http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Greenhouse Library 시뮬레이션과 deep deterministic policy gradient 를 사용한 최적 온실 에너지 사용에 관한 연구
안주연 ( Juyeon Ahn ),박현지 ( Hyunji Park ),서현권 ( Hyun Kwon Suh ) 한국농업기계학회 2021 한국농업기계학회 학술발표논문집 Vol.26 No.1
Greenhouses Library is an open-source library for dynamic modeling of the indoor climate of a greenhouse developed in the Modelica language. The library was developed to provide an open modeling framework to simulate the climate inside the greenhouse and greenhouse energy interaction. The complex dynamics exist inside the greenhouse. The various environmental conditions also form a complex interaction between the elements. With the existing rule-based environmental control, it is challenging to achieve optimal indoor control inside a complex greenhouse. However, recently, artificial intelligence via reinforcement learning has shown excellent performance in a wide range of fields, especially in fields that require continuous control, such a building environmental control. In this study, artificial intelligence was applied to optimize the use of energy inside the greenhouse. DDPG (Deep Deterministic Policy Gradient) algorithm was implemented using MATLAB Reinforcement Learning toolbox. The agents were trained using Greenhouse Library, which was assumed to provide a viable and practical training approach. Despite a long training period over several repetitions, energy consumption in the greenhouse and other control factors was not improved but rather showed unrealistic control values. The finding of this study should be further investigated in future research.