Design And Implementation Of Oilfied Water Meter Calibration System

Water flooding is a frequently used technique for oil exploitation,which aims to increase oil production by keeping the reservoir pressure constant.To achieve efficient water flooding,water meters are required for real time monitoring the water flow and volume.However the water meters need to be calibrated before they are used.In this study we develop an automatic water meter calibration system based on the image recognition technology.The system greatly improved the work efficiency and economic benefit as it realize the automatic calibration and information management.Firstly,we analysis the error detection problem as well as the calibration theory.Then we provide an overall design of the system including the hardware implementation strategy and the corresponding software design.The hardware system is designed by considering the technical principles of the calibrating apparatus.The software is developed for automatic operation of calibration process and information management of test data.Secondly,the recognition algorithm of water meter images is studied in this paper.We employ the image morphology method and the horizontal projection method to accurately locate the extract the characters.For the half-character-recognition problem,we propose a template matching algorithm by optimizing the correlation coefficients,which leads an improvement recognition accuracy.In addition,the method of Hough circular detection and color segmentation are used to accurately locate and extract the indicator region.The indicator direction determination problem is solved by an efficient method named the maximum distance method.Finally,we combine the recognition results of characters and indicators to obtain the readings of the water meter.The host computer program developed by LabVIEW software based on the Real-Time libraries,where the image recognition algorithm is coded by the LabVIEW Math-Script and the Local database is built by ORCLE.As a result,the automatic calibration and information management are successfully re alized.Finally experimental results demonstrate that the proposed method has a high image recognition accuracy.The success ratio of image recognition exceed s 98% and the required time is also satisfied.