This paper combines FEKF with the residual 2? Test method to detect the fault of the signal. The simple supported beam under the moving load is used to verify and apply different fault types to the vibration signal. The result shows that the residual 2? Test method can identify the fault signal accurately and can be correct. After the fault signal is identified, the fault signal is excluded to ensure the correctness and stability of the recognition results. Through the application of the fault detection function, even if the fault exists the signal, FEKF can still ensure that the accuracy of the recognition result is not affected by the fault signal and reflects its good fault tolerance.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU317
本文编号：2141677
[Abstract]:In the health monitoring system of the bridge, the damage identification of bridge structure is very important. At present, the method of structural damage identification based on the time domain vibration signal is developing very rapidly. The method of damage recognition based on time domain method includes extended Calman filter and least square method. The extended Calman filter method (Extended Kalman Filtering Met) Hod, EKF) is a real-time recursive algorithm which can be applied to the structure of nonlinear systems and identify the physical parameters of the structure. This method can accurately identify the position and degree of change of structural parameters. However, there are two limitations in the traditional extended Calman filtering method: (1) the augmented state vector includes the physical parameters of the structure, When the structure is complex, the larger dimension of the state vector will lead to low computational efficiency; (2) EKF belongs to a centralized filter structure, its fault tolerance is poor and the fault of the vibration signal will affect the recognition results. In order to improve the limitation of the EKF method, the federal extended Calman filter (Federal Extended Kalman Filtering Met) is applied in this paper. Hod, FEKF) the structure damage identification.FEKF structure is flexible, small calculation and good fault tolerance, can accurately identify the fault signal, avoid the fault signal caused by the error of recognition. Through modal coordinate transformation, the modal coordinate instead of structural displacement response to expand the state vector, and use modal truncation method, The main research contents of this paper are as follows: (1) FEKF is used in the damage identification of bridge structures widely used in navigation. The recurrence formula of FEKF damage identification based on structural free vibration response is derived, and an example of free vibration of simple beam structure is constructed. The FEKF is used to identify the damage of different damage conditions under different noise conditions. The results show that FEKF has good applicability and effectiveness for the damage identification of bridge structures with free vibration, and the identification results of the damage parameters are of high accuracy and good stability. (2) at present, the extended Calman filter is used to move to the movement of the bridge. The bridge structure under load is very rare for damage identification. However, in practical engineering, the bridge structure is in the service stage, it is difficult to detect its free vibration signal. In order to meet the requirements of on-line damage identification of bridge in the service stage, this paper proposes a method to identify the damage of bridge structure under the moving load with FEKF. The recursive formula of FEKF damage identification based on the vibration response of bridge structures under moving loads is derived, and a simple supported beam under moving load is constructed to identify the structural damage with FEKF under different noise and different conditions. The results show that the FEKF can use the forced vibration of the bridge structure under the action of moving load. The dynamic response signal can effectively identify the damage position and damage degree of the structure. (3) at present, the damage identification of the bridge structure is usually based on the completely correct hypothesis of the sensor measurement signal. However, when the sensor fails, the vibration signals obtained through the sensor will also be wrong because of the sensor failure, resulting in structural damage. There is a mistake in the result of the injury identification. This paper combines FEKF with the residual 2? Test method to detect the fault of the signal. The simple supported beam under the moving load is used to verify and apply different fault types to the vibration signal. The result shows that the residual 2? Test method can identify the fault signal accurately and can be correct. After the fault signal is identified, the fault signal is excluded to ensure the correctness and stability of the recognition results. Through the application of the fault detection function, even if the fault exists the signal, FEKF can still ensure that the accuracy of the recognition result is not affected by the fault signal and reflects its good fault tolerance.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU317
【参考文献】
相关期刊论文 前6条
1 尚久铨;卡尔曼滤波法在结构动态参数估计中的应用[J];地震工程与工程振动;1991年02期
2 谢献忠,易伟建;结构物理参数时域识别的子结构方法研究[J];工程力学;2005年05期
3 范新南;徐立中;苏丽媛;;基于联邦滤波技术的吨位智能测量系统[J];计算机仿真;2006年06期
4 邱恺;荣军;陈天如;吴训忠;;基于信息补偿的自适应联邦滤波算法[J];数据采集与处理;2007年03期
5 郭惠勇,张陵,蒋建;基于遗传算法的二阶段结构损伤探测方法[J];西安交通大学学报;2005年05期
6 顾双春;;基于衰减记忆滤波法的桥梁变形监测数据分析[J];公路与汽运;2009年04期
，本文编号：2141677
本文链接：https://www.wllwen.com/jianzhugongchenglunwen/2141677.html
