Files
Abstract
Electricity Price Forecasts have been a very important part of the energy industry for a long time. Traditionally, it was used by utilities, but after the liberalization of electricity markets around the world, they are of prime importance not only for utilities, but also for portfolio managers, aggregators, retailers and generation companies. These different players can plan the dispatch and consumption schedule of power in accordance with price forecasts in order to maximize profit. Apart from this, electricity prices, in general, can exhibit extreme volatility. This volatility can be attributed to the change in electricity demand and the amount of energy generated/used from renewable sources (since the price of power generated from renewable sources can be very low). This calls for better modeling of electricity prices. The aim of this study is to explore various machine learning techniques like linear regression, gradient boosting, random forests, support vector machines and artificial neural networks for forecasting electricity prices of the German market. The main focus of this research is to explore the use of meta heuristic algorithm like particle swarm optimization to search for the best suited feature space for each algorithm.