基于多智能体钻井故障诊断系统的研究

发布时间：2018-10-10 17:41

【摘要】：安全是石油开采质量的重要保证,石油钻井是石油开采的重要组成。然而,由于人类的大肆开采,导致现存的石油资源常常处于复杂的地质结构环境之中。因此,油田勘探和钻井的难度越来越高,而伴随钻井时发生的故障也越来越频繁。时常发生的钻井故障不但会极大地延长开采时间、增加开采成本,而且还会带来人员伤亡、经济损失。因此,如何合理地、科学地、实时地诊断出钻井故障来达到排除故障、降低成本与风险是一个重要的、有意义的研究课题。本文针对石油钻井系统的复杂性、诊断方法的多样性以及故障的并发性,利用多智能体系统(MAS, Multi-Agent System)的分布性、实时性、并行性和协作性,提出了基于MAS的钻井故障诊断系统的功能结构总体框架。该框架以钻井系统为研究对象,结合了多智能体故障诊断技术,能够合理地、科学地、实时地诊断出钻井故障,为现场专家提供有力的决策支持,以此来排除钻井故障。此外,本文分析了以钻井工艺为核心的基础理论,研究了钻井过程中特征参数的获取方法、常见故障类型及其发生机理,为石油故障诊断系统的设计与实现打下理论基础。在MAS故障诊断框架中,我们共研究并提出了五大智能体,每种智能体都有其自身的定位和功能。这五大智能体包括：任务分解、任务分配、任务诊断、协作和决策智能体。其中,任务分解智能体根据多层次全方位分解策略,将钻井的总体任务分解为若干子任务；任务分配智能体根据改进的合同网分配规则,将这些子任务分配给诊断智能体；而诊断智能体根据各自不同的诊断方法(如故障树、模糊推理等诊断方法),对钻井故障进行诊断并将诊断结果发送给决策智能体；最后,决策智能体根据仲裁知识库中的规则和方法,对诊断结果进行整合并作为最终结果向用户输出。在MAS故障诊断框架的基础上,本文设计了钻井诊断系统的软件结构以及系统中的数据显示模块、石油钻井参数设置模块、数据库连接模块、故障诊断模块等。完成了五大智能体的软件实现,最后将多智能体诊断技术应用到软件开发中,实现了钻井工程的故障诊断功能。以丰富的理论研究为基础,证实了本文所提出的系统的可行性和科学性,为该系统的进一步完善和最终的实现打下了夯实的基础,指明了方向。
[Abstract]:Safety is an important guarantee of petroleum exploitation quality, and oil drilling is an important component of petroleum exploitation. However, the extant petroleum resources are often in complex geological environment due to human exploitation. Therefore, the exploration and drilling of oil fields are becoming more and more difficult, and the faults that occur during drilling are becoming more and more frequent. The frequent drilling failure will not only prolong the mining time and increase the mining cost, but also bring casualties and economic losses. Therefore, it is an important and meaningful research topic how to diagnose drilling faults reasonably, scientifically and in real time to get rid of them and reduce the cost and risk. Aiming at the complexity of oil drilling system, the diversity of diagnosis methods and the concurrency of faults, this paper makes use of the distributivity, real-time, parallelism and cooperation of multi-agent system (MAS, Multi-Agent System). The functional framework of drilling fault diagnosis system based on MAS is presented. This framework takes drilling system as research object and combines multi-agent fault diagnosis technology. It can diagnose drilling faults reasonably, scientifically and in real time, and provide strong decision support for field experts to eliminate drilling faults. In addition, this paper analyzes the basic theory based on drilling technology, studies the methods of obtaining characteristic parameters, common fault types and their occurrence mechanism in drilling process, and lays a theoretical foundation for the design and implementation of oil fault diagnosis system. In the framework of MAS fault diagnosis, we have studied and proposed five agents, each agent has its own location and function. These five agents include task decomposition, task allocation, task diagnosis, collaboration and decision agent. The task decomposition agent decomposes the overall task of drilling into several sub-tasks according to the multi-level and omni-directional decomposition strategy, and the task allocation agent assigns these sub-tasks to the diagnostic agent according to the improved contract net assignment rules. According to their different diagnosis methods (such as fault tree, fuzzy reasoning and so on), the diagnosis agent diagnoses the drilling fault and sends the diagnosis result to the decision agent. According to the rules and methods in the arbitration knowledge base, the decision agent integrates the diagnosis results and outputs them to the user as the final results. Based on the MAS fault diagnosis framework, this paper designs the software structure of the drilling diagnosis system and the data display module, the oil drilling parameter setting module, the database connection module, the fault diagnosis module and so on. Finally, the multi-agent diagnosis technology is applied to the software development, and the fault diagnosis function of drilling engineering is realized. Based on the abundant theoretical research, the feasibility and scientific nature of the system proposed in this paper are confirmed, which lays a solid foundation for the further improvement and final realization of the system and points out the direction.
【学位授予单位】：西南石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TE28;TP18

【参考文献】