Date of Award


Document Type


Degree Name

Master of Science (MS)


College of Science and Mathematics


Computer Science

Thesis Sponsor/Dissertation Chair/Project Chair

Michelle Zhu

Committee Member

Weitian Wang

Committee Member

Jiaying Wang


The complexity and nonlinearities of the modern power grid render traditional physical modeling and mathematical computation unrealistic. AI and predictive machine learning techniques allow for accurate and efficient system modeling and analysis. Electricity consumption forecasting is highly valuable in energy management and sustainability research. Furthermore, accurate energy forecasting can be used to optimize energy allocation. This thesis introduces Deep Learning models including the Convolutional Neural Network (CNN), the Recurrent neural network (RNN), and Long Short-Term memory (LSTM). The Hourly Usage of Energy (HUE) dataset for buildings in British Columbia is used as an example for our investigation, as the dataset contains data from residential customers of BC Hydro, a provincial power utility company. Due to the temporal dependency in time-series observation data, data preprocessing is required before a model can be created. The LSTM model is utilized to create a predictive model for electricity consumption as output. Approximately 63% of the data is used for training, and the remaining 37% is used for testing. Various LSTM parameters are tested and tuned for best performance. Our LSTM predictive model can facilitate power companies’ resource management decisions.

File Format