A Predictive Load-Aware and Multi-Scale Energy-Behavior Optimization Algorithm for Decentralized Multi-Agent Systems in Dynamic Power Networks
Kuppani SathishProfessor, Tirumala Engineering College, Narasaraopet, Pal Nadu, Andhra Pradesh, India. ksathish@tecnrt.org; skuppani@gmail.com0000-0001-9581-6819
Dr.M. VaralatchoumyProfessor & Head, Department of Artificial Intelligence and Machine Learning, Cambridge Institute of Technology, K.R. Puram, Bengaluru, India. kvl186@gmail.com0000-0003-3720-9644
Dr.K. VinuthaAssistant Professor, Department of ISE, BMS Institute of Technology and Management, Bengaluru, India. vinuthak_ise2014@bmsit.in0000-0002-2165-4738
B.N. ShwethaAssistant Professor, Department of Engineering and Technology, S Vyasa School of Advanced Studies, S-VYASA (Deemed to be University), Bengaluru, India. shwetha434@gmail.com0000-0001-8521-0864
Dr. Venkateswararao PulipatiAssociate Professor, Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Bowrampet, Hyderabad, Telangana, India. pvenkat2004@gmail.com0000-0002-2765-8318
Dr.Y. Jeevan Nagendra KumarProfessor, Head of the Department, Department of Information Technology, Gokaraju Rangaraju Institute of Engineering and Technology, Hyderabad, India. jeevannagendra@gmail.com0000-0003-2577-8065
Multi-agent systems that run on decentralized multi-agent stacks on dynamic power networks are known to experience enduring issues connected to energy efficiency, coordination, and adaptability in the face of time-varying loads and restricted communication. The fact is that the majority of current decentralized control approaches are mainly based on reactive decision-making and do not include the possibility of predicting future energy requirements, which contributes to inefficiencies and system unreliability. To overcome such problems, the current paper has suggested a predictive load-aware, multi-scale energy-behaviour optimization algorithm, named DECO-MARS, targeting the area of decentralized multi-agent power systems. DECO-MARS incorporates predictive load-conscious consensus control with a two-layered optimization structure that optimizes and coordinates the local energy constraints and global coordination goals simultaneously, and proactively and scalably optimizes the decentralized control. The IEEE 13-bus distribution test feeder is used to test the proposed algorithm based on realistic and time-varying load conditions and renewable generation conditions. The results of the simulations indicate that the total energy loss under the DECO-MARS is only 18.2 kWh as opposed to 25.1 kWh under a consensus-only control and 27.4 kWh under a local optimization, which is a significant enhancement in the energy efficiency. The framework has a voltage stability of 0.029 p.u. Voltage deviation index, which is much lower compared to the baseline techniques. DECO-MARS also has a 94.0 % success rate of tasks, a control latency of 1.2 seconds, and a normalized coordination score of 0.92, which is better than existing decentralized methods on all of the metrics assessed. The findings indicate that predictive intelligence and multi-scale optimization can be a significant improvement to the reliability, efficiency, and coordination of decentralized power networks. DECO-MARS can be used in distributed and ubiquitous energy systems such as smart grids, edge-controlled power networks, and autonomous energy-aware cyber-physical systems.