Research On Fault Diagnosis Of Beam Pumping Unit Based On FBH-SC Algorithm

The method of sucker-rod pumping is used in a large proportion of the oil extractioa Beam pumping unit is one of sucker-rod pumping. A majority of pumping units are dispersed in wildness area which has terrible environmental conditions.Moreover, the working condition of pumping units beneath the well is complex, so it is difficult to diagnose and handle the faults of pumping units in time.It affects the output and economic benefit of oil field severely. Hence, it is important to analyzer the working condition of pumping units timely for improving the productivity of oil extraction and the economic profit. In the process of oil field production, dynamometer card is used to analyze the working condition of pumping units as the principal reference.Extracting feature vector from dynamometer cards and using unsupervised lea lning algorithm to learn with extracted feature vector datasets to diagnose fault is solution of this paper) Firstly, dynamometer cards of ground are translated to dynamometer cards of underground. Secondly, it is necessary to normalize dynamometer cards of underground. Lastly, this paper uses two feature extraction methods which were proposed by our research group to construct two feature vector datasets.To make use of advantage of SC algorithm applies to any data distribution and Black Hole(BH) algorithm is not sensitive to initial clustering center, this paper proposes Black Hole Spectral Clustering(BH-SC) algorithm and avoids disadvantage of SC algorithm is sensitive to initial clustering center and data distribution has influence on BH algorithm Because of original BH algorithm exists repeating computation, this paper proposes Fast Black Hole (FBH) algorithm and applies new algorithm to SC algorithm and proposes Fast Black Hole Spectral Clustering (FBH-SC) algorithm and computation time of this algorithm reduces significantly compared to BH-SC algorithm. This articled applies BH-SC algorithm and FBH-SC algorithm to two feature vector datasets of dynamometer cards and average clustering accuracies of clustering are both above 80% and this meets requirements of Oil field.It is need to assign clustering number and scale parameter before using BH-SC algorithm and FBH-SC algorithm To solving this problem, this article introduces a new indicator function and accomplishes clustering process when this function achieves optimal value. This function achieves optimal value means the clustering result is best, the clustering number and scale parameter is optimal. FBH algorithm which is proposed in chapter 4 is used to search procedure. This algorithm of automatically determining the number of clusters is applied to two feature vector datasets of dynamometer cards and the optimal number of clusters are equal to realistic number of diagnosis and mean accuracy of clustering is on the 80% and this satisfies the actual need. This paper adds new sample for diagnosis into Curve Feature Vector dataset and uses algorithm of automatically determining the number of clusters to cluster new dataset and determines condition types of new sample for diagnosis.