Research On Multi - Agent Drilling Fault Diagnosis System

Security is an important guarantee of petroleum extraction quality, and petroleum drilling is an important part in the oil exploitation. However, due to human unbridled exploitation, the existing oil resources are often under a complex geological structure and environmental conditions. So petroleum exploration and drilling are increasingly difficult, and drilling fault occurs more and more frequently. Frequent drilling fault will not only greatly prolong the production time, increase production costs, it will also lead to casualties and economic losses.This article Aimed at the complexity of the oil drilling system, the diversity of fault diagnosis method and the concurrency of system failures, using multi-agent system (MAS, Multi-Agent System) distribution, real-time, parallelism and collaborative, puts forward an overall framework of the function structure of drilling fault diagnosis system which is based on MAS. This framework focused on drilling system, combined with the multi-agent technology of fault diagnosis, so it could diagnoisis the drilling fault in real-time. Then this framework provided strong decision support for the field experts to exclude the drilling faults. In addition, this article analyzed the basic theory which puts the drilling technology as the core of the basic theory, studied the method to obtain characteristic parameters in the process of drilling, this articled aslo studied the common failure types and its mechanism, laied a theoretical basis for the design of the petrulum drilling fault diagnosis system.In the frame of the MAS fault diagnosis, we total studied and put forward five intelligent agents, each agent has its own positioning and function. The five agents include task decomposition, task allocation, task diagnosis, coordination and decision-making agent. Among them, task decomposition according to the strategy of multi-level comprehensive decomposition, decomposed the overall task of drilling into several subtasks; According to the improved contract net distribution rules, task allocation agent distributed these subtasks to diagnosis agent; And according to their different methods (such as fault tree and fuzzy reasoning diagnosis method), the diagnosis agent diagnosised the drilling faults and sent the results to the decision-making agent; Finally, the decision agent according to the arbitration rules and methods of knowledge base, integrated the diagnosis results, and output the results as the final results to the user.On the basis of MAS fault diagnosis framework, this article designed the software architecture of drilling diagnosis system and data display module of the system, drilling parameter setting module, etc. Completed the realization of five agent software, finally the multi-agent diagnosis technology is applied to software development, realized the function of fault diagnosis of drilling engineering. On the basis of abundant theoretical research, proved the feasibility and scientific of the proposed system in this paper, laid a solid foundation, and pointed out a direction for the further perfecting and final realization of this system.