基于支持向量机沉没度预测的潜油泵冲次优化研究  

Optimization Frequency of Stroke Based on Submergence Depth SVM Forecasting of Submersible Pump

作　　者：于德亮[1] 齐维贵[1] 邓盛川[1] 张永明[1] 王新民[2] 李鑫[3] 

机构地区：[1]哈尔滨工业大学电气工程及其自动化学院,黑龙江省哈尔滨市150001 [2]大庆油田采油工程研究院,黑龙江省大庆市163000 [3]东北石油大学机械科学与工程学院,黑龙江省大庆市163000

出　　处：《中国电机工程学报》2011年第27期138-144,共7页Proceedings of the CSEE

基　　金：黑龙江省工信委产业化基金资助项目(08020017)~~

摘　　要：在石油开采过程中,利用直线电机驱动潜油往复泵是一种新型举升方式。这种举升方式可使抽油机具有较好的可控性。在这一背景下,提出一种基于沉没度预测的潜油往复泵冲次优化方法。文中对在采油厂采集的沉没度数据进行时间序列的输入空间重构,基于支持向量机(support vector machines,SVM)建立沉没度预测模型。以抽油机的经济效益为目标,以预测得到的沉没度为参量,对直线电机的冲次进行优化。采用沉没度–冲次子区间匹配方法改进原优化算法,进一步降低原方法的计算量和数据量。改进的优化算法可以更好地适应现场计算机对计算量的限制。该优化方法可有效地提高往复泵的产油量,并避免直线电机长期工作于满载或过载状态。In the process of oil production, the submersible reciprocating pump is driven by linear motor, such pump application is novel lifting method which has been considered as with good controllability. On this background, an optimization method for stroke frequency of submersible reciprocating pump based on submergence depth forecasting was presented. The submergence depth data in field was measured pretreated to form submergence depth time series, and then reconstruction phase space of input samples. The submergence depth forecasting model was established based on support vector machines （SVM）. The economic benefit of pumping unit was used as optimization objective and submergence depth was used as parameters to optimize the stroke frequency of linear motor. In order to meet the need of field devices, the submergence depth and frequency of stroke subinterval method was adopted to establish an improved optimization method to further decrease computational complexity and data quantity. The optimization method of submersible pump can increase oil production and avoid long time full load or overload of linear motor.

关 键 词：潜油往复泵 直线电机 沉没度预测 冲次优化 支持向量机 

分 类 号：TM85[电气工程—高电压与绝缘技术]