正确吊装集装箱是列车安全货运的基本前提之一

正确吊装集装箱是列车安全货运的基本前提之一 集装箱装车时需要确定的主要是水平方向的位置,因此理想状态是从垂直方向拍摄整个场 景的水平图,然而实际情况并不允许这样做.一方面得到整个水平图需要无限高的摄像机位 置... 集装箱 工业 正确吊装集装箱是列车安全货运的基本前提之一,因此集装箱与车匹的位置配准相当重要.由于铁轨的弯曲与列车停靠位置的不确定性,传统装车方式只能依靠吊车司机的肉眼观察与经 验完成配准,这种方式效率低下而且事故率高,误差大,严重的位置偏差甚至对列车的运行安 全带来威胁,因此必须采用更先行的科学技术来辅助完成集装箱的配准工作.可选的位置配准技术 方案 气瓶 现场处置方案 .pdf气瓶 现场处置方案 .doc见习基地管理方案.doc关于群访事件的化解方案建筑工地扬尘治理专项方案下载 主要有激光测距与图像处理两种,激光测距技术无法适应货运站的剧烈震动的环境,因此采用图像处理的方案.合适的图像可以反映集装箱与车箱的位置关系,然而,这种图像的采集并不容易,位置关系的反映也是仿射的和不稳定的,必须先在图像中找到车箱和集装箱的 位置,然后将它们的大小比例还原,才可能得到它们的实际位置关系.本文系统地研究了利用 图像进行集装箱定位所面临的图像采集,数据模型,物体定位等关键问题,并实现了系统的软硬件开发,实践证明,本文所设计的定位系统误差在10厘米以内,操作方便,性能可靠,具有较高的理论意义与实用价值. 1 集装箱装车时需要确定的主要是水平方向的位置,因此理想状态是从垂直方向拍摄整个场景 的水平图,然而实际情况并不允许这样做.一方面得到整个水平图需要无限高的摄像机位置或 与场景相同大的摄像头孔径,这两个条件显然都无法达到.另一方面,集装箱与车箱的尺寸很接近,从垂直方向看车箱很可能被完全遮挡.因此,只能采用侧上斜拍的方式.根据吊装环境的实际情况,可供选择的拍摄点有吊车室和司机室上的三个位置,既图1中的P1,P2,P3.其中P2,P3所拍摄到的图像中会包含司机室,为减少后续处理的复杂度,本文的研究主要使用P1作为拍摄点. 图1摄像机的位置 2 集装箱定位的主要目的是要获得集装箱与目标车箱的相对位置差异,以便指导吊车司机进行装车作业.为此,必须首先在图像中识别出集装箱与车箱的位置.由于高度与水平距离的不同,并且不同集装箱的大小可能也不相同,集装箱与车箱在图像中的表现也各不相同,要对它们的位置确定并不容易. 2.1 从形状上看,集装箱都是长方体,比较容易识别.然而实际能拍摄到的图像中并不能得到集装 箱的整体图像,并且不同物流公司的集装箱在颜色,大小等方面都不相同,很难将集装箱图像用于精确定位.幸运的是,作业过程中所使用的吊具是始终固定的,并且总能定位到集装箱的 中心位置,因此只需要确定吊具的位置就能获得集装箱的位置.对于吊具系统,我们可以先获得其去背景的照片(如图2),然后再在实际图像中进行目标定位.在获得吊具位置后,则可以根据吊具与集装箱的位置关系获得集装箱的颜色等信息,最后根据颜色信息得到集装箱的整体 轮廓和区域. 图2吊具系统 2.2 虽然车箱是长方体容器,但由于集装箱的遮挡,能拍摄到的通常只是车箱的一个侧面(如图3所示).车箱的颜色与集装箱颜色差别较大,但却与路基颜色接近,因此只能从形状特征识别它.从图3也可以看出,车箱最明显的标识就是它的上边由两条相邻的平行线组成,由于摄像机的 高度固定,这两条平行线的距离可视为固定值.另外,粗略定位很容易由人工完成,也就是说拍照的方向可以固定为如图3所示的角度,车箱位置处于集装箱下方.于是车箱定位问题可以转化为自下而上搜索两箱平行直线的问题. 图3车箱的照片 3 由于车箱与集装箱相对摄像机的距离不相同,图像中直接观察到的位置关系并不等价于它们 实际的位置关系.成像过程实际上是从三维到二维的映射过程,为从图像中获得实际事物的位 置关系,必须恢复这种映射.从二维图像恢复到三维的映射是个困难的问题,通常还是病态的(ill-posed)问题,但集装箱的定位问题有一些有利条件使问题简化.首先,集装箱与车箱的方向通常都是平行或接近平行的,因此在计算它们位置关系时通常只需要考虑它们在某一个截面 上的位置关系,也就是说我们只需要考虑将一个面映射到一条线的关系.其次,摄像机的高度是固定的,车箱高度也固定,因此可以事先确定摄像机到车箱的竖直高度H1.同时,我们还可以得到吊具的宽度,从而可以从图像中的吊具宽度推算集装箱的宽度和位置等信息.根据各种已知条件,我们可以抽象出如图4所示的门吊系统成像模型,其中F为摄像机的焦距. 图4门吊系统的射影模型 4 定位系统的最终研究目标是要确定集装箱与车箱的位置关系,从而指导吊装作业.根据前面的 分析 定性数据统计分析pdf销售业绩分析模板建筑结构震害分析销售进度分析表京东商城竞争战略分析 ,在图4所示的射影模型中,已知数据有:吊具宽度W1,吊具中心与摄像机的水平距离W3,摄像机的焦距F,摄像机与车箱的竖直距离H1等,要估计的目标是集装箱与车箱的水平距离 W2. 2W1,,2首先,由射影定理与勾股定理可以得到摄像机与集装箱的竖直距离为:F,W3,其,,L3,, 中L3可以在图像中测得.再根据三角形相似可以得到集装箱的高度 2L4，L7W1,,2H2,F,W3 ,,L5L3,, 2(L4，L7，L5)(L4，L7)W1,,2于是可得到摄像机到集装箱的水平距离为: F,W3,,,FL5L3,, 2(L4，L7，L5)L7W1,,2摄像机到车箱的水平距离为: F,W3 ,,FL5L3,, 2(L4，L7，L5)L4W1,,2 W2,F,W3,,FL5L3,, 最后可得到车箱与集装箱的水平距离: 5 为提高生产效率,保证货运安全,上述定位系统被运用到成都某货运站中.系统运行性能稳定, 系统误差较小.为测试系统的精度和稳定性,我们通过现场实验获得一组性能数据,实验中针对每个尺寸的集装箱和车箱分别做20交测试,如表1所示. 表1定位系统的误差统计 车箱集装平均绝对最大绝最小绝 宽度 箱宽误差(mm) 对误差 对误差 度 从表1可以看出,系统实际运行的最大误差为31mm ,远远小与系统设计的目标范围(100mm).同时系统的误差波动不大,有利于保持货车的平衡. 6 实践证明,集装箱精确定位系统能有效帮助货运站完成吊装任务,不仅减少了吊装工人反复校正的过程,而且使吊装位置的误差由200mm的范围减小到30mm的范围,为集装箱安全运输提供了技术上的保障.系统的不足之处在于,吊装的位置调整仍然需要工人参与,操作结果总存在较大误差,下一步的工作将是研究将定位系统与控制系统结合,实现吊装工作的自动化和智能化. Since the correct lifting of containers is one of the basic preconditions for safe train shipping, so the position register of containers and vehicles is very important. Due to the bend of rail and the uncertainty of train mooring, the traditional loading can only be registered by a driver’s experience and his eyes. This method is of lower efficiency and high accident frequency as well as large error. The severe location deviation may even cause threat to the safe running of a train, so a more advanced technology should be used to assist to finish container register. The technique proposals for local register we can choose are mainly laser distance measurement and image treatment. Since the former technique fails to adapt to the violent shake, so we use image register proposal. A proper image is able to reflect the local relation between containers and car carriage. However collecting this image is not easy and the reflection of location relation is only imitated reflect and unstable, so we must find the locations of carriage and container in the images and reduce the locations based on the actual sizes, by which, we can obtain the actual location relation. This article has studied some key problems such as how to collect images, digital model, object positioning by using images to orientating containers and this article has realized the development of software and hardware. It has been proved that the error of this positioning system designed in this article is within ten centimeters. This positioning system is of convenient operation and reliable functions and has higher theory significance and practical value. 