基于数据挖掘的油管柱腐蚀预测预警问题研究与系统开发

发布时间：2018-03-22 03:26

本文选题：腐蚀预警　切入点：数据挖掘　出处：《西安电子科技大学》2015年硕士论文　论文类型：学位论文

【摘要】：油管柱腐蚀是影响石油开采时间成本以及物资成本的一个重要因素。油管柱腐蚀会降低油管柱的工作寿命,降低石油开采效率。如何准确预测工作环境中油管柱腐蚀速率,更换与维护油管柱从而降低石油开采的时间与物资成本,是石油开采行业的一个重要问题。如果能从大量的油管柱腐蚀试验数据中发现出油管柱的腐蚀规律,可以减少大量人力物力资源的消耗。由于目前在油管柱腐蚀预测领域上对腐蚀数据的处理大多数使用传统的方法,分析维度低,大量腐蚀数据的信息没有得到有效的利用。本文引入数据挖掘技术,利用这一技术提取腐蚀数据的有效信息。本文首先对油管柱腐蚀预警问题进行分析,总结油管柱腐蚀预警特征,然后对油管柱腐蚀预警需要解决的问题进行分析,总结出解决油管柱腐蚀预警问题至少需要满足要求,在此基础上,提出运用数据挖掘技术解决油管柱腐蚀预警问题的方法。方法的基本思想是通过数据挖掘方法分析油管柱腐蚀试验数据,得出腐蚀规律,在使用中把油管柱工作环境参数传递给腐蚀规律,计算腐蚀速率,及时反馈给管理人员。本文分别将朴素贝叶斯算法,贝叶斯网络算法等贝叶斯算法,C4.5算法,随机森林算法等决策树算法,以及人工神经网络算法应用于油管柱腐蚀规律发现,预测和预测模型的创建,并对其效果进行分析。分析结果显示,基于人工神经网络算法的数据挖掘方法更适合解决油管柱腐蚀预警问题。本文对数据挖掘技术中的人工神经网络算法进行了详细的研究,分析其预测模型建立以及训练过程,得出人工神经网络对任意函数有很好的逼近能力,通过调整优化参数可以很好地应用在油管柱腐蚀试验数据挖掘中。本文将人工神经网络算法应用到油管柱腐蚀预警数据挖掘程序中。油管柱腐蚀预警数据挖掘程序是油管柱腐蚀预警系统的预测模块,通过对油管柱腐蚀数据进行分析并处理,建立预测模型并对需要预测的环境条件进行油管柱腐蚀速率预测。本文基于数据挖掘技术与Weka工具使用人工神经网络模型对油管柱腐蚀预测数据挖掘程序进行开发,利用Tomcat,Maven和Nexus工具,对其Web程序进行开发。为验证研究成果的可行性与有效性,本文设计并开发了一个基于数据挖掘的油管柱腐蚀预测程序,介绍了系统的架构设计、流程设计、程序设计和关键代码实现。应用效果表明,本文提出的数据挖掘技术Web技术解决油管柱腐蚀预警问题的方法是可行有效的。
[Abstract]:The corrosion of tubing string is an important factor affecting the time cost and material cost of oil production. The corrosion of tubing string will reduce the working life of tubing string and reduce the efficiency of oil production. How to accurately predict the corrosion rate of tubing string in working environment? Replacing and maintaining tubing string so as to reduce the time and material cost of oil production is an important problem in petroleum production industry. If the corrosion law of tubing string can be found from a large number of corrosion test data of tubing string, It can reduce the consumption of a lot of human and material resources. Because the traditional method is used to deal with the corrosion data in the field of pipeline corrosion prediction, the dimension of analysis is low. The information of a large number of corrosion data has not been effectively utilized. In this paper, data mining technology is introduced to extract the effective information of corrosion data. This paper summarizes the characteristics of tubing string corrosion warning, then analyzes the problems that need to be solved, and concludes that the problem of tubing string corrosion warning should at least meet the requirements. This paper puts forward a method of using data mining technology to solve the problem of pipe string corrosion warning. The basic idea of the method is to analyze the corrosion test data of tubing string by data mining method, and to obtain the corrosion law. In use, the operating environment parameters of tubing string are transferred to corrosion law, corrosion rate is calculated, and feedback is given to managers in time. In this paper, Bayesian algorithms such as naive Bayes algorithm, Bayesian network algorithm, etc. The decision tree algorithm such as stochastic forest algorithm and artificial neural network algorithm are applied to the discovery, prediction and prediction of the corrosion law of tubing string, and the effect of the model is analyzed. The data mining method based on artificial neural network algorithm is more suitable to solve the problem of pipeline corrosion warning. In this paper, the artificial neural network algorithm in data mining technology is studied in detail, and its prediction model establishment and training process are analyzed. It is concluded that the artificial neural network has good approximation ability to arbitrary functions. In this paper, the artificial neural network algorithm is applied to the data mining program of tubing string corrosion early warning data mining. Sequence is the prediction module of tubing string corrosion warning system. By analyzing and processing the corrosion data of tubing string, Based on the data mining technology and Weka tool, the pipeline corrosion prediction data mining program is developed by using artificial neural network model. In order to verify the feasibility and validity of the research results, a pipeline corrosion prediction program based on data mining is designed and developed in this paper. The architecture design and flow design of the system are introduced. Program design and key code implementation. The application results show that the method of data mining technology Web proposed in this paper is feasible and effective to solve the problem of pipe string corrosion warning.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TE983;TP311.13

【相似文献】