01. Universidade Federal Rural de Pernambuco - UFRPE (Sede)
URI permanente desta comunidadehttps://arandu.ufrpe.br/handle/123456789/1
Navegar
2 resultados
Resultados da Pesquisa
Item Controle de temperatura de estufas utilizando método de controle supervisório(2020-10-29) Araújo, Guilherme Matheus de; Gonçalves, Glauco Estácio; http://lattes.cnpq.br/6157118581200722; http://lattes.cnpq.br/1992687962116742Greenhouses are structures used to set ideal environmental conditions for plant development. They are easy to build and maintain and vary in shape, structural components and other factors. They are made to adjust according to the needs of those who implent them, to a giving group of plants. One of the main factors that influence plant growth is the temperature inside the greenhouse. In order to reach the ideal temperature conditions for each group of plants automatically, a temperature control system, which models the actuators of a greenhouse environment and the events that occur in it, was developed. The validation of this control system was made possible thanks to the creation of a simulator developed according to the principles of mass and energy balance. The communication between the two entities takes place through a REST application interface. Three modeling proposals and rules for the control system are presented in this work. The main objective is to determine which one has the best feasibility and efficiency ratio in controlling the temperature inside the greenhouse, considering energy consumption, error rate in ºC and the percentage of time that the environment remains within ideal conditions. All proposals are based on the control of the states of two specific actuators in a greenhouse: the heating and cooling systems. The simulated scenarios use specific configurations for the greenhouse and a small tomato crop, in 24 hours periods for days chosen from different seasons. The control models showed different levels of efficiency, with up to 33% disparity.Item Um currículo de aprendizagem por reforço para recompensas modeladas no Lunar Lander(2021-07-19) Albuquerque, Renilson da Silva; Sampaio, Pablo Azevedo; http://lattes.cnpq.br/8865836949700771; http://lattes.cnpq.br/3364503614448061Reinforcement learning is a machine learning paradigm where the agent learns to solve problems interacting with an environment, executing actions in a trial and error sequence. For each action performed, the agent receives a reward from the environment indicating how effective it was in solving the whole problem. The agent’s objective is to maximize the total reward received. However, in some reinforcement learning problems, the agent needs to learn complex tasks receiving uninformative rewards, leading to the credit assignment problem that slows the agent’s training process. Reward shaping and curriculum learning are techniques that can speed up agent training time by separating the problem into smaller tasks to be solved sequentially, applying smaller and informative rewards for each action performed. Lunar Lander is a simplified 2D simulator used as a benchmark for reinforcement learning solutions to the optimization problem on landing control of a lunar module. However, its standard rewards system assigns much more punitive rewards for the use of the engines, not being very constructive for the agent, which can lead to the credit assignment problem. Hence, this work proposes a curriculum using two additional shaped reward models and runs experiments that aim to minimize the Lunar Lander learning time. This work found that both the new models and the curriculum were more effective in training the Lunar Lander agent compared to the standard rewards model.