1. How to capture positioned image When loading containers, the horizontal direction should be assured, so a horizontal view by shooting the entire scene from vertical direction is most perfect. However, such a shooting is not realistic, because on one hand, to obtain the entire view needs infinitely high camera location or a same large camera aperture with the shooting scene, which are both not allowed; on the other hand, since the sizes of container and carriage are fairly close and viewing carriage in vertical direction may be completely kept out, so we can only take a slanting shoot in a side-upper location. Based on the practical situation of hoisting circumstance, there are three shooting points in picture one of crane room and driver room can be chosen, see P1, P2 and P3. The pictures taken in P 2 and P3 will have driver room and in order to lower the complex of picture’s later treatment, this article will focuses on using P1 as shooting point. The location of Camera in Picture one 2. Positioning the images of container and carriage The goal for positioning container is mainly to obtain the relative location discrepancy between container and aimed carriage so as to guide crane drivers to load container, therefore firstly, it is a must to identify the locations of container and carriage in the picture. The height and horizontal distance are different and sizes of different containers may be also different, so container and carriage in the picture look different, thus determining their locations is not so easy. 2.1. The positioning of container Viewing from figure, all containers are cuboid and can be easily identified, however, the practical picture taken are not a entire picture; furthermore, the containers in different logistics companies have different colors and sizes, so it is quite difficult to use container image for accurate positioning. Luckily, the lifting tool in operation tasks are always fixed and can be positioned into central location of container, so we can obtain container location by assuring the location of lifting tool. In terms of lifting tool system, we can firstly obtain its picture removing its background (see Picture Two) and then making goal positioning in the practical picture. When obtaining the location of lifting too, we can get some information, including the color of container, etc. based on the location relation between lifting tool and container and finally get the overall profile and region of container based on color information. Picture Two: The system of lifting tool 2.2. The positioning of carriage The carriage is a cuboid vessel, but due to the container’s sheltering, what we can shoot is generally only a side profile of carriage (please see Picture Three). Carriage color is quite different from that of container, but similar to the color of roadbed, therefore, we can identify carriage color from figure character. From Picture Three, we can see that the most evident mark in carriage is two close parallel lines on top of carriage. The distance of the two parallel lines can be considered as a fixed value since the camera height is fixed. Besides, a cursory positioning can be easily finished, that means that the shooting direction can be fixed as the angle in Picture Three and carriage location is in lower side of container. Therefore, carriage positioning problem can be transformed to searching two parallel lines from bottom to top. Picture Three: the picture of carriage 3. The projection model of crane system Since the distances between carriage and