Model-data-based switching adaptive control for dense medium separation in coal beneficiation. (May 2020)

• Record Type: Journal Article
• Authors: Dai, Wei; Zhang, Lingzhi; Fu, Jun; Chai, Tianyou; Ma, Xiaoping
• Is Part Of: Control engineering practice. Volume 98(2020)
• Publication Date: 2020-05
• Subjects: Dense medium separation -- Switching adaptive control -- Data-driven -- Unmodeled dynamics; Automatic control -- Periodicals; 629.89
• DOI: 10.1016/j.conengprac.2019.104241
• Languages: English
• ISSNs: 0967-0661

Abstract

Abstract: Dense medium separation (DMS) is an important process for the coal industry, but its time-varying and strongly nonlinear characteristics make it difficult to operate stably. To address this problem, this paper first adopts knowledge-based linear model and data-driven virtual unmodeled dynamics to formulate the DMS process. Then a model-data-based switching adaptive control approach is proposed to achieve the online adjustment of set point of dense medium density, which is a key operation parameter for the low-ash coal in production. To realize adaptive ability of the control system, the online parameter estimation methods for the linear model and virtual unmodeled dynamics are adopted, and their convergence are fully discussed, followed by the stability and convergence analysis of the closed-loop system. Finally, the effectiveness is demonstrated by using real data and semi-physical simulations. Highlights: A model-data-based switching adaptive control method is proposed for DMS process. A random vector function link network (RVFLN) is adopted to design data-driven virtual unmodeled dynamics compensator. The stability and convergence of proposed method are fully analyzed. The method feasibility is illustrated based on on-site data and semi-physical simulation.