A neural network model for short-term load forecasting | |
| Author | Valenzona, Marion Loreto |
| Call Number | AIT Thesis no.CS-98-22 |
| Subject(s) | Neural networks (Computer science) Electric power-plants--Load |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Advanced Technologies |
| Publisher | Asian Institute of Technology |
| Abstract | For the past decade, numerous studies have been conducted on the application of neural networks to the short-term load forecasting task. Results obtained show acceptable accuracy during the forecast of normal days. However, it is also observed that there is a decrease in the accuracy during the "anomalous" load periods usually characterized by the occurrence of holidays and other national events. This study attempts to develop a neural network based on the combined (unsupervised/supervised) approach wherein the unsupervised stage serves as data preprocessing to identify clusters of similar load profiles. Such task is performed by means of the Kohonen Self Organizing Map (SOM). On the other hand, the supervised stage is assigned the proper forecasting task and is implemented using a multilayer perceptron with backpropagation learning algorithm. Simulations are conducted for the short-term load forecast of some selected anomalous load periods of Thailand. Results obtained indicate that the hybrid model provides a considerable improvement in the forecasting accuracy of these anomalous load periods. |
| Year | 1998 |
| Type | Thesis |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Sadananda, Ramakoti; |
| Examination Committee(s) | Yulu, Qi ;Dhadbanjan, Thukaram; |
| Scholarship Donor(s) | H.M. King of Thailand; |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology,1998 |