【作者】 罗旋；

【导师】 王世杰； 吕晓仁；

【作者基本信息】 沈阳工业大学 ， 机械设计及理论， 2013， 博士

【摘要】 螺杆泵采油是一种新型的人工举升方式，具有结构简单、高效节能、机组占地面积小等特点，尤其适宜于高粘度、高含砂量、高油气（水）比原油开采，已经在国内外的油田生产中普遍使用。在螺杆泵采油过程中，转速的选择十分重要，直接影响到泵的效率和使用寿命。如果转速选择不合理，将引发油井抽空“烧泵”、泵效下降以及使用寿命缩短等诸多问题。目前，国内外已经有研究人员采用建立最佳转速模型的方法来选择合理的转速，其不足之处在于没有考虑泵的结构参数和油井工况的耦合影响，而且所建立的模型收敛速度慢、精度低；没有对转速的智能控制问题进行深入研究，对提高泵效、延长寿命的效果仍不明显。为此，本文从理论和实验两方面开展了采油螺杆泵转子转速优化方法的研究，旨在进一步提高螺杆泵采油作业中转速控制技术水平。论文首先分析了螺杆泵结构和工作原理，综述了螺杆泵采油技术的国内外发展状况，重点对螺杆泵采油技术领域中存在的关键技术问题进行了剖析，确定了论文工作的切入点。速度是影响螺杆泵定子橡胶寿命的主导因素，研究他们之间的关系正是论文工作的核心。但是速度对橡胶磨损的影响又受制于螺杆泵结构参数和油井工况参数，这种影响呈非线性耦合形态。对此，文中在梳理了两大类影响因素的基础上，对解决非线性耦合关系问题的方法进行了考证，提出基于人工神经网络技术建立螺杆泵转速优化模型的思路，选取原油温度、原油粘度、螺杆泵泵端压差和螺杆泵定子橡胶磨损量作为输入量，以螺杆泵转速作为输出量，模拟实际工况并以其为参考考察优化效果。文中阐述了基于四种典型的人工神经网络模型（BP网络、RBF网络、Elman网络和GA-BP网络）进行优化的具体算法及优化结果与实际值的对比分析过程，从中遴选了比较理想的优化模型。在螺杆泵采油系统的实际工作过程中，转速需根据实际工况的变化而变化。为实现螺杆泵转速的实时控制，自行开发了螺杆泵转速优化系统。文中介绍了软硬件开发环境、软件中内含的算法以及基于此平台对螺杆泵转速及其影响因素的测量和实时调控等操作过程。为验证螺杆泵转速优化模型的有效性，并对转速优化问题深入研究，自行设计并研制了能够模拟实际油井工况的螺杆泵转速优化结果检验平台。该实验平台以环-块式摩擦磨损试验机为主体，结合所开发的螺杆泵转速优化系统，能够完成螺杆泵转速优化模型实效性的验证以及不同控制方式实效性的试验研究。在螺杆泵转速优化效果检验平台上，对基于人工神经网络优化模型获得的螺杆泵转速优化结果进行了实验分析。采用不同试验方案（橡胶配方）进行了试验，对比分析实验结果可以看出，基于BP神经网络优化模型优化出的转速对减轻螺杆泵定子橡胶磨损量的效果比较明显。最后，对进一步完善基于人工神经网络的螺杆泵转速优化平台提出了建设性思路，给出了优化系统的开放性架构和与实际信号采集子系统的接口。更多还原

【Abstract】 As a new artificial lift method, progressing cavity pump (PCP) oil production hasbeen widely used in the oil fields at home and abroad for its characteristics includingsimple structure, high efficiency energy saving, small floor area and so on. PCP isespecially suitable for the oil well of production high viscosity oil, high sand content andhigh gas content. The choice of rotor speed of PCP speed is very important in the oilproduction; the rotor speed will directly affect the pump efficiency and service life. If theselection of rotor speed is unreasonable, it will cause “burn pump”, lower pump efficiency,shorten service life and other problems. Recently, researchers at home and abroad haveused the method of establishing the best speed model to choose reasonable speed. Thedisadvantage of the best speed model is that it does not consider the coupling influence ofgeometric parameters and structure parameters, and the convergence speed of establishedmodel is slower and accuracy is lower. Meanwhile, the researchers have not further studyin the problem of speed intelligent control, and the effect is still not obvious to improvepump efficiency and prolong life. Therefore, this paper studied PCP speed optimizationmethod from the theory and experiment both aspects, and trying to improve the level ofPCP speed control technology.In the first place, the PCP structure and its working principle were analyzed in thearticle. Then, development of the PCP technology both at home and abroad wassummarized. The key problems of oil production technology of PCP were deeply discussedand the breakthrough point of this work was determined.Speed is the dominant factor of affecting the service life of PCP stator rubber;therefore the core of this paper was study the relationship between them. However, theinfluence of speed on rubber wear depends on structure parameters and working conditionparameters and the influence shows a nonlinear coupling relationship. Therefore, on thebasis of combing the two groups of influence factors, the paper made a textual research forthe method of solving nonlinear coupling. The thought of establishing optimization model of PCP speed on the basis of artificial neural network was proposed. In order to simulatethe actual working condition and investigate optimization effect, the oil temperature, oilviscosity, pressure difference of pump ends and wear loss of stator rubber were selected asinput, and PCP speed as output. Four typical artificial neural network models (BP network,RBF network, Elman network and GA-BP network) were described in the paper. From theanalysis results between optimization results and actual value, the ideal optimization modelwas selected.The rotor speed needs to be changed according to the changes of working conditionsin the process of oil production. In order to realize the real-time control for speed, a PCPspeed optimization system was developed. The development environment of hardware andsoftware, algorithm in the software, measurement and real-time control for PCP speed andits influence factors utilizing this system were introduced in the paper.In order to verify the effectiveness of PCP speed optimization model and make furtherstudy on speed optimization, a test platform of PCP speed optimization which can simulatethe actual well conditions was self-designed and developed. The main structure of the testplatform was ring-block type wear testing machine and combining with the developedspeed optimization system, it can complete the effectiveness verification of PCP speedoptimization model and the experimental research for different control modes.On the test platform, PCP speed optimization results obtained by artificial neuralnetwork optimization model were analyzed in experiment. Experiments with differentschemes were carried on, and from the results it can be seen that the wear loss of statorrubber was obviously reduced with the speed which is optimized by the BP neural networkoptimization model.Finally, this paper proposed a constructive idea to improve PCP speed optimizationplatform based on artificial neural network. Moreover, it gave the open architecture ofoptimization system and the subsystem interface between actual signal acquisitions.更多还原