container and camera are different, so the viewed location relations in picture are not their practical local relations. The imaging process is actually a mapping process from three dimension to two dimension. In order to obtain the location relations of practical objects from pictures, such mapping process must be recovered. To recovering three dimension to two dimension is difficult and generally even ill posed. But some advantages in positioning container make such a problem simple. Firstly, the directions of container and carriage are generally parallel or close to parallel, so generally we need only consider the location relation on some section when calculating their location relation, that is we should only consider the relation mapping one surface onto one line; secondly, camera height is fixed and carriage height is also fixed, so the vertical height, H1 from camera to carriage can be confirmed beforehand. At the same time, we can get width of lifting tool, thereby calculating some information, including container width and location. Based on all given conditions, we can abstract an imaging model of crane system like Picture Four, in which, F is camera focal length. Picture Four: The projection model of crane system 4 The estimation for correcting container’s location The ultimate goal for studying positioning system is to make clear the relations between locations of container and carriage so as to guide hoisting operation. Based on the above analysis, in the projection model in Picture Four, the numbers we have known are width of lifting tool W1, the horizontal distance between lifting tool center and camera, W3, focal length of camera, F, the vertical distance between camera and carriage, H1, etc. and the estimated goal is the horizontal distance between container and carriage, W2. Firstly, we can get the vertical distance by Projection’ Theorem and Pythagoras' 2Theorem:W1,,2F,W3, among which, L3 can be measured in picture. Then the ,,L3,, height of container can be got based on triangle 2L4，L7W1,,2similarity:, thus the horizontal distance between H2,F,W3,,L5L3,, 2(L4，L7，L5)(L4，L7)W1,,2camera and container is: F,W3and the horizontal ,,FL5L3,, 2(L4，L7，L5)L7W1,,2distance between camera and carriage : F,W3, finally ,,FL5L3,, we can get the horizontal distance between carriage and container: 2(L4，L7，L5)L4W1,,2W2,F,W3 ,,FL5L3,, 5 In order to promote production efficiency and ensure a safe freight, the above positioning system has been used in some freight terminal in Chengdu. The result is that the system performs steadily and system error is relatively small. To test the system’s precision and stability, we have obtained a group performance data through field experimentation and we have made twenty tests respectively for containers and carriage in all sizes, please see Table One. Table One Error Statistics of the Positioning System carriacontaiMAE Max. Min. ge ner (mm) absolute absolutwidth width error e error From Table One, we can see that the maximum error when the system is running is 31mm, which is much smaller than the system designed range (100mm). Meanwhile, the system’s error fluctuation is limited, which can help to keep truck’s balance. 6 The practice has proved that container accurate positioning system can help to finish hoisting effectively in freight terminal. The system has not only reduced hoisting workers’ repeated correction and also lower the error of hoisting location from 200mm to 30mm, thus guaranteeing a safe container transportation. The disadvantage of this system lies in the demand for workers in adjusting hoisting location and larger error of operation result. The following job is to study the combination of positioning system and control system and how to realize the automation and intelligence of hoisting task. 本文系牛档搜索（Niudown.COM）根据用户的指令自动搜索的结果，文中内涉及到的 资料 新概念英语资料下载李居明饿命改运学pdf成本会计期末资料社会工作导论资料工程结算所需资料清单 均来 自互联网，用于学习交流经验，作品其著作权归原作者所有。不代表牛档搜索（Niudown.COM）赞 成本文的内容或立场，牛档搜索（Niudown.COM）不对其付相应的法律责任！