Date of Award
6-1-2022
Thesis Type
masters
Document Type
Thesis (Restricted Access)
Divisions
eng
Department
Department of Electrical Engineering
Institution
Universiti Malaya
Abstract
Accurate electricity price forecasting (EPF) is important for the purpose of bidding strategies and minimizing the risk for market participants in the competitive electricity market. However, accurate prediction is very challenging due to complex nonlinearity in electricity prices. Therefore, forecasting accuracy highly depends on the nature of time series. An improved deep-learning framework is proposed for short and mid-term EPF which consists of four modules: time-series data pre-processing, the deep learning-based prediction methodology, spike prediction module and reliability checking of prediction model. The feature pre-processing module is based on linear trend of the correlated features of electricity price series and test time series for unit root by augmented dickey fuller (ADF). In addition, the time series data is transformed with box-cox transformation method for better training process. Firstly, the prediction module combines linear scaled hyperbolic tangent (LISHT) with the long short-term memory (LSTM) and compared with bidirectional long short-term memory (BiLSTM) which is a recurrent neural network (RNN) to adjust complex nonlinear features and improve the precision of day ahead prediction. The residual autocorrelation determined in the reliability check section. Secondly, an optimized gated recurrent unit (GRU) which incorporates bagged tree ensemble (BTE) is developed in the recurrent neural network (RNN) architecture for the mid-term EPF. A tanh layer is employed to optimize the hyperparameters of the heterogeneous GRU with the aim to improve the model's performance, error reduction and predict the spikes. This study is performed based on the Australian price, load and renewable energy supply data from five major economical states New South Wales (NSW), Queensland (QLD), South Australia (SA), Tasmania (TAS), Victoria (VIC). The experimental results obtained show that the proposed EPF framework performed better compared to previous techniques.
Note
Dissertation (M.A.) – Faculty of Engineering, Universiti Malaya, 2022.
Recommended Citation
Md Rashed, Iqbal, "Time series analysis and improved deep learning model for electricity price forecasting / Md Rashed Iqbal" (2022). Student Works (2020-2029). 1218.
https://knova.um.edu.my/student_works_2020s/1218