Swarm Reinforcement Learning using DEEPSO-Q with Advantage for Operational Planning of Energy Plants

Autor:	Kenjiro Takahashi, Yoshikazu Fukuyama
Rok vydání:	2020
Předmět:	Electric power system Mathematical optimization Computer science Operational planning Swarm behaviour Particle swarm optimization Reinforcement learning Differential (mechanical device) Energy (signal processing)
Zdroj:	ICAIIC
DOI:	10.1109/icaiic48513.2020.9065286
Popis:	This paper proposes a new swarm reinforcement learning technique and its application to operational planning of energy plants (OPEPs) in buildings and factories. The proposed technique is differential evolutionary particle swarm optimization based q-learning with advantage (DEEPSO-Q w/A). The proposed DEEPSO-Q w/A based method can reduce engineering man-hour to develop the planning system and operational costs of the system. The results of the proposed method are compared with those of the conventional PSO-Q and DEEPSO-Q based methods. It is verified that the proposed method can reduce energy costs the most among the three applied methods.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::2bf43efd4c150a0da4bcd96ba83dca8b https://doi.org/10.1109/icaiic48513.2020.9065286 Zobrazit plný text záznamu