A Study Of Data-Excavating Technique In Oil And Gas Geochemical Exploration

This thesis is focused on the study of data-mining technique in the intelligent decision-making auxiliary system. A data-mining system based on the in-situ investigating information and the in-situ analyzing and testing data is designed and realized. In order to manage the system and select the proper algorithm,an illustration table of the algorithm and a rule-selecting database of the algorithm are proposed in the model management of this system. The model management of the data mining is realized by using the above tables and database,as well as some process control functions,which makes the system universal. The sub-set of the data pretreatment system contains several key algorithms and models,which not only makes it possible to establish data-mining models based on the concise and representative data,but also shortens the data-mining time,and improves the quality of data ming. The property-selecting calculating method of the rough-set is improved and a concept of mutuality between the properties and the conditions is proposed. Based on the information gain concept in the information theory,the selecting and algorithm of the property sub-set is completed,which improves the efficiency of the algorithm of the data ming and the quality of the knowledge discovery. The data-ming system established in this thesis can be used not only in in-situ oil and gas geochemical exploration,but also in many other fields as an auxiliary tool in the decision making process.