Abstract:
Power system expansion planning begins with a forecast of future load requirements.
Estimation of both demand and energy requirements are crucial to effective system
planning. Load forecasting is also important for operational decision making by utilities
and for contract evaluations and evaluations of various sophisticated financial products
on energy pricing offered by the market.
Load forecasting is concerned with the prediction of hourly, daily, weekly and annual
values of the system demand and peak demand. Long-term load forecasting is an integral
process in scheduling the construction of new generation facilities and in the
development of transmission and distribution systems.
Empirical mode decomposition (EMD) technique has been applied for short-term load
forecasting, identification of weather sensitive component of electrical load and multi
scale analysis of daily peak load. A long-term load forecasting method using EMD is
developed in this thesis. The proposed methodology is entirely based on the historical
load data. No econometric factor like GDP, change in literacy rate, industrialization, new
electricity connectivity etc. are considered in the forecasting process due to the lack of
reliable data.
In this work, EMD technique is used on the historical load data of Bangladesh Power
System (BPS) to decompose it into intrinsic oscillatory components, called intrinsic mode
functions (IMFs) and a residue. Daily, weekly and monthly ratios of each IMF and the
residue are evaluated. Expected values of daily, weekly and monthly ratios of each IMF
and the residue are determined. A ratio factor for each day of each IMF and the residue is
calculated. The annual peak value of each IMF and the residue are forecasted using least square 2nd
order polynomial regression. These forecasted values are multiplied with corresponding
ratio factor of each day to forecast the daily peak values of each IMF and the residue.
Finally, the forecasting of electrical load is obtained by summing up the forecasted IMFs
and the residue.
The proposed methodology is validated through statistical error evaluation process and
comparison with other standard techniques. The developed methodology is applied to
forecast the daily peak load of BPS for a 10 year period, from year 2013 to year 2022.
