Research Into Expert System Mechanism On Computing Remaining Oil

As the key factor to improve recovery ratio and development effectiveness, analysis of remaining oil plays an important role in reservoir management. Along with high-speed development of oilfield, especially enters into high water-contained stage, there were some phenomena occurred: the distribution of remaining oil got scattered; condition under the shaft changed significantly; contradiction between inner layer broke out; the result of production logging contained much more water than before. Because of those phenomena, analysis of remaining oil saturation became a hard nut to crack.There are many methods for computing remaining oil saturation, but in view of different geological condition, it requires abundant artificial experience and expert knowledge. This thesis combines expert system with analysis of remaining oil and realizes achievements below: firstly, knowledge base has been built up; secondly, according to analytic method, control area method and sandstone area method all based on layer quantitative analysis method, it accomplishes analysis, calculation and adjustment of remaining oil; finally, users can view distribution of remaining oil both in contour diagram and data sheet.We have studied theory about rule-based expert system and frame-based expert system in the designing of expert knowledge base and confirm the thought of designing knowledge base which adopted rule-frame knowledge representation method. Rules are divided into data rules, control rules and correct rules. Data rules are the core of reasoning model, which describe status of active data and show the influence they produced. They are composed by main rules, slot aggregate and sub-rules three parts. Control rules can extract effective data rules and replenish declaration of them. Correct rules are the rules in charge of correcting remaining oil' s data.Inference mechanism algorithm selection of remaining oil has been set up in this thesis, mainly involving judgment upon precondition and definition of evaluation architecture. Decrease coefficient of confidence level is put forward which make the evaluation more creditable. Finally, combining rule base, we realize selective preference of analysis and computing remaining oil for guaranteeing intelligence and extendibility of remaining oil algorithm.According to the model we mentioned above, we implement analysis, computing and drawing of remaining oil with daqing bachang oil recovery, which provide strongly data foundation for decision maker.