基于粗糙集和神经网络的油气钻井作业安全评价模型研究

发布时间：2018-12-11 18:44

【摘要】：钻井作业是石油和天然气勘探开发活动中最主要的事故频发区之一。钻井作业现场的隐患数量和人员违章数量常常居高不下,大量的隐患和人员违章极易诱使安全事故的发生。安全事故一旦发生,便会造成人员损失、设备损坏和坏境污染,也会对经济效益和社会效益产生巨大影响。如何保证钻井作业的安全进行,预防事故的发生始终是钻井作业行业需要重点关注的问题。因而,全面辨识和分析油气钻井作业系统中的危险源,了解钻井作业现场的安全状态是必要的。建立一套适合油气钻井作业安全评价模型是当前油气钻井作业所急需解决的问题。本文的研究旨在为钻井作业的安全评价提供有效的评价方法,为钻井作业的安全监管人员提供实时、客观的决策依据。本文的研究是钻井作业安全管理走向科学化、信息化的一种全新探索,对提升钻井公司的安全管理水平有重大的意义。油气钻井作业是一个复杂的系统工程,该系统的最大特点是动态性、随机性和模糊性。影响钻井作业安全的因素众多,各因素之间相互制约。钻井作业进行安全评价是一种非线性问题。考虑到BP神经网络具有很好的非线性映射能力,粗糙集对不完备和不确定信息的强大分析能力,本文采用粗糙集和神经网络来构建钻井作业安全评价模型。本文主要开展一些几个方面的研究：(1)了解安全评价和钻井作业安全评价的国内外研究现状；(2)综合辨识钻井作业过程中的危险源,从人的不安全行为和物的不安全状态两个方面出发对其进行分析,建立钻井作业安全评价指标体系；(3)使用粗糙集和神经网络的松耦合模型做钻井作业安全的定性评价和使用神经网络做钻井作业安全的定量评价。在做定性安全评价时,本文首先使用粗糙集对样本数据做属性约简。然后,基于最小条件属性集选取神经网络的训练样本和测试样本。最后,构建神经网络模型,使用训练样本对其训练,使用测试样本进行预测。之后,本文使用神经网络做了钻井作业安全的定量评价,分别使用训练样本和测试样本对网络进行了训练和测试；(4)钻井作业安全评价模型的设计,主要包括：系统的总体设计,粗糙集模块的程序设计和神经网络模块的程序设计。
[Abstract]:Drilling is one of the most important accident-prone areas in oil and gas exploration and development activities. The number of hidden troubles and the number of personnel violating regulations are always high in drilling operation, and a large number of hidden dangers and personnel violations are easy to induce the occurrence of safety accidents. Once a safety accident occurs, it will cause loss of personnel, equipment damage and environmental pollution. It will also have a great impact on economic and social benefits. How to ensure the safety of drilling operation and how to prevent accidents are always the key issues for the drilling industry to pay attention to. Therefore, it is necessary to identify and analyze the hazard sources in the oil and gas drilling system and to understand the safety state of the drilling site. It is an urgent problem to establish a set of safety evaluation model for oil and gas drilling operation. The purpose of this paper is to provide an effective evaluation method for the safety evaluation of drilling operations, and to provide real-time and objective decision basis for the safety supervisors of drilling operations. The research of this paper is a new exploration of drilling operation safety management, which is scientific and information, and has great significance to improve the safety management level of drilling companies. Oil and gas drilling is a complex system engineering, the system is characterized by dynamic, randomness and fuzziness. There are many factors influencing the safety of drilling operation, and each factor restricts each other. Drilling safety evaluation is a nonlinear problem. Considering that BP neural network has good nonlinear mapping ability, rough set has strong ability to analyze incomplete and uncertain information, this paper uses rough set and neural network to construct safety evaluation model of drilling operation. This paper mainly carries out some research in several aspects: (1) to understand the current situation of safety assessment and safety evaluation of drilling operations at home and abroad; (2) identify the dangerous sources in drilling operation synthetically, analyze the unsafe behavior of human and the unsafe state of objects, and establish the evaluation index system of drilling operation safety; (3) the loosely coupled model of rough set and neural network is used to evaluate the safety of drilling operation qualitatively and quantitatively. In the qualitative security evaluation, the rough set is first used for attribute reduction of sample data. Then, the training samples and test samples of neural network are selected based on the minimum conditional attribute set. Finally, the neural network model is constructed, the training sample is used to train it, and the test sample is used to predict it. After that, the neural network is used to evaluate the safety of drilling operation quantitatively, and the network is trained and tested using training samples and test samples respectively. (4) the design of the safety evaluation model of drilling operation mainly includes: the overall design of the system, the programming of the rough set module and the program design of the neural network module.
【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

【参考文献】