Simulación de un controlador basado en el algoritmo Q-learning en un sistema de lodos activados

Published in: Innovation in Engineering, Technology and Education for Competitiveness and Prosperity: Proceedings of the 11th Latin American and Caribbean Conference for Engineering and Technology
Date of Conference: August 14-16, 2013
Location of Conference: Cancun, Mexico
Authors: Carlos Andrés Peña Guzmán
Jaime Lara-Borrero
Refereed Paper: #115

Abstract:

In English:
This document presents a computational tool for control in a activated sludge reactor through simulations, based on the concept of reinforcement learning, where an agent runs an unknown environment and take action to achieve a specific objective. The control is execute to seraching flow for air injected into a reactor aerobic and the recycle sludge flow rate, to obtain a substrate of less than 100 mg/L and dissolved oxygen concentrations between 1 and 2 mg/L.


In Spanish:
Este documento presenta una herramienta computacional para el control en un tanque de aireación de los lodos activados mediante simulaciones, la herrameinta se basa en el concepto de aprendizaje por refuerzo, donde un agente recorre un ambiente desconocido y realizando acciones para alcanzar un objetivo especifico. El control se ejecuta buscando un caudal de aire a inyectar a un reactor aeróbico y un caudal de recirculación de lodos, para obtener un efluente con una concentración inferior a 100 mg/L de DQO y concentraciones de oxígeno disuelto entre 1 y 2 mg/L.