PROJECT TITLE :

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

ABSTRACT:

In deregulated energy markets, accurate electricity price forecasting (EPF) is critical and has a direct impact on the power system's optimal management. Taking into account all of the important aspects that go into calculating energy pricing, some of which are stochastic, is a difficult undertaking. First, Grey correlation analysis is used in this article to identify the most effective parameters in the EPF problem and to eliminate redundant elements based on low correlation grades.

Then, to denoise data sets from various sources independently, a deep neural network with stacked denoising auto-encoders was used. After that, a dimension reduction procedure is used to identify the most important properties of the input data while ignoring the less important ones. Finally, the rough structure artificial neural network (ANN) was used to forecast the electricity price for the next day.

The proposed method is tested on data from Ontario, Canada, and the forecasted results are compared to different structures of ANN, support vector machine, and long short-term memory—benchmarking methods in this field—as well as forecasting data from the independent electricity system operator (IESO). In addition, the findings of this paper show that the proposed method is effective in terms of reducing error criterion and improves forecasting error by roughly 5-10% when compared to IESO. This is an outstanding achievement in the subject of EPF.