石油炼制与化工 ›› 2023, Vol. 54 ›› Issue (5): 101-107.

• 控制与优化 • 上一篇    下一篇

基于SVDD的减压直拔生产沥青的原油种类挖掘方法

董叶伟,秦康,陈莹,吴昊,何跃,芦泽龙   

  1. 中石化石油化工科学研究院有限公司
  • 收稿日期:2022-09-01 修回日期:2023-01-30 出版日期:2023-05-12 发布日期:2023-05-12
  • 通讯作者: 秦康 E-mail:qinkang.ripp@sinopec.com
  • 基金资助:
    工信部公共服务平台项目

TPYE SELECTION OF CRUDE OIL FOR STRAIGHT-RUN ASPHALT PRODUCTION BASED ON SVDD METHOD

#br#   


  • Received:2022-09-01 Revised:2023-01-30 Online:2023-05-12 Published:2023-05-12
  • Contact: Kang Qin E-mail:qinkang.ripp@sinopec.com

摘要: 减压直拔生产沥青是生产道路沥青的主要方式,其经济性与能效性具有显著优势,但该生产过程对其原油的选择通常依赖定性分析及历史经验,缺乏完备的理论数据支持。通过收集大量沥青生产成功案例,提取对应原油的物性数据并进行处理与分析,结合支持向量数据描述(SVDD)单值分类器特性,建立基于SVDD的减压直拔生产沥青的原油种类识别模型。基于该模型对中国石化原油数据库中273种原油进行数据挖掘与识别,最终得到46种适合通过减压直拔方式生产沥青的原油,并通过可行性分析明确了单值分类器在油品数据挖掘中的有效性,为炼化企业经营决策提供了理论支持,明确了生产沥青的原油选择空间,降低原油试错成本,提高沥青生产的可行性。

关键词: 沥青生产, 原油, 数据挖掘, 支持向量数据描述, 分类器

Abstract: Straight-run asphalt production is the main way to produce road asphalt, which has significant advantages in economy and energy efficiency. However, the selection of crude oil in this production process usually depends on qualitative analysis and historical experience while lacking complete theoretical data support. Through collecting a large number of successful cases of asphalt production, extracting the physical properties data of corresponding crude oil and processing and analysis, combined with the SVDD single-value classifier characteristics, a support vetor data description(SVDD-based) model for identifying the types of bituminous crude oil produced by straight-run asphalt was established. Based on the model, 273 types of crude oil in the SINOPEC crude oil database were mined and identified, and 46 types of crude oil suitable for straight-run asphalt production were finally obtained. Furthermore, through feasibility analysis, the validity of single-value classifier in crude oil data mining was confirmed, which provided theoretical support for the management decision of refinery enterprises. This method not only can clarify the selection space of oil for asphalt production, but also reduces the cost of trial and error of crude oil selection and improves the feasibility of asphalt production.

Key words: asphalt production, crude oil, data mining, SVDD, classifier