摘要
介绍了基于BP神经网络对有杆抽油机井下示功图的一种模式识别方法。首先利用有杆抽油机运动数学模型把井上采集数据变换为井下数据,然后通过几何变换的方法提取井下数据特征值,训练BP神经网络,分析训练效果,最后使用非训练数据样本对识别准确性进行验证。通过仿真数据验证,识别效果较好、速度快。该方法建立的BP网络具有复杂度较低、速度快、效果好的优点。
The method based on BP artificial neural network for recognition of downhole power indicator diagram of oil well rod pumping machine is presented. The up-well data are transferred into down-well data by the motion mathematical model of the rod pumping machine,then the downhole eigenvalues are extracted through a geometric transformation method. The BP neural network is trained, and the training effect is analyzed. Finally, the identification accuracy is verified by the non-training data samples. Through simulation verification,it is proved that this method has high speed and good effect for recognition of downhole power indicator diagram of rod pumping machine based on BP neural network.
出处
《电气传动自动化》
2016年第1期38-42,共5页
Electric Drive Automation
关键词
有杆抽油机
BP神经网络
模式识别
示功图
oil well rod pumping machine
BP artificial neural network
Pattern recognition
power indicator diagram
