基于相似识别的抽油机井故障检测与诊断研究

Fault detection and diagnosis of oil pumping unit based on similarity feature recognition

抽油机井故障检测与诊断对管控油田平稳生产、保证油田效益意义重大。论文研究了基于相似识别的故障检测与诊断方法,采用融合二分类和三元组损失的深度学习网络提取示功图图形趋势特征,构建抽油机标准特征库与故障特征库并进行更新维护,在检测到与标准特征库不相似的示功图时检索故障特征库以实现诊断。实验结果表明,该方法能够有效检测抽油机井故障,减少误报率,并在检测到异常时检索故障库实现故障诊断。

Fault detection and diagnosis of oil pumping unit is of great significance to maintain the stability and benefits of the oil production processes.In this paper,a fault detection and diagnosis method based on similarity feature recognition is studied.A deep learning network that combines binary classification and triple loss is used to extract the trend characteristics of indicator diagrams.Then,standard feature library and fault feature library are established and updated for each pumping unit,so that indicator diagram which is dissimilar to the standard feature library can be detected and diagnosed by searching its fault feature library.The experimental results indicate that the method can effectively detect the faults of oil pumping unit,reduce the false alarm rate,and retrieve the fault library to diagnosis when an abnormality is detected.