In recent decades, the traditional monopolistic energy exchange market has been replaced by deregulated, competitive marketplaces in which electricity may be purchased and sold at market prices like any other commodity. As a result, the deregulation of the electricity industry has produced a demand for wholesale organized marketplaces. Price predictions, which are primarily meant to establish the market clearing price, have become a significant factor to an energy company’s decision making and strategic development. Recently, the fast development of deep learning algorithms, as well as the deployment of front-end metaheuristic optimization approaches, have resulted in the efficient development of enhanced prediction models that are used for electricity price forecasting. In this paper, the development of six highly accurate, robust and optimized data-driven forecasting models in conjunction with an optimized Variational Mode Decomposition method and the K-Means clustering algorithm for short-term electricity price forecasting is proposed. In this work, we also establish an Inverted and Discrete Particle Swarm Optimization approach that is implemented for the optimization of the Variational Mode Decomposition method. The prediction of the day-ahead electricity prices is based on historical weather and price data of the deregulated Greek electricity market. The resulting forecasting outcomes are thoroughly compared in order to address which of the two proposed divide-and-conquer preprocessing approaches results in more accuracy concerning the issue of short-term electricity price forecasting. Finally, the proposed technique that produces the smallest error in the electricity price forecasting is based on Variational Mode Decomposition, which is optimized through the proposed variation of Particle Swarm Optimization, with a mean absolute percentage error value of 6.15%.

This paper introduces an alternative forecasting approach that leverages the application of visibility graphs in the context of multivariate energy forecasting for a regional airport, which incorporates energy demand of diverse types of buildings and wind power generation. The motivation for this research stems from the urgent need to enhance the accuracy and reliability of load forecasting in microgrids, which is crucial for optimizing energy management, integrating renewable sources, and reducing operational costs, thereby contributing to more sustainable and efficient energy systems. The proposed methodology employs visibility graph transformations, the superposed random walk method, and temporal decay adjustments, where more recent observations are weighted more significantly to predict the next time step in the data set. The results indicate that the proposed method exhibits satisfactory performance relative to comparison models such as Exponential smoothing, ARIMA, Light Gradient Boosting Machine and CNN-LSTM. The proposed method shows improved performance in forecasting energy consumption for both stationary and highly variable time series, with SMAPE and NMRSE values typically in the range of 4–10% and 5–20%, respectively, and an $R^2$ reaching 0.96. The proposed method affords notable benefits to the forecasting of energy demand, offering a versatile tool for various kinds of structures and types of energy production in a microgrid. This study lays the groundwork for further research and real-world applications within this field by enhancing both the theoretical and practical aspects of time series forecasting, including load forecasting.

The electricity wholesale market is inherently volatile in a deregulated market structure where market participants like power generators and retailors drive the price of electricity. Timely forecasting of the wholesale market prices by market participants has become of utmost importance in order to maximize on profits and minimize on risks. This report presents a hybrid method comprised of an extreme learning machine and a fuzzy inference engine to forecast price intervals using historical wholesale price extreme values (price maximum and minimum), historical load, generation and congestion hours, forecasted temperature and power outage data. This hybrid forecasting method has been tested on RTO Pennsylvania-New Jersey-Maryland (PJM) interconnection for the period July 1 ^st , 2018 to February 8 ^th , 2019, and is compared with individual extreme learning machine and the non-linear autoregressive neural network.