新型RL模型及其在抽油机节电控制中的应用  

A new RL model and its application to power control of well pumping unit

作　　者：丁宝[1] 张延丽[2] 孙立峰 

机构地区：[1]哈尔滨工业大学电气工程与自动化学院,黑龙江哈尔滨150006 [2]哈尔滨工业大学机电工程学院,黑龙江哈尔滨150006 [3]北京威英智通科技有限公司,北京100085

出　　处：《石油学报》2003年第3期110-112,共3页Acta Petrolei Sinica

摘　　要：针对低产井采油过程中游梁式抽油机存在的“泵空”和“干抽”问题 ,采用以新型RL模糊神经网络模型为核心的电动机间歇启停控制方案 ,因而减少了抽油机的无效行程 ,节约了大量电能。该控制系统具有良好的自适应性能 ,在环境不断变化的过程中 ,自动调节自身参数及结构 ,依据采样数据和模糊专家数据库 ,确定出合理的停机时间。基于该模型开发的抽油机节电控制器已在大庆、胜利、辽河、中原、大港各大油田采油厂投入使用。实践表明 ,这种控制器工作可靠 ,抽油机起停时间合理 ,在保证采油量的前提下 ,节电率达 3 0 %以上 ,实现了抽油机控制的自动化与智能化。There are the problems of “pump emptiness” and “evacuation” in rocker pumping unit during petroleum exploitation of the low-production well.So,a new RL fuzzy neyral network model for controlling the interim operation of pumping unit was presented.This scheme can decrease the backlash of pumping unit and save much electrical energy.The control system possesses the favorable self-adaptability.With the change of environment,it can regulate the parameters and structure of itself and determine the reasonable idle hour according to the sampled data and fuzzy-expert database.The controller of pumping unit based on the model has been successfully used in the oil recovery plants of Daqing,Shengli,Liaohe,Zhongyuan and Dagang Oil Fields.Experiments showed that the works of controllers and the start-stop time of pumping unit were reasonable.The energy-saving rate ran up to 30 percent with the precondition of oil output equality.The automatic and intelligent control of well pumping unit may be realized.

关 键 词：游梁式抽油机 智能控制器 模糊神经网络 模糊Takagi—Sugeno模型 模糊专家数据库 

分 类 号：TE93[石油与天然气工程—石油机械设备]