水基动力无杆抽油机故障敏感因子提取方法研究

摘要

Rod-less pumping with water as power as a new technology, has not yet put into use, the research of the fault is not mature. In order to improve the accuracy of separability and diagnosis of fail signal from the pumping oil machine system, aimed at the typical failure of the pumping oil machine, it normalizes the collected fail signals and denoises them with EEMD. A method based on time frequency feature parameter extraction and a failure analysis method based on manifold learning to reduct dimension are studied. Finally, it makes use of the difference between distance and Manifold to ifnd out the sensitive factor. The results show that this method can extract the sensitive factor of the failure of the pumping unit, and lay the foundation for the following research.%水基动力无杆抽油机作为一项新技术，尚未大规模的投入使用，对其故障的研究还不成熟。为了提高抽油机系统故障信号的可分性和诊断正确率，针对抽油机的典型故障，对采集到的故障信号进行数据归一化处理以及EEMD降噪。研究了基于时频域特征参数提取的方法以及基于流形学习的降维故障诊断方法。最后提出采用距离差异度和流形差异度寻找故障的敏感因子。研究结果表明，此种方法能够提取抽油机故障的敏感因子，为后续的研究奠定了基础。