Distributed Pareto Reinforcement Learning for Multi-objective Smart Generation Control of Multi-area Interconnected Power Systems

Yin, Linfei; Cao, Xinghui; Sun, Zhixiang

doi:10.1007/s42835-022-01099-y

Cited by 2 publications

(2 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed approach demonstrates superior performance compared to existing methods in terms of settling time, frequency, and tie-line power deviations for different usage cases. In [12], the research addresses the multi-objective control problem (MOCP) in smart generation control (SGC) of multi-area interconnected power systems (MAIPSs). The proposed distributed Pareto reinforcement learning (DPRL) approach combines reinforcement learning with multiple Q matrices, enabling dynamic control strategies online while optimizing multiple objectives.…”

Section: Literature Reviewmentioning

confidence: 99%

“…Work in [2], on the other hand, employs Distributed Coordination AGC with Adaptive Chaotic Gray Wolf Algorithm, which is rated "High" in complexity due to the adoption of a complex optimization algorithm. Other models that lean towards higher complexity include those utilizing reinforcement learning (Work in [12]), chaos game optimization (Work in [11]), and fuzzy logic (Work in [3]). Delay: Delay in control systems refers to the time it takes for the controller to process input data and produce an output.…”

Section: Empirical Analysismentioning

confidence: 99%

See 1 more Smart Citation

Comparative Analysis of Various Controllers used for Automatic Generation Control in Electrical Power Systems

2023

IJFMR

View full text Add to dashboard Cite

In the rapidly evolving landscape of power systems, achieving a stable and efficient Automatic Generation Control (AGC) remains paramount. The exigency for this research stems from the critical role AGC plays in maintaining frequency and power balance amidst dynamic system changes. The consequences of inadequacies in AGC manifest as frequency deviations, system instabilities, and potential blackouts, with widespread socio-economic implications for different use cases. Historically, various controllers have been proposed and implemented to address AGC challenges. However, existing methodologies have displayed inherent limitations. Predominant issues include lack of adaptability to non-linearities, slow response times, inability to cater to the multifaceted nature of modern power grids with renewable integrations, and susceptibility to uncertainties and disturbances. This comprehensive review delves deep into the spectrum of controllers available for AGC, encompassing conventional PI/PID controllers, robust controllers, adaptive controllers, intelligent controllers, and the more recent forays into neural network-based and fuzzy logic controllers. Each controller's salient features, operational dynamics, advantages, and shortcomings are meticulously dissected. Furthermore, this paper elucidates the contexts in which each controller type excels or is found wanting, providing invaluable guidance for system designers and operators. In essence, this review serves as a holistic compass for stakeholders involved in power system operation and design. It offers both a retrospective understanding and prospective insights, assisting decision-makers in choosing the most suitable controller for a given scenario. Whether one aims to optimize AGC for a traditional grid, a grid with high renewable penetration, or a hybrid system, this paper promises to be an indispensable reference for researchers.

show abstract

Section: Literature Reviewmentioning

confidence: 99%