油田水表检定系统的设计与实现

发布时间：2018-09-05 17:26

【摘要】：在油田的开采过程中,对油田进行注水是保持油层压力和提高油田产量的重要手段。为了能够注水有效,科学开发油田,需要利用水表对注水的流量和体积进行严格监控。水表在使用前必须由计量单位进行误差检定,以确定水表能否投入使用。本课题以图像识别技术为核心,设计一套自动化程度较高的水表检定系统,以实现水表检定过程的自动化和数据管理的信息化,对提高计量单位的工作效率和经济收益有重要意义。首先,本文基于误差检定的方法和控制任务,设计了水表检定系统的总体实现方案。针对检定误差的分析流程和检定装置的工艺原理,给出了检定系统的硬件实现方案;针对系统提出的检定流程的自动化操作和测试数据的信息化管理,给出了软件的整体方案和实现方法。其次,本文对水表图像的识别算法进行了研究。对于字符区域的定位,本文采用了图像形态学处理方法和水平投影方法,完成了字符区域的粗略定位和精确提取。对于水表中的半字识别问题,本文设计了改进的模板匹配算法,优化了匹配算法的相关性系数,提高了半字识别的正确率。对于指针区域的定位,本文采用了Hough圆检测和颜色分割方法,完成了指针区域的粗略定位和和精确提取。对于指针方向的判读,本文设计了最大距离法,可以有效地对指针方向进行判读。最后通过字符和指针的工作方式计算出水表的读数。然后,基于LabVIEW语言设计了上位机软件。利用LabVIEW Real Time等相关类库编写了实时控制程序,利用LabVIEW Math Script编写了图像识别算法,利用Oracle数据库完成了本地数据库的创建,实现了检定过程自动化和数据管理信息化的要求。最后,验证了系统检定流程的合理性,以及图像识别的正确率和执行时间。测试表明,检定系统满足了设计需求,图像识别率可以达到98%以上,识别时间满足实时性要求。
[Abstract]:In the process of oilfield production, water injection is an important means to maintain reservoir pressure and increase oil production. In order to develop oil fields scientifically and effectively, water meters should be used to strictly monitor the flow and volume of water injection. In order to determine whether the water meter can be put into use, the meter must be calibrated by the unit of measurement before it is used. Based on the image recognition technology, a water meter verification system with high degree of automation is designed in this paper to realize the automation of water meter verification process and the information of data management. It is of great significance to improve the work efficiency and economic benefits of the unit of measurement. Firstly, based on the error verification method and control task, the overall realization scheme of water meter verification system is designed. According to the analysis flow of the verification error and the process principle of the verification device, the hardware realization scheme of the verification system is given, and the automatic operation of the verification process and the information management of the test data are presented. The overall scheme and implementation method of the software are given. Secondly, the recognition algorithm of water meter image is studied in this paper. In this paper, the method of image morphological processing and horizontal projection are used to locate and extract the character region roughly. In this paper, an improved template matching algorithm is designed to optimize the correlation coefficient of the matching algorithm and improve the accuracy of half-word recognition. In this paper, Hough circle detection and color segmentation are used to locate the pointer region, and the rough location and accurate extraction of the pointer region are completed. In this paper, the maximum distance method is designed for pointer direction interpretation, which can effectively interpret pointer direction. Finally, the reading of the water meter is calculated by the work of characters and pointers. Then, the upper computer software is designed based on LabVIEW language. The real-time control program is written by using LabVIEW Real Time and other related class libraries, the image recognition algorithm is written by LabVIEW Math Script, the local database is created by using Oracle database, and the requirement of automation of verification process and information of data management is realized. Finally, the rationality of the system verification process and the correct rate and execution time of image recognition are verified. The test results show that the verification system meets the design requirements, and the recognition rate can reach more than 98%, and the recognition time can meet the real-time requirements.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TE938;TP391.41

【参考文献】