游梁式抽油机工况在线诊断系统及方法的研究

发布时间：2018-01-16 08:04

本文关键词：游梁式抽油机工况在线诊断系统及方法的研究　出处：《沈阳工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：游梁式抽油机采油系统在当前石油生产中占有很大比重。抽油机通常工作环境恶劣,工况复杂,系统设备易于损坏,严重影响了油田采油效率和经济效益;所以及时掌握油井工况,对实现采油系统自动监控和科学管理而言十分重要。示功图是判断抽油机井工作状况的重要依据之一,但传统的示功图采集方法存在传感器成本高且易损坏、工序繁杂反馈信息慢等不足。因此,本文在详细分析抽油机采油系统工作特性及损耗的基础上,提出通过电参数间接测量油井示功图的方法,并采用支持向量机技术对抽油机工况进行智能诊断。本文以能量守恒为依据,运用复变矢量法建立抽油机悬点位移的数学模型,并推演出电参数与光杆载荷关系及悬点载荷与位移关系的函数关系式。同时运用Matlab软件对电参数法间接测量抽油机示功图方法进行计算分析。针对示功图的图形特点,运用不变矩方法,提取示功图的7个不变矩特征值,并结合抽油机运行时的基本参数组成综合特征参数,建立综合特征参数样本库。并在此基础上,研究基于支持向量机的油井工况分类识别方法,分析了不同核函数以及不同惩罚因子C与核宽度g的参数组合对分类准确率的影响,并通过LibSVM分别对不变矩特征参数样本和综合特征参数样本进行仿真实验。理论分析和仿真实验表明,电参数法间接测量的等效示功图可以有效反映油井工况;综合特征参数结合支持向量机的游梁式抽油系统工况诊断方法,可以更准确的对油井工况进行识别。
[Abstract]:The oil extraction system of beam pumping unit occupies a large proportion in the current oil production. The pumping unit usually works in a bad environment and complex working conditions, and the system equipment is easy to be damaged, which seriously affects the oil recovery efficiency and economic benefit. Therefore, it is very important to realize the automatic monitoring and scientific management of oil production system to grasp the working conditions of oil wells in time, and the indicator diagram is one of the important basis for judging the working condition of pumping wells. However, the traditional method of collecting indicator diagram has the disadvantages of high cost and easy damage of sensor, slow feedback information and so on. Therefore, this paper analyzes the working characteristics and losses of oil extraction system of pumping unit in detail. A method of indirectly measuring well indicator diagram by electrical parameters is presented, and support vector machine (SVM) technology is used to make intelligent diagnosis of pumping unit working conditions. This paper is based on conservation of energy. The mathematical model of suspension displacement of pumping unit is established by using complex variable vector method. The functional relationship between electric parameter and light rod load and between suspension load and displacement is derived. Meanwhile, the method of indirect measurement of power diagram of pumping unit by electric parameter method is calculated and analyzed by Matlab software. The graphic features of the work graph. The method of moment invariant is used to extract the eigenvalues of 7 invariant moments of the dynamometer, and combined with the basic parameters of the pumping unit to form the comprehensive characteristic parameters, and the sample database of the integrated characteristic parameters is established. The classification and recognition method based on support vector machine (SVM) is studied, and the effect of different kernel function and the combination of different penalty factor C and kernel width g on the classification accuracy is analyzed. The simulation experiments of moment invariant parameter samples and synthetic feature parameter samples are carried out by LibSVM. The theoretical analysis and simulation results show that. The equivalent indicator diagram measured indirectly by the electric parameter method can effectively reflect the working conditions of the oil well. Combining characteristic parameters with support vector machine (SVM), the condition diagnosis method of beam pumping system can be used to identify the working conditions of oil wells more accurately.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TE933.1

【参考文献】