Predicción de la tendencia con índices porcentuales de los precios de bolsa horarios del mercado eléctrico usando clasificadores con parámetros adaptativos y varias fuentes de información
Predicción de la tendencia con índices porcentuales de los precios de bolsa horarios del mercado eléctrico usando clasificadores con parámetros adaptativos y varias fuentes de información
Autores
Director
Holguín Londoño, Mauricio
Autor corporativo
Recolector de datos
Otros/Desconocido
Director audiovisual
Editor/Compilador
Editores
Universidad Tecnológica de Pereira
Tipo de Material
Fecha
2021
Cita bibliográfica
Título de serie/ reporte/ volumen/ colección
Es Parte de
Resumen
La correcta predicción del Precio Horario en la Bolsa de Energía (PHBE) puede facilitar el mercado eléctrico a los Generadores Distribuidos y Autogeneradores a Pequeña Escala para que se integren en los diferentes sistemas de distribución energéticos. Dicha integración acarrea importantes beneficios para los usuarios generadores, como la remuneración económica por la energía excedentaria que inyecten en la red, y otros
beneficios para el sistema eléctrico y el medio ambiente. Por consiguiente, se presenta una metodología que permite predecir, a corto plazo, la tendencia con índices porcentuales del PHBE en el mercado eléctrico colombiano, haciendo uso de Redes Neuronales Artificiales (ANNs) y Modelos Ocultos de Markov (HMMs), con parámetros adaptativos al mercado y con base en dos fuentes de información. Con la implementación propuesta se confirma que los clasificadores ANNs son una herramienta potente y flexible para pronosticar la tendencia del PHBE, también se evidencia que hay muchas formas de utilizar los HMMs para mejorar el desempeño en la clasificación. Además, se muestra que el uso de índices porcentuales en la predicción
de la tendencia del PHBE es posible y es un indicador que aporta mayor relevancia en la predicción, en contraste con la sola tendencia de subida o de bajada del precio. De esta forma, los analistas y otros agentes del mercado pueden tener una idea más acertada para la toma de decisiones.
The correct prediction of the Hourly Price on the Energy Stock (PHBE) can facilitate the electricity market for Distributed Generators and Small-Scale Self-Generators to be integrated into the different energy distribution systems. This integration brings important benefits for the generating users, such as the economic remuneration for the surplus energy that they inject into the grid, and other benefits for the electrical system and the nature. Therefore, a methodology is presented that allows predicting, in the short term, the trend with percentage indices of the PHBE in the Colombian electricity market, making use of Artificial Neural Networks (ANNs) and Hidden Markov Models (HMMs), with adaptive parameters to the market and based on two sources of information. With the proposed implementation it is confirmed that the ANN classifiers are a powerful and flexible tool to forecast the trend of the PHBE, it is also evident that there are many ways to use the HMMs to improve the performance in the classification. In addition, it is shown that the use of percentage indices in the prediction of the trend of the PHBE is possible and is an indicator that provides greater relevance in the prediction, in contrast to the only trend of rising or falling of the price. In this way, analysts and other market agents can have a better idea for decision-making.
The correct prediction of the Hourly Price on the Energy Stock (PHBE) can facilitate the electricity market for Distributed Generators and Small-Scale Self-Generators to be integrated into the different energy distribution systems. This integration brings important benefits for the generating users, such as the economic remuneration for the surplus energy that they inject into the grid, and other benefits for the electrical system and the nature. Therefore, a methodology is presented that allows predicting, in the short term, the trend with percentage indices of the PHBE in the Colombian electricity market, making use of Artificial Neural Networks (ANNs) and Hidden Markov Models (HMMs), with adaptive parameters to the market and based on two sources of information. With the proposed implementation it is confirmed that the ANN classifiers are a powerful and flexible tool to forecast the trend of the PHBE, it is also evident that there are many ways to use the HMMs to improve the performance in the classification. In addition, it is shown that the use of percentage indices in the prediction of the trend of the PHBE is possible and is an indicator that provides greater relevance in the prediction, in contrast to the only trend of rising or falling of the price. In this way, analysts and other market agents can have a better idea for decision-making.