(12f) Development of a Biologically-Inspired Approach for Advanced Adaptive Control of Clean Energy Systems

Over the last decades, biologically-inspired control techniques have emerged as attractive alternatives for process control applications. For example, a Biologically-Inspired Optimal Control Strategy (BIO-CS) has been recently proposed and illustrated via examples including a nonlinear fermentation process^1-3 and a hybrid energy system that integrates different process components⁴. Also, a low-dimensional system associated with the Acid Gas Removal (AGR) unit of an Integrated Gasification Combined Cycle (IGCC) plant was controlled using BIO-CS considering setpoint tracking cases⁵. Specifically, BIO-CS is an agent-based control strategy that is inspired by the ant’s rule of pursuit idea in combination with gradient-based optimal control concepts^1-5. In previous applications of BIO-CS, the multi-agent aspects of the algorithm were studied in detail and different termination criteria were analyzed to accommodate computational time limitations. However, the existence of plant-model mismatches as well as multiple disturbances in the control of high-dimensional systems has not yet been addressed using BIO-CS. To fill these gaps, in this presentation, BIO-CS is further extended to incorporate an adaptive component based on an Artificial Neural Network (ANN) method into the controller formulation. The proposed framework is then implemented for tackling high-dimensional control structures of the IGCC-AGR process simulation considering more realistic scenarios that could be encountered in practice.

For this study, a Multi-Input-Multi-Output (MIMO) system from the IGCC-AGR process in DYNSIM (software used for dynamic simulations of chemical processes) is chosen. In particular, CO₂/H₂S absorption units of the IGCC-AGR process are analyzed for the control structure selection. Simplified dynamic models are derived for BIO-CS employing system identification techniques, such as the autoregressive model with exogenous inputs (ARX). The optimal control trajectories are then computed by BIO-CS using the simplified model to maintain multiple outputs at their desired setpoints. The controlled/output variables considered in this implementation are the compositionsof the outgoing stream and the temperature of the incoming solvent related to the CO₂ absorption unit. The manipulated/input variables are the flowrates of the recycled solvent and the refrigerant associated with the same unit. Similar variables are selected for the H₂S absorption system, thus resulting in an overall high-dimensional system with multiple control islands. The proposed controller framework is designed in MATLAB and the control laws are communicated to the IGCC-AGR process simulation in DYNSIM by employing a MATLAB-DYNSIM link. This application represents the realistic scenario of the inherent mismatch between the plant and the model used by the controller. To mitigate this mismatch, an ANN-based adaptive component that has online learning capabilities is incorporated into the BIO-CS formulation. Using the information of the tracking errors, outputs, and available states, the adaptive BIO-CS brings the system back to the desired operating point. Preliminary results on the implementation of the proposed framework for the CO₂ capture island show promising capabilities in terms of maintaining the system at the required level of carbon capture. In addition, these results demonstrate the potential of the BIO-CS with adaptive component framework to tackle multiple challenges, such as nonlinearities, high dimensionality and plant-model mismatches that are commonly encountered in the process control industries.

References:

1. Lima F. V., Li S., Mirlekar G. V., Sridhar L. N. and Ruiz-Mercado G. J., “Modeling and advanced control for sustainable process systems”. Sustainability in the Analysis, Synthesis and Design of Chemical Engineering Processes, G. Ruiz-Mercado and H. Cabezas (eds.), Elsevier, 2016.
2. Li S., Mirlekar G. V., Ruiz-Mercado G. J. and Lima F. V., “Development of chemical process design and control for sustainability”. Processes, 4(3):23, 2016.
3. Mirlekar G. V., Li S. and Lima F. V., “Design and implementation of a Biologically-Inspired Optimal Control Strategy (BIO-CS) for chemical process control”. Submitted for publication.
4. Mirlekar G. V., Pezzini P., Bryden M., Tucker D. and Lima F. V., “A Biologically-Inspired Optimal Control Strategy (BIO-CS) for hybrid energy systems”. To appear in Proceedings of 2017 American Control Conference.
5. Mirlekar G. V. and Lima F. V., “Design and implementation of a Biologically-Inspired Optimal Control Strategy (BIO-CS) for advanced energy systems”. In AIChE Annual Meeting, San Francisco, CA, 2016.