Research Of Petroleum Job Summary Verification Agent-based Intelligent Analysis System

Petroleum job summary verification system is an application system which is used to collate the content of job summary. A lot of data generated during the oil well production has been stored and processed. The data used for finishing verification indicates the mistakes produced in the period of the job content. In order to find out the regulation of common mistakes from information of different level to improve the efficiency and quality of job summary verification, there is an urgent need appearing some new technique and system to help manager change the vast data into information and knowledge for using. The combination of data mining technique and Multi-Agent technique is applied to Petroleum job summary verification. Petroleum job summary intelligence analysis system based on Agent is designed. The problems such as low efficiency and verification overlook in old system have been overcome by the new system.Petroleum job summary intelligence analysis system is studied in thesis. An application based on the combination of data mining technique and Multi-agent technique is studied and a data mining model based on Multi-Agent technique is brought up. So the intelligence analysis problem in the system is solved. Oilfield water supplying payment warning system for verifying the data mining model is also described.Traditional data mining divides the process of data mining into several necessary phases. The thing in every phase is accomplished by human. Intelligent Agent has such features as adapting to the outside environment, using its own knowledge to solve problems, learning automatically, cooperating with other Agents .These features make it a potential way by constructing Agent to resolve the problem that logic is successive but physics is interrupted.The organization and structure of data mining model are described in detail from the view of constructing Agent. And each of Agent is assigned some kind of task and the function and design pattern are introduced in detail. The effective accomplishment of multi Agent cooperation relies on perfection of cooperation .Agent cooperation is defined from two aspects: cooperation mechanism and message protocol. General segment-whole scheme method modulates the task coordination of Agent on the whole.The message transmission among Agents is finished by extended Agent communication language (KQML).To implement the cooperation of Agents, performative reflecting the purpose and faith of Agent is extended on the basis of the original performative in the system. The meaning analysis of extended performative is made clear.The method of constructing model based on greedy network-best competition is proposed in the process of data mining in thesis. The intelligence and autonomy of Agent is used to implement the function of mining automatically. Human factor is injected into the data mining system and the building of data mining pattern is enriched from the view of concept.