The objective of this research is to develop a new class of intelligent adaptive control architecture with scalable algorithms for complex systems control and optimization. The approach is to provide a unified control framework using an introduced system supervision concept that coherently incorporates adaptive controller and intelligent learning schemes with special emphasis on control of power system dynamics.
Intellectual Merit:
The intellectual merit is in extensive design, development and evaluation of a next generation intelligent control paradigm. Modern power systems are becoming increasingly stressed, due to growing demand and deregulation. An important solution for the control of modern power network is to provide intelligent hybrid frameworks that are adaptable to dynamic environment (system-centric). The goal of this CAREER program is to develop such a novel architecture based on the concept of supervisory loops that monitor system dynamics. The result is a resilient control framework that can adapt seamlessly to both designed and emergent conditions.
Broader Impact:
The broader impact includes a potential breakthrough in emerging and largely untapped system-centric controllers for seamless integration of renewable and non-renewable energy sources; economic and viable continuous power dispatch; sustainable and transportable next generation energy systems; and security against power disruptions caused by inadvertent events or malicious intent. As these approaches are important in many other areas of scientific applications, they will therefore have global value. This research also present unique opportunities to student researchers, especially from under-represented and minority groups and foster inter- and cross- disciplinary research collaborations with academic and scientific laboratories.
. The approach is to provide a unified control framework using an introduced system supervision concept that coherently incorporates controller and intelligent learning schemes with special emphasis on control of power system dynamics. Such system-centric controllers can optimize and control the power grid and have a reliable, resilient and robust sytem in the presence of higher penetration of renewables and at the same time optimize the local controllers (such as PSS) to perform better in during transient stability conditions. Outcomes: First a new concept of system-centric controller has been formulated. These algorithms are then designed and tested for resiliency, scalability, accuracy and reliability. The proposed highly sophisticated intelligent optimal controller in the form of action-critic neural network integrated to linear controllers performs during the operation of the power grid and maintain the grid stability. These designs are evaluated on a one machine test system and multiple machine two area power system models, augmenting conventional PSS and adaptive PSS. It has been observed that the proposed architecture can be used as a local controller or wide area controller. The control architecture can evolve from a simple linear controller to sophistitacted optimal controller based on system configurations. Another important feature is that, the online control tuning can be adjusted based on stochastic or deterministic problems and thus we can choose this as a value priority scheme. The architecture has provided a successful integrated platform and breakthrough in modern power system control and stability. The architecture is further designed using an optimal time shifting capable Dual Hueristic Programming (DHP) action critic neural network. The DHP based neural network aguments a conventional linear adaptive controller with value priorities on different local control objectives.This architecture is further tested on a multi-machine power system for inter-area mode oscillations. Then a wide area network based on ths architecture has been designed and developed. This architecture provides global objective function evaluations and interacts with local controller based on this method. The results indicates that the introduction of DHP substantially improves transient stability margin and damps inter-area model oscillations faster which has a direct corelation to power transfer capability. We have completed designing and deploying the control algorithms using a FPGA based hardware. The architecture can work with existing sensors and relays such as PMU's and other local controllers. The real-time test shows the proof of the proposed archtiecture. The proposed design can be implemented in the power grid increasing the controllability of the grid and improving the grid level resiliency, robustness, reliabiltiy and stability during current and emergent conditions of the grid.
