Content uploaded by Eric Tatara
Author content
All content in this area was uploaded by Eric Tatara on Jan 10, 2014
Content may be subject to copyright.
A preview of the PDF is not available
Agent-Based Control of Spatially Distributed Chemical Reactor Networks
Abstract and Figures
Large-scale spatially distributed systems provide a unique and difficult control challenge because of their nonlinearity,
spatialdistribution and generally high order. The control structure for these systems tend to be both discrete and distributed
as well and contain discrete and continuous elements. A layered control structure interfaced with complex arrays of sensors
and actuators provides a flexible supervision and control system that can deal with local and global challenges. An adaptive
agent-based control structure is presented whereby local control objectives may be changed in order to achieve the global
control objective. Information is shared through a global knowledge environment that promotes the distribution of ideas through
reinforcement. The performance of the agent-based control approach is illustrated in a case study where the interaction front
between two competing autocatalytic species is moved from one spatial configuration to another. The multi-agent control system
is able to effectively explore the parameter space of the network and intelligently manipulate the network flow rates such
that the desired spatial distribution of species is achieved.
Figures - uploaded by Eric Tatara
Author content
All figure content in this area was uploaded by Eric Tatara
Content may be subject to copyright.
... This framework was extended to control system performance assessment and modification (Schäfer & Cinar, 2004). The advent of agent-based systems enabled distributed AI implementation, leading to hierarchical agent-based supervision systems with distributed monitoring, diagnosis, and control functionality to supervise and regulate distributed processes for comprehensive multi-layered monitoring, diagnosis, and control unit, section and plant level operation (Perk, Shao, Teymour, & Cinar, 2012;Perk, Teymour, & Cinar, 2011;Tatara, North, Hood, Teymour, & Cinar, 2005). The data analytics, ML, and AI experience we gained in this journey provided the opportunity to apply these techniques to automated drug delivery, focusing on automated insulin delivery to people with diabetes with a multivariable artificial pancreas (mAP). ...
An adaptive-learning model predictive control (AL-MPC) framework is proposed for incorporating disturbance prediction, model uncertainty quantification, pattern learning, and recursive subspace identification for use in controlling complex dynamic systems with periodically recurring large random disturbances. The AL-MPC integrates online learning from historical data to predict the future evolution of the model output over a specified horizon and proactively mitigate significant disturbances. This goal is accomplished using dynamic regularized latent variable regression (DrLVR) approach to quantify disturbances from the past data and forecast their future progression time series. An enveloped path for the future behavior of the model output is extracted to further enhance the robustness of the closed-loop system. The controller set-point, penalty weights of the objective function, and constraints criteria can be modified in advance for the expected periods of the disturbance effects. The proposed AL-MPC is used to regulate glucose concentration in people with Type 1 diabetes by an automated insulin delivery system. Simulation results demonstrate the effectiveness of the proposed technique by improving the performance indices of the closed-loop system. The MPC algorithm integrated with DrLVR disturbance predictor has compared to MPC reinforced with dynamic principal component analysis linked with K-nearest neighbors and hyper-spherical clustering (k-means) technique. The simulation results illustrate that the AL-MPC can regulate the glucose concentrations of people with Type 1 diabetes to stay in the desired range (70–180) mg/dL 84.4% of the time without causing any hypoglycemia and hyperglycemia events.
... Although models and algorithms used to implement the control tasks may vary from plant to plant, most chemical plants today operate with a combination of these two hierarchies as shown in Figure 5. Decisions of decentralized subsystem controllers are coordinated to achieve the overall plant objectives. How to select the subsystems, design of regulatory and optimizing structures for each subsystem and developing feed forward/feedback information sharing strategies among the decentralized subsystem controllers have been studied in the past and they are still open research issues [Lesser, 1999; Tatara et al. 2006]. Figure 5. Hierarchical Control ...
Before embarking on the subject matter, a brief introduction to my background would put the title of this paper into perspective. I spent my academic career teaching and doing research in process control. During the last ten years I have served in university administration. While engaged in academic affairs, I came to realize that I benefited significantly from my systems and process control background. In retrospect I find systems thinking useful for conceptualizing and addressing a wide range of complex issues that come up in university management. Human factor, uncertainty and difficulty of modeling university dynamics make university administration a challenging task. Despite these difficulties, I believe that systems thinking in general and process control in particular can provide helpful guidance towards transforming academic administration into a learning and continuously improving system.
