基于RBFNN的FDPSO火灾爆炸波及钻井区概率分析  

FDPSO Fire and Explosion Spreading to Drilling Area Probability Analysis Based on RBFNN

作　　者：杨源[1] 余建星[1] 杜尊峰[1] 晋文超[1] 冯加果 蒋天芳 

机构地区：[1]天津大学水利工程仿真与安全国家重点实验室,天津300072 [2]中海油研究总院,北京100027 [3]天津德赛海洋船舶工程技术有限公司,天津300451

出　　处：《天津大学学报》2012年第10期917-923,共7页Journal of Tianjin University(Science and Technology)

基　　金：国家科技重大专项资助项目(2011ZX05026-006);国家自然科学基金创新研究群体科学基金资助项目(51021004);国家自然科学基金资助项目(51179126)

摘　　要：浮式钻井生产储油轮(FDPSO)是一种具备钻井、采油、储存、外输等多项功能的新型浮式生产装置.FDPSO的功能高度集成和密集布置带来了各模块风险的相互影响问题.针对由此引发的一类突出危险事件,即FDPSO装置上外部设备发生的火灾爆炸波及钻井区,展开事件发生概率研究.在定量风险分析的基础上,引入经改进粒子群算法优化的径向基函数神经网络进行概率预测,并验证了该模型的可用性.研究成果有助于设计人员回避繁琐的定量风险分析过程,快速确定不同布置情况下外部火灾爆炸波及钻井区的概率,从而通过合理设计有效控制该风险.Floating, drilling, production, storage and offloading vessel （FDPSO） is a new floating production installa- tion capable of drilling, production, storage and offloading. The highly integrated functions and dense layout on FDPSO make each module interact with the other when risk occurs. This paper studies the occurrence probability of a kind of highly dangerous events resulting from this interaction, namely the adjacent equipment fire and explosion spreading to drilling area. Based on quantitative risk analysis, radial basis function neural network optimized by a modified particle swarm optimizer was introduced to predict the occurrence probability, and the availability of this model was verified. Research results can help designers avoid complex and cumbersome quantitative risk analysis procedure, and quickly determine the occurrence probability of adjacent equipment fire and explosion spreading to drilling area under different layout conditions, so that this risk will be controlled effectively through reasonable design.

关 键 词：浮式钻井生产储油轮 火灾爆炸 径向基函数神经网络 粒子群算法 

分 类 号：TE58[石油与天然气工程—油气田开发工程]