摘要：To enhance operational efficiency at natural gas stations, it is essential to establish the functional connection topology of station components. This includes the reliability block diagram and fault tree model during reliability assessment and optimization. The complex technological processes and interactions among various components require accurate and efficient component identification. In this paper, a novel detector for component identification has been established based on the Detection Transformer (DETR). It directly predicts object bounding boxes and categories without post-processing, therefore, enabling end-to-end detection. To test the proposed model, a comprehensive dataset has been generated combining data augmentation. Experimental results showed the well performance of the proposed model in identifying natural gas station equipment. The identification accuracy is 89.94%, which is the state of the art of the existed model. The proposed model can provide effective technical support for safe operation and equipment management, and also shows strong ability to deal with different scenarios and environments. ? The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

ISSN号：1876-1100

卷、期、页：卷1333：43-54

发表日期：2025-01-01

期刊分区（SCI为中科院分区）：无

收录情况：EI（工程索引）,CPCI-S（科技会议录索引）

发表期刊名称：Lecture Notes in Electrical Engineering

通讯作者：梁馨月,何宇轩,覃小东

第一作者：苏怀,张劲军,杨兆铭

论文类型：会议论文

论文概要：梁馨月,苏怀,张劲军,何宇轩,覃小东,杨兆铭，End-to-End Detection for Key Equipment in Natural Gas Station with DETR，Lecture Notes in Electrical Engineering，2025，卷1333：43-54

论文题目：End-to-End Detection for Key Equipment in Natural Gas Station with DETR