Electricity load forecasting for power system planning and operation by using artificial neural network
The electrical power system performance is obtained when generation keeps pace with demand. The generation, transmission, and distribution companies need a method for forecasting electrical load to maximize the security, efficiency, and economic utilization of their electrical infrastructure. The ne...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/102772/1/YakubHusseinMohamedMSKE2022.pdf.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The electrical power system performance is obtained when generation keeps pace with demand. The generation, transmission, and distribution companies need a method for forecasting electrical load to maximize the security, efficiency, and economic utilization of their electrical infrastructure. The necessity of an appropriate model for forecasting the short-term electric power load provides the information required for the system's routine process management and unit commitment. The STLF has many notable merits, including reduction of operational costs, maintains the efficiency of energy markets, and provide a deeper understanding of the monitored system's dynamics. The STLF indicates the anticipated electric load for a time period ranging from a few hours to a few days. This is accomplished by inputting the day's form, hour, temperature, previous day loads, previous week loads and previous 24-hour loads into the proposed algorithm to forecast the short-term load. An Artificial Neural Network is created using the MATLAB2021a Simulation Software to validate the proposed Back Propagation algorithm's efficiency. The Artificial Neural Network is a mathematical method that simulates the human brain's thought processes. The Artificial Neural Network can be built and trained to take previous load data and the weather information as inputs and gives an output of the forecasted load. Thus, there are some missing data in the dataset that will surely affect the system’s accuracy. Therefore, the forecasting accuracy was evaluated by calculating different error metrics such as the MAPE, the MAE, the APE, the Daily Peak error, the MSE and the derived RMSE. For the proposed algorithm, the range of some hourly load forecasting error metrics lies between 1934378.64MW2 and 579905.75MW2 for the MSE, 2.97% and 6.17% for the Daily Peak error, and 3.15% and 6.54% for the MAPE. On the other hand, the overall MAPE of the proposed system is 5.9% error, while the compared multiple linear regression model gives an overall MAPE of 10.43% error. Therefore, the proposed model has least error than the MLRM. The proposed back propagation algorithm results demonstrate the method's superior performance and demonstrate that it can be used in realistic systems for forecasting short-term electricity load. |
|---|