(188c) Development of Biomimetic Approaches for Intelligent Control System Design, Monitoring and Optimization of Advanced Energy Systems

Traditionally, control systems have been designed based on assumed a priori knowledge of the process system. In these systems, a number of input variables are manipulated to accomplish disturbance rejection and/or servo control performance. While it is possible to adapt the process model based on available data, the control structure and the controllers rarely change. Thus, the knowledge learned during process operation is lost or remains underutilized. Furthermore, due to the desire for high efficiency under nominal as well as off-design conditions, ever-tightening environmental regulations, requirements for increased plant availability, and decreasing profit margins, control and optimization objectives of today’s operating plants are rapidly changing. Moreover, performances and characteristics of the key equipment items can significantly change over a period of time resulting in significant changes in the performance of the control system. Thus, it is desired that the plant control system should be agile and adapt quickly to the dynamic changes and requirements. To address these challenges, in this presentation, biomimetic approaches are introduced for control structure design, intelligent control and optimization. The proposed methods are applied to an Integrated Gasification Combined Cycle (IGCC) process with CO₂capture.

Biological systems differ from the traditional process control systems in distinct ways. For example, self-organization, distributed intelligence, adaptability, intelligent monitoring, cognition, and decision capabilities are some of the powerful characteristics of the biological world that can be effectively utilized in process control. At the top, the central nervous system (CNS) integrates the information from and coordinates the activities of all parts of the bodies (for bilaterian animals). Inspired by these distinct characteristics, a novel biomimetic approach to control system design has been developed. In particular, a suite of methodologies and algorithms has been developed to accomplish: (i) self-organization of the control structure for maximizing the plant’s operating profit by mimicking the function of the cortical areas in the human brain, (ii) design of distributed, agent-based and adaptive controllers that mimic the ant’s rule of pursuit combined with artificial neural network ideas, (iii) intelligent monitoring of the controllers powered with cognition and decision capabilities that mimic the artificial immune systems, and (iv) seamless coordination and integration in the control system that mimics the CNS.

The developed methodologies and algorithms are tested in a large-scale, plant-wide model of an Integrated Gasification Combined Cycle (IGCC) plant with CO₂ capture. The present work shows that the biomimetic approaches can offer superior control system performance in comparison to the existing control system design techniques.