稠油开采方式决策混合智能系统

REALIZATION AND APPLICATION OF THE MIXED INTELLIGENCE SYSTEM IN SCREENING OF VISCOUS CRUDE OIL PRODUCTION SCHEMES

以往稠油开采方式都是通过人工筛选的 ,不同的专家可能得出几种不同的筛选结果 ,而每种结果的好坏难以评价。针对这一情况 ,提出并设计了 1种更为可靠的筛选途径 ,即利用混合智能系统 ,将符号专家系统与人工神经网络紧密地融合在一起 ,通过符号专家系统优化神经网络的结构 ,同时用神经网络获取稠油开采方式筛选所需的参数贡献度知识。该系统实现了知识的模糊表示 ,知识库的层次性组织、知识的模糊推理等。用现场资料进行检验 ,系统筛选结果与实际情况吻合很好。该混合智能系统的开发 ,也为其广泛用于石油勘探开发的其它领域开辟了一条新路。

Traditionally the screening of viscous crude oil production schemes are conducted manually, in which every experts may arrive at different results and the superiority of these are hard to judge. The mixed intelligence system which is more reliable is put forward. In the system, the symbolic expert system (SBWS) and the artificial neural network are integrated, the former being aimed at optimizing the structure of the latter while the latter being used to acquire the contribution degrees of screening of viscous crude oil production schemes' character parameters. With the system, fuzzy expression of expert knowledge, multi-level organization of the knowledge base, fuzzy inference are realized. A satisfactory result was yielded in the field testing. The development of the system also opens a new perspective toward a more extensive utilization of the mixed intelligence system in other areas of oil exploration and development.