面向多源不确定性数据的往复压缩机决策级融合诊断方法  

Decision-level fusion diagnosis method of reciprocating compressor facing multi-source uncertain data

作　　者：张宇婷 段礼祥[1,2] 李兴涛 张馨月 ZHANG Yuting;DUAN Lixiang;LI Xingtao;ZHANG Xinyue(College of Safety and Ocean Engineering,China University of Petroleum(Beijing),Beijing 102249,China;Key Laboratory of Oil and Gas Production Safety and Emergency Technology,Ministry of Emergency Management,Beijing 102249,China;China National Oil and Gas Exploration and Development Corporation,Beijing 102249,China)

机构地区：[1]中国石油大学(北京)安全与海洋工程学院,北京102249 [2]应急管理部油气生产安全与应急技术重点实验室,北京102249 [3]中国石油国际勘探开发有限公司,北京102249

出　　处：《中国安全生产科学技术》2024年第9期112-119,共8页Journal of Safety Science and Technology

基　　金：中国石油天然气集团有限公司科学研究与技术开发项目(ZLZX2020-05-02)。

摘　　要：为解决不确定性高的数据源使多源信息融合诊断模型精度降低的问题,提出1种面向不确定性数据的往复压缩机决策融合诊断方法。构建基于GRU-AlexNet网络的初步诊断模型,得到往复压缩机各传感器信号的初始诊断结果,并引入余弦相似度与置信熵的概念构建联合指标改进传统DS证据理论,结合初步诊断结果进行多源信号决策融合诊断。研究结果表明:在对往复压缩机故障的加速度、位移、压力信号(不确定性数据)融合诊断试验中,融合诊断准确率高达99.98%,相较于单一信号源诊断结果分别提高约9.27,5.13,48.30个百分点。该方法可在较大程度上降低不确定性信息对于融合诊断结果的影响,具有良好的容错性与稳定性,可有效提高往复压缩机使用过程中各类故障识别的准确性,进而提高设备的稳定性,保证其良好工作状态。研究结果对保障相关企业安全生产、提高设备产出能力具有重要参考意义。In order to solve the problem that the accuracy of multi-source information fusion diagnosis model is reduced due to high uncertain data sources,a decision fusion diagnosis method of reciprocating compressor facing uncertain data was proposed.A preliminary diagnosis model based on the GRU-AlexNet network was constructed to obtain the initial diagnosis results of each sensor signal of the reciprocating compressor,then the concepts of cosine similarity and confidence entropy were introduced to build a joint index to improve the traditional DS evidence theory,and the multi-source signal decision fusion diagnosis was carried out combining with the preliminary diagnosis results.The results show that in the experimental study on the acceleration,displacement,and pressure signal(uncertain data)fusion diagnosis of reciprocating compressor faults,the fusion diagnosis accuracy was 99.98%,which was 9.27,5.13,and 48.30 percentage points higher than the single signal source diagnosis results,respectively.This method greatly reduces the influence of uncertain information on the fusion diagnosis results,and has good fault tolerance and stability.It can effectively improve the accuracy of various types of fault identification,thereby improving the stability of equipment and ensuring its good working condition.It is of great significance to ensure the work safety of enterprises and improve the output capacity of equipment.

关 键 词：往复压缩机 智能诊断 不确定性数据 多源信息融合 DS证据理论 

分 类 号：X937[环境科学与工程—安全科学] TH17[机械工程—机械制造及自动化]