摘要
因抽油机井的拖动装置匹配功率影响因素多且无特征,造成部分电动机功率利用率低、能耗高。基于大数据挖掘技术,建立了LSTM神经网络训练模型,利用人工智能算法分类识别,形成不同抽油机机型的拖动匹配技术,按照能耗最优的原则制定了匹配模板,识别出电动机功率匹配不合理井群,并给出不合理单井电动机匹配方案。以高启动双速电动机为例,按电动机匹配方案现场验证106口井,电动机功率匹配符合率达到98.1%,装机功率下降了21.34%,无功节电率达到了22.6%,综合节能率达到了7.34%,为拖动设备匹配及互换提供了技术支撑。
Due to the multiple and characteristic factors affecting the matching power of driving device in pumping well,some motors have resulted in low power utilization and high energy consumption.Based on big data mining technology,the LSTM neural network training model has been established,and artificial intelligence algorithms have been used for classification and recognition to form driving matching technology for different types of pumping units.The matching templates have been formulated according to the principle of optimal energy consumption,identified unreasonable well groups for motor power matching,and provided unreasonable single well motor matching schemes.Taking high startup dual speed motor power as an example,the on-site 106 wells are verified according to the motor matching schemes,and the compliance rate of motor power matching is 98.1%.The installed power has been decreased by 21.34%,a reactive power saving rate has reaches by 22.6%,and the comprehensive energy conservation rate has been up to 7.34%,which provides technical support for matching and exchanging driving equipment.
作者
姜凯馨
JIANG Kaixin(No.4 Oil Production Plant of Daqing Oilfield Co.,Ltd)
出处
《石油石化节能与计量》
CAS
2024年第7期6-10,共5页
Energy Conservation and Measurement in Petroleum & Petrochemical Industry
关键词
抽油机井
神经网络
相关系数
功率匹配
电动机互换
pumping well
neural network
correlation coefficient
power matching
motor exchange
