Analysis of short term load forecasting techniques / Tan Vy Luoh

Author

Vy Luoh Tan

Date of Award

8-1-2019

Thesis Type

masters

Document Type

Thesis (Restricted Access)

Divisions

eng

Department

Faculty of Engineering

Institution

University of Malaya

Abstract

Nowadays, the implementation of advanced technology load and the introduction of multiple renewable energy sources to the grid have created major impacts to the electricity utilities provider with problems of power fluctuation, over generation and conventional power interruption. Therefore, short term load forecasting (STLF) is widely implemented as a necessary technique in power system planning and operation to ensure the power system is functioning in reliable and secure condition. In this report, three common numerical STLF techniques including Multiple Linear Regression (MLR), Curve Fitting and Bagged Tree Regression are proposed to forecast one-day ahead load profile with a yearly historical load data. The algorithms for each respective techniques are modelled in MATLAB Toolbox for simulation purpose. Forecasted curve of three techniques are obtained for evaluation with the diagnosis statistics including mean absolute percentage error (MAPE), mean absolute error (MAE), standard deviation absolute percentage error (StdAPE) and standard deviation absolute error (StdAE). The relative error between actual load and forecasted load is computed and used to compare the performance among three STLF techniques. As a result, bagged tree regression has lower relative error in MAPE and StdAPE which can be used to indicate it is more accurate STLF technique compare to the othertwo STLF techniques studied in this paper.

Note

Research Report (M.A.) - Faculty of Engineering, University of Malaya, 2019.

11169-tan.pdf (1092 kB)

This document is currently not available here.

Share

COinS