This paper addresses the impact of decomposition on the closedloop performance and computational efficiency of model predictive control (MPC) of nonlinear process networks. Distributed MPC structures with different communication strategies are designed for regulation of an integrated reactor−separator process. Different system decompositions are also considered, including decompositions into local controllers with minimum interactions obtained via community detection methods. The closed-loop performance and computational effort of the different MPC designs are analyzed. Through such a comprehensive comparison, tradeoffs between performance and computation effort, and the importance of systematic choice of the system decomposition, are documented.
A combined distributed moving horizon estimation and distributed model predictive control architecture is proposed to address the distributed output−feedback control problem for nonlinear process systems. Community detection based on modularity maximization is used to generate separate optimal decompositions for the estimation and control problems on the basis of suitable graphs. The process of benzene alkylation with ethylene is used as a case study to illustrate the application and computational advantages of the proposed control strategy.
The synthesis of a model-based control structure for general linear dissipative distributed parameter systems (DPSs) is explored in this manuscript. Discrete-time distributed state measurements (called process snapshots) are used by a continuous-time regulator to stabilize the process. The main objective of this article is to identify a criterion to minimize the communication bandwidth between sensors and controller (snapshots acquisition frequency) using linear systems analysis and still achieve closed-loop stability. This objective is addressed by adding a modeling layer to the regulator. Theoretically, DPSs can be well described by low dimensional ordinary differential equation models when represented in functional spaces; practically, the model accuracy hinges on finding basis functions for these spaces. Adaptive proper orthogonal decomposition is used to identify statistically important basis functions and establish locally accurate reduced order models which are then used in controller design. The proposed approach is successfully applied toward thermal regulation in a tubular chemical reactor.
This paper focuses
on the development of a rigorous model for isothermal
CO2 adsorption columns which describes the spatiotemporal
dynamics of CO2 concentrations in the bulk and solid bed
by a set of partial and location-varying ordinary differential equations.
By considering both dispersion and convection phenomena, the model
provides the spatiotemporal behavior of the adsorption rate and circumvents
the unphysical simplifying assumptions of linear driving force and
uniform adsorption rates through the column length invoked in previous
modeling efforts. The proposed model is then employed to compute physical
quantities originating from material conservation laws such as the
adsorption rate constant and CO2 adsorption capacity from
a set of experimental data without using empirical parameter assumptions
invoked in previous research. The spatiotemporal dynamics of CO2 adsorption in an aminosilica packed bed are successfully
predicted by the proposed model. The adsorption rate constant and
capacity of the bed are then identified using a set of experimental
CO2 concentration measurements at the adsorption column
outlet by solving a dynamic optimization problem using a shooting
method formulation. Finally, the adsorption enthalpy is computed by
employing the heat of adsorption data to validate the estimated parameters
of the system.
The output feedback control problem for a class of nonlinear distributed parameter systems with limited number of continuous measurement sensors that describes a wide range of physico-chemical systems is investigated using adaptive proper orthogonal decomposition (APOD) method. Specifically, APOD is used to initiate and recursively revise locally valid reduced order models (ROMs) that approximate the dominant dynamic behavior of such physico-chemical systems. The controller is designed based on ROMs by combining a robust state controller with an APOD-based nonlinear Luenberger-type switching dynamic observer of the system states to reduce measurement sensors requirements. The important static observer requirements on the number of measurement sensors (that must be supernumerary to the ROM dimension) and their location is circumvented by synthesizing dynamic observers. Three different approaches are introduced to recursively compute the dynamic observer gains at the ROM revisions. The stability of the closed-loop system is proven via Lyapunov and hybrid system stability arguments without invoking the separation principle between control and observation. The proposed method is successfully used to regulate a physico-chemical system that can be described in the form of the Kuramoto-Sivashinsky equation when the process exhibits significant nonlinear behavior.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.