Ensemble Learning-based Electricity Price Forecasting for Smart Grid Deployment

Abstract

In a deregulated electricity market, forecasting electricity prices is essential to help stakeholders with the decision making process. Electricity price forecasting is an inherently difficult problem due to its special characteristics of dynamicity and nonstationarity. Further deployment of the smart grid for power distribution enables a two-way communication between the supplier and the consumer. This enables the supplier to price the energy based on the consumption feedback from the consumer, and the consumer can schedule their consumption behavior to achieve optimal utilization. In our research, our goal is to tackle dynamicity involved in price forecasting. We employ three different machine learning-based prediction algorithms, and propose two versions of ensemble based models (fixed weights and varying weights) which make predictions by taking into account the individual predictions from the three algorithms.The input features are selected from a pool of features derived from information such as past electricity price data, weather data, and calendar data. A wrapper method for feature selection is used in which the model is continuously trained and updated in order to select the best feature set. The performance of the proposed method is evaluated and compared with the published results of the state-of-the-art Pattern Sequence-based Forecasting (PSF) method on the same data sets and our method is observed to provide superior results.

Date of Award	Jun 2013
Original language	American English
Supervisor	U Zeyar Aung (Supervisor)