Sucker rod pumping diagnosis using valve working position and parameter optimal continuous hidden Markov model. (November 2017)

• Record Type: Journal Article
• Authors: Zheng, Boyuan; Gao, Xianwen
• Is Part Of: Journal of process control. Volume 59(2017)
• Page Start: 1
• Page End: 12
• Publication Date: 2017-11
• Subjects: Sucker rod pumping diagnosis -- Dynamometer card -- Valve working position -- Barycentric decomposition -- Continuous hidden Markov model; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2017.09.007
• Languages: English
• ISSNs: 0959-1524

Abstract

Highlights: The paper analyzes the mechanism of dynamometer card. A novel method, which is based on the curvature and the barycentric decomposition, is adopted to locate valve working positions of dynamometer cards at different working conditions. Seven novel geometric features are extracted from dynamometer card according to qualitative and quantitative analysis. The continuous hidden Markov model is first time applied in the field of sucker rod pumping diagnosis and the parameters of this model are optimized by the clonal selection algorithm to improve its classified performance. Actual production data is adopted in simulation. Abstract: Down-hole operating condition diagnosis based on dynamometer card is a key subject for sucker rod pumping in oil extraction engineering. In this technology, feature extraction and diagnostic model are two indispensable elements. To accurately and automatically diagnose the operating condition by computer, a novel diagnostic method for sucker rod pumping is proposed. The first novel idea is to extract seven geometric features, which are obtained from dynamometer card using barycentric decomposition algorithm and valve working position. The second novel idea focuses on the use of continuous hidden Markov model (CHMM) to create classifiers for diagnosing the down-dole operating conditions and then clonal selection algorithm (CSA) is used to optimize the selection of initial parameters for CHMM. Finally, the proposed method is tested on an oil field dynamometer card set. Furthermore, this technique is compared with some other existing approaches. The simulation results demonstrate that the performance using the method proposed in this paper is satisfactory.