自组织竞争神经网络在江苏油田有杆抽油系统故障诊断中的应用

有杆抽油系统的故障诊断技术是国内外采油工程技术中的一个重要研究课题。通过示功图的不同形状特征可以反映抽油机的不同工作状态。将自组织竞争神经网络应用于示功图的识别与分类，建立了一个自组织竞争神经网络模型对示功图进行自动聚类，从而实现故障诊断的自动化。应用江苏油田的实测示功图数据进行实验，可以看出自组织竞争神经网络具有良好的分类能力和泛化性能，是实现油田抽油系统故障诊断的有效技术，具有很强的实用价值和广泛的应用前景。

Application of Self-Organizing Competitive Neural Network to Fault Diagnosis of Sucker Rod Pumping System in Jiangsu Oilfield

Fault diagnosis of suck rod pumping system is an important research subject of oil extraction engineering. By using the computer-aided diagnosis system, we are able to learn the precise real-time information about the status of the sucker rod pumping system, which is important to the realization of long distance remote control of oil extraction and the improvement of economic interest of petroleum industry. In practice, the down-hole conditions are mostly observed by dynamometer cards. Down-hole conditions of suck rod pumping system are reflected by different shapes of dynamometer cards. Fault diagnosis of sucker rod pumping system is actually a pattern recognition and classification technique. Self-organizing competitive neural network is used in this paper to classify dynamometer cards in fault diagnosis of sucker rod pumping system. A self-organizing competitive neural network model is constructed to achieve automatization of fault diagnosis by automatic clustering of dynamometer cards. The model was applied to data measured from Jiangsu oilfield. Results from experiments show that the model has good classification capability and generalization capability. In conclusion, self-organizing competitive neural network is an effective method to automatically diagnose fault of sucker rod pumping system. It has great practical value and good application future.