(190c) Optimal Control Structure Design for Cyber-Physical Systems

Although significant body of work exists in the open literature on controller design, research in the area of optimal control structure design which is primarily focused on optimal selection of controlled variables (CVs) is still nascent. Traditionally, control structure design has been done by leveraging process knowledge and heuristics. Recently, optimal selection of CVs by considering plant economics as well as controllability with due consideration of transport lag and closed loop interactions has been proposed for optimal control structure design (Jones et al. 2014a, 2014b). To the best of our knowledge, work on controlled variable selection for cyber-physical systems (CPS) is practically non-existing in the current open literature. CPS are flexible, where virtual components can be readily arranged in new configurations and/or new virtual component(s) can be readily added thus resulting in likely change in the optimal control structure. Due to high complexity and strong interaction between virtual components and the plant hardware in CPS, optimal control structure design becomes challenging. In this work, a novel approach with superior computational efficiency and performance is presented for a cyber-physical system (CPS).

The proposed method involves a three-stage procedure: a priori analysis, controlled variable selection, and a posteriori analysis. The controlled variable selection algorithm results in a large-scale, constrained, mixed-integer multi-objective optimization problem. For solving this problem, artificial intelligence via evolutionary algorithms is incorporated into the solution strategy. In particular, a metaheuristic heterogeneous multiagent optimization framework is deployed. This approach embodies novel features of an intelligent system by exhibiting coordination and information sharing amongst several agents making the controlled variable selection process amenable for online application.

The algorithm is applied to a hybrid gas turbine (GT)-fuel cell CPS, where a virtual fuel cell interacts with the gas turbine recuperated cycle. Fuel cell hybrid power systems exhibit immense potential for unparalleled electrical power generation efficiency (Tucker et al., 2005). The solid oxide fuel cell (SOFC)-GT hybrid power system has been developed under the Hybrid Performance (Hyper) project at the U.S. Department of Energy’s National Energy Technology Laboratory (NETL) in Morgantown, WV. The Hyper facility at NETL makes use of a numerical model for the SOFC that can produce 300 -700 kW electric power coupled with a 120 kW turbine. Thermo-hydraulic effects of the SOFC are captured by a series of pressure vessels and piping and by controlling the natural gas flowrate to a burner. Multivariable control of this system is notably challenging (Tsai et al., 2010). Large number of candidate controlled variables with strong interaction, multi-scale dynamic response, and process nonlinearities result in a challenging problem for controlled variable selection. Our study shows strong tradeoffs between economic performance and controllability of this system along with significant variabilities under off-design conditions. Posteriori analysis of the controlled variables yield a control structure that exhibits superior control and economic performance.

References

Jones D, Bhattacharyya D, Turton R, Zitney S, “Plant-Wide Control System Design: Secondary Controlled Variable Selection”, Computers & Chemical Engineering, 71, 253-262, 2014a

Jones D, Bhattacharyya D, Turton R, Zitney S, “Plant-Wide Control System Design: Primary Controlled Variable Selection”, Computers & Chemical Engineering, 71, 220-234, 2014b

Tsai A, Banta L, Tucker D, Gemmen, R, "Multivariable robust control of a simulated hybrid solid oxide fuel cell gas turbine plant", Journal of Fuel Cell Science and Technology, 7, 041008, 2010

Tucker D, Lawson L, Gemmen, R, "Characterization of air flow management and control in a fuel cell turbine hybrid power system using hardware simulation", In ASME 2005 Power Conference, American Society of Mechanical Engineers, 959-967, 2005.