In order to diagnose the fault of the pump oil well, a fault diagnosis method based on the wavelet packet and the neural network is introduced. Firstly, the dynamometer card of the pumping well is gathered and normalized, then these data are decomposed by using three layers of wavelet packet. It makes up eight energy eigenvectors which are regarded as the input eigenvector of the RBF network. The experiment indicates that the method can not only detect the fault of the pumping oil well but also can recognize the fault type of it. It shows that the method is very effective for safety protection and fault diagnosis in the pumping oil well.

No abstract

The surface dynamometer cards are important working condition data of sucker-rod pumping system. It has a very important practical significance for the analysis of transmission system and the diagnosis of oil production condition of sucker-rod pumping system. The pump dynamometer cards are important reference for the diagnosis of oil production condition, and its key technology is the identification of pump dynamometer cards. A new similar pattern recognition algorithm based on root-mean-square error (RMSE) is proposed, a theoretical model of the similarity matching algorithm based on RMSE is established, and the algorithm is studied and analyzed. The three-dimensional vibration mathematical models for the surface dynamometer cards are created, by which the surface dynamometer cards can be transformed to the pump dynamometer cards. The accuracy, reliability and stability between the algorithm of RMSE similarity matching and the classical algorithms of similarity pattern matching are studied. The research shows that the resistance to the graphics deformation of RMSE algorithm is the highest among all algorithms. The application of RMSE algorithm and classic similarity matching algorithms to the identification of real pump dynamometer cards and the fault diagnosis of oil wells indicates that the RMSE algorithm has very high identification reliability and accuracy. The remarkable feature of the RMSE algorithm is that it has very high identification accuracy for small difference, while the classical similarity matching algorithms do not have this feature.

No abstract

Co-Kriging (CK) modeling provides an efficient way to predict responses of complicated engineering problems based on a set of sample data obtained by methods with varying degree of accuracy and computation cost. In this work, the Gaussian random process (GRP) is introduced to construct a novel combination CK model (CK-GRP) to improve the prediction accuracy of the conventional CK model, in which all the sample information provided by different correlation models is well utilized. The features of the new model are demonstrated and evaluated for a numerical case and an engineering application. It is shown that the CK-GRP model proposed in this work is effective and can be used to improve the prediction accuracy and robustness of the CK model.