Hybrid Forecasting Model for Short-Term Electricity Market Prices with Renewable Integration

Abstract

In recent years, there have been notable commitments and obligations by the electricity sector for more sustainable generation and delivery processes to reduce the environmental footprint. However, there is still a long way to go to achieve necessary sustainability goals while ensuring standards of robustness and the quality of power grids. One of the main challenges hindering this progress are uncertainties and stochasticity associated with the electricity sector and especially renewable generation. In this paradigm shift, forecasting tools are indispensable, and their utilization can significantly improve system operation and minimize costs associated with all related activities. Thus, forecasting tools have an essential key role in all decision-making stages. In this work, a hybrid probabilistic forecasting model (HPFM) was developed for short-term electricity market prices (EMP) combining wavelet transforms (WT), hybrid particle swarm optimization (DEEPSO), adaptive neuro-fuzzy inference system (ANFIS), and Monte Carlo simulation (MCS). The proposed hybrid probabilistic forecasting model (HPFM) was tested and validated with real data from the Spanish and Pennsylvania-New Jersey-Maryland (PJM) markets. The proposed model exhibited favorable results and performance in comparison with previously published work considering electricity market prices (EMP) data, which is notable.

A Novel Electricity Price Forecasting Approach Based on Dimension Reduction Strategy and Rough Artificial Neural Networks

“…Moreover, statistical methods perform poorly dealing with price spikes in the electricity markets [4]. Recently, researchers have been interested in hybrid approaches, combining statistical methods with other forecasting methods in order to limit the shortcomings of these procedures [5]- [7]. The probabilistic methods are other kinds of approaches in EPF studies.…”

A Novel Ensemble Algorithm for Solar Power Forecasting Based on Kernel Density Estimation

“…This study is an example of numerous others applying probabilistic and ML techniques for electricity price forecasting which has been increasing exponentially in the past decade as shown by Reference [6]. The latter shows that, while non-existent before 2003, probabilistic methods (or hybrid ones) have quickly gained ground as one of the main approaches used contemporarily for price forecasting [19][20][21]. The analysis in Reference [22] has shown that, for the case of price forecasting, while combining different forecasts in an ensemble framework does not necessarily always bring about improved accuracy, it does contribute to more reliable forecasting by decreasing the risk associated with an individual method.…”

A comprehensive review on integrating renewable energy sources into smart grid : Challenges and Solutions

“…From three vital methods of forecasting (differentiated such as classical stastistical technique, computational intelligent method and hybrid algorithms) time series method-most ususal classical method of forecasting is explained along with their mathematical expressions in [39]. The hybrid probabilistic forecasting model (HPFM) using one of the modified versions of PSO (DEEPSO) is introduced in [42]. Reults were analysed with the help of monte-carlo simulation model with the desire of finding the range of forecasted values without growing the forecast error.…”