Inteligencia artificial para sistemas adaptables
Identifying the moment when maintenance and improvement processes should be executed in any infrastructure system is a very complex task, since these systems are characterized by evolving over time. Similarly, the execution of these activities depends on factors that are unknown when designing the m...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Estado: | Versión publicada |
| Fecha de publicación: | 2021 |
| País: | Colombia |
| Institución: | Universidad de los Andes |
| Repositorio: | Séneca: repositorio Uniandes |
| Idioma: | español |
| OAI Identifier: | oai:repositorio.uniandes.edu.co:1992/50975 |
| Acceso en línea: | http://hdl.handle.net/1992/50975 |
| Access Level: | acceso abierto |
| Palabra clave: | Inteligencia artificial Aprendizaje por refuerzo (Aprendizaje automático) Carreteras Ingeniería |
| Sumario: | Identifying the moment when maintenance and improvement processes should be executed in any infrastructure system is a very complex task, since these systems are characterized by evolving over time. Similarly, the execution of these activities depends on factors that are unknown when designing the maintenance and improvement programs, and are very difficult to predict. The objective of this project is to create an agent using artificial intelligence algorithms, especially machine learning and reinforcement learning algorithms. The algorithm builds a policy that guarantees that the changes are made to the system only when required and guarantees that the size of the updates is adequate. This ensures that the system works in optimal conditions, always seeking to maximize the profits obtained. For this, a model that simulates the behavior of the real environment and a specific agent was created, which, through interaction with the environment, learns about the changes that occurred in the system over time and identifies which are the actions that generate the most benefit. |
|---|