【作者】 陈编；

【导师】 肖纯；

【作者基本信息】 武汉理工大学 ， 控制科学与工程， 2012， 硕士

【摘要】 本文所述桥梁重载实时在线识别系统采集前端采用光纤光栅传感器,相对于传统的重载识别具有很大的优势,是桥梁健康监测的一个重要组成部分,已经成为国内外该领域研究的重点。传统载重检查需要车辆停止静态称重,容易造成交通堵塞；采用的传感器主要是基于机电原理设计的,在性能上有其特定的弊端。而本识别装置,首先采集传感器采用的是光纤光栅传感器,在抗干扰性、体积小容易安装、测量值范围空间分辨率,线性范围等性能上都优于传统机电原理的传感器；由于是实时在线识别,车辆在动态运行中就被检查识别,不容易造成交通堵塞。由于动态检查对车辆静态重量以外的因素：车速、检查处桥面平整度、车辆自身的震动比较敏感,在动态识别克服传统识缺点的基础上,保证识别精度是本文识别系统的重点。本文所述桥梁重载实时在线识别主要包括两部分：相关数据如动应变、车速等信息的采集、显示和储存,超重车辆的识别。数据的采集、显示和储存部分通过网口将光纤光栅解调仪等下位机装置的数据传入服务器,通过实时曲线显示采集的数据,通过数据库保存采集的数据,以备以后数据的查询。采集部分涉及网口编程,储存部分涉及利用AD0方式访问数据库,由于采集频率高,数据量大,采集、显示和储存同时进行,所以系统开辟了多个线程,提高CPU利用率。超重车辆识别部分,由于对最终识别结果的影响因素较多,所以将BP神经网络算法引入到本文识别部分。针对传统BP神经网络只能达到局部最优,收敛速度较慢的缺点,利用遗传算法优化传统的BP神经网络,使得优化后的BP神经网络既能提高收敛速度又能达到全局最优。实验结果表明,本文识别系统能在克服传统静态称重的基础上,满足一定误差要求(5%)的现场应用,相对于单纯的动态识别提高了精度。更多还原

【Abstract】 Data acquisition terminal of real-time online identification system of weigh-in-motion of vehicles on bridge described in this article uses fiber bragg grating sensors.There are a great of advantages compared to traditional weigh-in-motion recognition. It is an important part of the bridge health monitoring.It is becoming a focus on domestic and international research in this field. During traditional load check, vehicles need to be stopped and acceptd static weighting.It is likely to cause a traffic jam; sensors used here are mainly based on the electrical and mechanical principles, there are many specific drawbacks in performance. During the identification device, first,acquisition sensors use fiber Bragg grating are superior to traditional sensors based on the electrical and mechanical principles in many aspects,for example, anti-interference, small size and easy to install, resolving accuracy of the scope of the measured values, linear range; during real-time online identification, vehicles Dynamically are checked which is not likely to cause traffic jam. During Dynamic check,many factors can cause changing of result outside of the static weighting:speed, flatness of Bridge deck on check, vibration of the vehicle, dynamic recognition need overcoming the shortcuts of traditional load check. It is important to ensure that accuracy of this identification system.Real-time online identification system of weigh-in-motion of vehicles on bridge consists of two parts:the collection、display and storage of relevant data, such as dynamic strain、speed and other information, the identification of overweight vehicles.In the part of data acquisition, display and storage, acquisition of data is through Ethernet port.Data of fiber grating demodulator machine device is transmissed into sevice.Real-time curve shows data collected through the database.The collected data are saved to database for preparing for data query in future. Acquisition relates to the programming of the Ethernet port. Storage relates to the use of ADO accessing to database. Due to the high frequency of collection and the large volume of data acquisition、display and store at the same time, the system has opened up multiple threads to improve CPU utilization.Final result of identification part of weigh-in-motion of vehicle is affected by many factors. the BP neural network algorithm is introducted to identify the part of this article. Due to traditional BP neural network can only reach a local optimum and convergent slower, genetic algorithms is used to optimize the traditional BP neural network.Optimized BP neural network can not only improve the convergence speed but also reach the global optimum.The experimental results show that this article recognition system can overcome shortcut of the traditional static weighing.lt meets some field applications with error requirement under5%, improvimg the accuracy relativs to a simple dynamic recognition.更多还原