复杂背景下仪表指针示数的识别与研究

发布时间：2019-06-02 03:16

【摘要】：在我国电力智能化系统快速发展的今天,电力自动化、无人值守的需求越来越紧迫。指针式仪表作为指示电力环境状态的量测工具普遍存在于电厂中。由于指针式仪表由人工判读完成,所以会出现效率低、误差大等现象。针对这种现象,本文研究并设计了复杂背景下仪表指针示数识别系统。由机器代替人对电厂仪表进行判读,进而达到提高工作效率、降低错误率的目的。本文通过深入了解指针式仪表的特点和构造,结合当前人工智能领域的先进技术,对仪表外形、仪表指针及表盘数字等相关组件进行了分析。本文提出了一种基于Adaboost分类器精确定位仪表的方法,该方法能有效解决因变电站复杂环境而导致的仪表难定位问题,相较于Hough变换的仪表定位具有较高的稳定性和鲁棒性。本文使用Sift算法实现旋转表盘和倾斜表盘中数字正确定位,利用Hough直线检测实现仪表指针提取,并依据平面向量夹角的计算公式完成对指针当前示数的判读。在对指针式仪表的研究和设计中,根据Sift算法的检测结果,本文提出了一种基于聚类算法的特征区域定位方法。该方法按照平面聚类划分的原则定位到Sift匹配点最密集的区域,实现对表盘数字高效定位。在对仪表指针提取的研究过程中,本文提出一种基于局部中心位置的指针重定位方法,该方法通过指针与表盘中心的位置关系利用Hough检测结果实现对指针位置重定位,进而完成对仪表指针更为精确的提取。在设计结构上,本课题采用手持终端进行现场信息采集,服务端对采集图像信息进行处理的C/S架构体系。根据实际需求实现了仪表指针识别系统,并完成了对COMPOUND GAUGE(真空压力)仪表指针示数的精确判读工作。
[Abstract]:With the rapid development of intelligent power system in our country, the demand of power automation and unattended is becoming more and more urgent. As a measuring tool to indicate the state of power environment, pointer instrument widely exists in power plant. Due to the manual interpretation of the pointer instrument, there will be low efficiency, large error and so on. In view of this phenomenon, this paper studies and designs the instrument pointer indication recognition system under the complex background. The instrument of power plant is interpreted by machine instead of human, so as to improve the working efficiency and reduce the error rate. Based on the deep understanding of the characteristics and structure of the pointer instrument and the advanced technology in the field of artificial intelligence, this paper analyzes the instrument shape, instrument pointer and dial number and other related components. This paper presents a method of accurate instrument positioning based on Adaboost classifier. this method can effectively solve the problem of instrument positioning caused by complex environment of substation, and has higher stability and robustness than Hough transform instrument positioning. In this paper, Sift algorithm is used to locate the numbers in rotating dial and inclined dial correctly, Hough line detection is used to extract instrument pointer, and the current indication number of pointer is interpreted according to the calculation formula of plane vector angle. In the research and design of pointer instrument, according to the detection results of Sift algorithm, a feature region location method based on clustering algorithm is proposed in this paper. According to the principle of plane clustering, this method can locate the densest area of Sift matching points, and realize the efficient location of dial numbers. In the research process of instrument pointer extraction, this paper proposes a pointer relocation method based on local center position, which realizes pointer position relocation by using Hough detection results through the position relationship between pointer and dial center. Then the instrument pointer is extracted more accurately. In the design structure, the handheld terminal is used to collect the field information, and the server is used to process the collected image information. According to the actual demand, the instrument pointer identification system is realized, and the accurate interpretation of COMPOUND GAUGE (vacuum pressure) instrument pointer indication is completed.
【学位授予单位】：山东理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TM62

【参考文献】