基于八链码形状匹配的抽油机井地面示功图凡尔开闭点识别

示功图凡尔开闭点的精准识别对抽油机井工况诊断、功图计产具有重要意义。针对目前示功图凡尔开闭点识别研究局限性,结合大数据环境下抽油机井生产特点,为进一步提高示功图凡尔开闭点识别精准性和实用性,提出一种基于八方向链码形状匹配的抽油机井地面示功图凡尔开闭点识别方法。首先对理论示功图和凡尔开闭点物理表示意义进行分析,用自定义的八方向链码对预处理后的地面示功图进行轮廓提取,再建立形状匹配库,在设定的迭代次数内进行轮廓光滑和链码压缩直至形状匹配成功,最后根据匹配形状定位规则定位凡尔开闭点。应用该方法对胜利油田某区块1650个地面示功图进行凡尔开闭点识别,并将定位的凡尔开闭点运用于60口抽油机井11种工况的识别,基于该方法的识别效果比已有的基于四方向链码多边形近似凡尔开闭点识别方法提高了近1%。结果表明精确识别凡尔开闭点有利于提高抽油机井的故障诊断率和产液量功图计算精准率,从而验证了该方法的有效性。

Valve Opening and Closing Points Recognition in Surface Dynamometer Cards from Sucker-rod Pumping Wells Based on 8-direction Chain Code and Shape Matching

To accurately recognize the valve opening and closing points of dynamometer cards is of great significance to fault diagnosis and oil yield calculation about dynamometer cards from sucker-rod pumping wells. Aiming at the limitations of valve opening and closing points recognition research and the oil production characteristics in the context of big data, and further improving the recognition accuracy and practicality, a novel method based on 8-directions chain code and shape matching to identify the valve opening and closing points of surface dynamometer cards from sucker-rod pumping wells is proposed. Firstly, by analyzing the theoretical dynamometer card and the physical expression meaning of valve opening and closing points, the contour of the preprocessing surface dynamometer card was extracted by 8-direction chain code. Then the shape matching database was established, and the extracted contour was smoothed and the chain code was compressed in a certain number of iterations until the shape matches with the database shape data. Finally, the valve opening and closing points of surface dynamometer cards were located by the positioning rule of the matched shape. The proposed method was applied to the valve opening and closing points recognition of 1650 surface dynamometer cards in a block in Shengli Oilfield, and the located valve opening and closing points were applied to eleven kinds of the working condition recognition from sixty sucker-rod pumping wells. Compared with the working condition recognition methods based on the proposed method and based on the existing 4-direction chain code and polygonal approximation method, the working condition recognition rate is improved by about 1%. The result shows that the accurate valve opening and closing points recognition can improve the fault diagnosis rate and the liquid production capacity calculation accuracy rate by dynamometer cards, thus validates the effectiveness of this method.