Supervised dictionary-based transfer subspace learning and applications for fault diagnosis of sucker rod pumping systems

摘要

Sucker rod pumping wells are systems that their operation states varies slowly. For a relatively new well, it is hard to collect all kinds of fault samples for training. Moreover, samples from different wells are not always have similar distributions, so directly using samples from other wells as the training data may hardly get good results. In this paper, a novel framework is proposed to solve the aforementioned problems. For the source data from one well and the target data from another well, a transform matrix is calculated to transfer these data into a common low dimensional subspace. In this subspace, the source data that contain all kinds of fault samples and the target data that lack some kinds of fault samples can be represented by a shared dictionary matrix. By introducing two idea regularization terms, the structure information of source data and target data are included into the dictionary learning process. So the obtained dictionary has discriminative ability. Extensive experiments are conducted to evaluate the effectiveness of the proposed method. (C) 2019 Elsevier B.V. All rights reserved.