Optimal Controlled Variable Selection for Cyber-Physical Systems

Abstract

The design of optimal control architectures and the optimal selection of controlled variables represents a critical task for maximizing the economic profitability of operating plants. Traditionally, controlled variables are selected based on heuristics or past experiences. For cyber-physical systems (CPS), where physical and virtual components can be integrated in the same unit to evaluate the dynamic response of innovative power cycles past experiences may not be available. Trial-and-error methods can be time-consuming and costly. Besides, the selected controlled variables may not be optimal from the economic perspective. A systematic approach to the selection of controlled variable for cyber physical systems is presented here. The method is applied to a solid oxide fuel cell (SOFC)-gas turbine (GT) CPS at the National Energy Technology Laboratory (NETL) in Morgantown, WV. The main result of this work is the synthesis of optimal controlled variable sets for a cyber physical system exhibiting trade offs between economics as well as controllolability of variables.

Multiobjective Optimal Controlled Variable Selection for a Gas Turbine–Solid Oxide Fuel Cell System Using a Multiagent Optimization Platform

“…The specific GT-SOFC system that is investigated here is the HyPer (hybrid performance) facility at the U.S. Department of Energy’s National Energy Technology Laboratory (NETL) in Morgantown, West Virginia, USA . Despite several studies on the dynamics of the HyPer facility and its interaction with the hardware, ,,− the CVs for this process are currently selected heuristically. The HyPer facility is a cyber-physical system , where a software model is used for representing the SOFC.…”

Biomimetic and Advanced Control Structure Design with Real Time Optimization

“…Furthermore, the impact of the decomposition was examined on various cases when the process shifts from nominal conditions and the losses are found to increase as the process deviates from nominal conditions, however no clear correlation was found from the decomposition and islanding on the increment of the loss. (Bankole et al, 2018b, Bankole et al, 2018c.…”

“…Again as in section 3.6.3.2, the optimization as defined in Eq. (3.45) was implemented on an Intel ® Xeon ® CPU E-5-1620 v2 with 32GB RAM using the heterogeneous multiagent framework programmed in MATLAB (Bankole et al, 2018b). A similar version with worst case loss defined in Eq.…”

“…A similar version with worst case loss defined in Eq. 3.26 was implemented in (Bankole et al, 2018c). The multiagent framework with parameters as described in Table 3.4.…”