Purpose
This paper is to describe development of the features and functions of Repast Simphony, the widely used, free, and open source agent-based modeling environment that builds on the Repast 3 library. Repast Simphony was designed from the ground up with a focus on well-factored abstractions. The resulting code has a modular architecture that allows individual components such as networks, logging, and time scheduling to be replaced as needed. The Repast family of agent-based modeling software has collectively been under continuous development for more than 10 years.
Method
Includes reviewing other free and open-source modeling libraries and environments as well as describing the architecture of Repast Simphony. The architectural description includes a discussion of the Simphony application framework, the core module, ReLogo, data collection, the geographical information system, visualization, freeze drying, and third party application integration.
Results
Include a review of several Repast Simphony applications and brief tutorial on how to use Repast Simphony to model a simple complex adaptive system.
Conclusions
We discuss opportunities for future work, including plans to provide support for increasingly large-scale modeling efforts.
It is highly desirable to have a statistical process monitoring (SPM) system that detects the abnormalities in process operations quickly with as few missed and false alarms as possible while the process operates under various operating conditions. An agent-based combined monitoring and fault detection framework is proposed in this study. In this framework, different SPM techniques compete with and complement each other to enhance detection speed and accuracy. SPM techniques from literature such as principal component analysis (PCA), multiblock PCA (MBPCA), and dynamic PCA (DPCA) techniques are implemented in this agent-based process supervision system. An agent performance assessment and agent management layer provides dynamic adaptation of the supervision system and improves the performance of SPM. The statistical information coming from each of the statistical techniques is summarized through a consensus mechanism. The performance of the agent-based consensus mechanism using different consensus criteria is tested for system disturbances of various magnitudes. The effectiveness of the proposed agent-based framework with different consensus criteria is evaluated based on fault detection times and missed alarm rates and the adaptation of the supervision system is illustrated.
An adaptive agent-based hierarchical framework for fault type classification and diagnosis in continuous chemical processes is presented. Classification techniques such as Fisher’s discriminant analysis (FDA) and partial least-squares discriminant analysis (PLSDA) and diagnosis tools such as variable contribution plots are used by agents in this supervision system. After an abnormality is detected, the classification results reported by different diagnosis agents are summarized via a performance-based criterion, and a consensus diagnosis decision is formed. In the agent management layer of the proposed system, the performances of diagnosis agents are evaluated under different fault scenarios, and the collective performance of the supervision system is improved via performance-based consensus decision and adaptation. The effectiveness of the proposed adaptive agent-based framework for the classification of faults is illustrated using a simulated continuous stirred tank reactor (CSTR) network.
An adaptive hierarchical framework for process supervision and fault-tolerant control with agent-based systems is presented. The framework consists of modules for fault detection and diagnosis (FDD), system identification and distributed control, and a hierarchical structure for performance-based agent adaptation. Multivariate continuous process monitoring methodologies and several fault discrimination and classification techniques are implemented in the FDD modules to be used by multiple agents. In the process supervision layer, the continuous intramodular communication between FDD and control modules communicates the existence of an abnormality in the process, type of the abnormality, and affected process sections to the distributed model predictive control agents. In the agent management layer, the performances of all FDD and control agents are evaluated under specific process conditions. Performance-based consensus criteria are used to prioritize the best-performing agents in consensus decision making in every level of process supervision and fault-tolerant control. The collective performance of the supervision system is improved via performance-based consensus decision making and adaptation. The effectiveness of the proposed adaptive agent-based framework for fault-tolerant control is illustrated using a simulated continuous stirred-tank reactor network. Copyright © 2011 John Wiley & Sons, Ltd.
Complexity is a very diversified and branched subject and, ironically, is itself quite complex. In this paper, although we present the different aspects and definitions of complexity, we concentrate on its chemical/biological engineering relevance, especially for reaction/diffusion and hydrodynamic processes. System theory is used as the common language to unify concepts, and emphasis is given to bifurcation, chaos as the basis of behavioral complexity and the configuration of processes as the basis for structural complexity. Natural processes are grouped under biocomplexity, while man-made processes are treated as complexity alone. We restrict our attention in this paper to systems that do not change their structure during the process, so that self-organizational criticality is explained, but not utilized. Computational complexity is intrinsically inherent in all the processes we consider, but it is not given much attention in this paper. Despite these severe limitations on the scope of our paper, the subject is still quite complex and branched, and this paper tries to bring it to the attention and interest of a wider spectrum of chemical/biological engineers in both academia and industry.
