基于高分辨率遥感影像的管道高后果区识别方法

1.国家管网集团西气东输公司;2.苍穹数码技术股份有限公司

1.PipeChina West East Gas Pipeline Company;2.KQ GEO Technologies Co. Ltd.

管道高后果区识别工作主要依据GB32167—2015《油气输送管道完整性管理规范》开展，但由于缺乏有效技术手段的支撑，标准中的定量识别判据对实际识别工作的指导作用受限，从而导致识别工作的效率以及识别结果的准确率不高。为此，提出基于高分辨率遥感影像的高后果区识别方法，采用卷积神经网络和条件随机场组合模型，通过管道中心线两侧缓冲区的建立和基于线性参考的管段划分，实现管道沿线建筑物的自动检测与提取，最后，应用几何计算完成全线高后果区识别。利用该方法对某管道高后果区进行了识别，识别的准确率与召回率分别为93％、90％，大幅提高了高后果区识别的效率与准确性。研究成果可为管道高后果区的高效、定量识别提供技术指导，具有良好的应用前景。（图4，表1，参27）

Presently, the High Consequence Areas (HCAs) are mainly identified according to Oil and gas pipeline integrity management specification (GB 32167－2015). However, due to the lack of effective technical means, the quantitative identification criterion in the standard has a limited guidance effect on the actual identification work, thus leading to the inefficiency of identification and the low accuracy of identification results. In this regard, the HCA identification method based on high-resolution remote sensing images was proposed. Meanwhile, the buildings along the pipeline were detected and extracted automatically based on the convolutional neural network and the conditional random field, in combination with the identification of buffer area and the Linear Referencing based pipeline segmentation. Thus, the identification of HCAs along the pipeline was completed through the geometric calculation. Moreover, the HCAs of a pipeline were identified with this method, of which the accuracy rate and recall rate were 93% and 90%, respectively, showing a great improvement in the efficiency and accuracy of HCA identification. Hence, the research results could provide technical guidance to the efficient and quantitative identification of HCAs along the pipeline, and the method has a good application prospect. (4 Figures, 1 Table, 27 References)