Recent research discussed several approaches to understand the relation between microscopic agent behavior and macroscopic
multi–agent system (MAS) behavior. A structured methodology to derive these models will have impact on MAS design, evaluation
and debugging. Current results have established the description of macroscopic behavior, including cooperation, by Rate Equations
derived from markovian agent–states transitions. Emergent phenomena elude these descriptions. In this paper, we argue that
mesoscopic modeling is needed to provide appropriate descriptions of emergent system behavior. The mesoscopic agent states
reflect the emergent behavior and allow for a deliberative implementation of the rules and conditions which cause the MAS
to self–organize as wanted. In a case study, we construct such a mesoscopic model for the socio-economic inspired Minority
Game. The mesoscopic description leads us to a deliberative implementation, which exhibits equivalent self–organizing behavior,
confirming our results.
The University of Chicago's Social Science Research Computing's RePast is a software framework for creating agent-based simulations using the Java language. It provides a library of objects for creating, running, displaying, and collecting data from agent-based simulations. In addition, RePast includes several varieties of charts for visualizing data (e.g. histograms and sequence graphs) and can take snapshots of running simulations and create QuickTime movies of such. This paper describes some of the major changes and new features provided by the recently released RePast 2.0.
Piece-wise affine and mixed logical dynamical models for discrete time linear hybrid systems are reviewed. Constrained optimal control problems with linear and quadratic objective functions are defined. Some results on the structure and computation of the optimal control laws are presented. The effectiveness of the techniques is illustrated on a wide range of practical applications.
Systems with high steady-state multiplicity and rich dynamic behavior are difficult to investigate using conventional reductionist methods. A network of more than five reactors hosting cubic autocatalytic reactions may potentially have more than 102 steady states and many distinct dynamic regimes, all for the same parameter set. This paper discusses how the static complexity of such systems can be measured to give a holistic picture. To achieve this, stochastic simulations were performed to statistically determine the bifurcation structure of the system, and the gathered information is summarized using a measure akin to fractal dimension. With this measure, the growth of static complexity is investigated as a function of the network size.
The static and dynamic behavior of the autocatalytic reaction R + 2P → 3P with decay P → D is studied in networks of coupled continuous stirred tank reactors (CSTRs). Numerical bifurcation studies of the system are performed, resulting in rich steady-state bifurcation structures with multiple steady states and isolas. The heterogeneity of the networks is influenced by the number of reactors as well as the network topology. It is shown that the number of steady states of the network increases with heterogeneity, thereby allowing those autocatalytic species to exist in the network that would normally not exist in the homogeneous environment of a single CSTR. Spatial patterns of stable steady states are evident in reactor networks. Dynamic simulation studies are performed to illustrate the transition from one stable state configuration to another or from stable steady states to periodic regimes.
Two types of nonlinear feedback control schemes are introduced and analyzed for their capability of recovering the original state of an isothermal continuous-flow stirred tank reactor with one robust cubic autocatalytic species, perturbed by a temporary disturbance of an invading cubic autocatalytic species in the inflow. The control objectives are to eliminate the invading species from the system and to restore the original state of the host species. The extent of applicability of the control design to different nonrobust invading species is studied, when the controller is tuned for a specific invader. Moreover, a time-delay feature is suggested in one of the control schemes developed to achieve the control objectives in systems with poor detection of invading species.
The competition of two species for a common resource is illustrated using the paradigm of autocatalytic replicators inhabiting a continuous stirred tank reactor (CSTR) environment that is continuously fed with the resource. In most cases presented, one species is robust (appears in reactor feed) while the other is not. The introduction of the second (invading) species into the CSTR via an unsustained disturbance has a strong effect on the steady-state and dynamic behavior of the first (host) species. New steady states are added to the bifurcation diagram that show that the invading species can coexist in the system with the host species when its growth and death characteristics are similar to those of the latter. The population levels of the host species are greatly reduced in these cases as a result of the considerable decrease in resource concentration at steady state. Open-loop strategies for the elimination of the invading species are developed and discussed. These strategies involve the manipulation of the reactor residence time to destabilize the states of coexistence.
Agent-based computer systems are surprisingly effective at solving complex problems. Built by combining autonomous computer routines, or agents, with low-bandwidth communication capabilities, these systems typically perform significantly better than the individual routines operating alone. One source of this improvement lies in the cooperative collaboration among the individual agents that compose the system. This work proposes a modular framework for implementing agent-based systems for engineering design. Using a variety of different algorithmic agents, the key ideas are highlighted by identifying multiple identical global optima for a non-convex optimization problem with numerous local minima. The results show that inter- and intra-agent collaboration have a significant impact on system performance. Further, the system can be parallelized with little or no penalty. By observing and analyzing the interactions among individual agents, we gain insights that will aid in the development and management of a conceptual design system for truly hard and large problems.