Short-term Load Forecasting

Shokuhian, Hamideh | 2014

519 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 45864 (44)
  4. University: Sharif University of Technology
  5. Department: Management and Economics
  6. Advisor(s): Fatemi Ardestani, Farshad; Barakchian, Mahdi
  7. Abstract:
  8. In this thesis we are going to forecast the hourly consumption of the electricity over the country with two models and then, combine them. The first model decomposes the consumption to a deterministic trend and a stochastic residual. The second one assumes that the trend part is also stochastic.Once the consumption is being predicted separately by the models, in the second part of the thesis, we will combine the results to get a final prediction. This prediction is going to be compared with the load forecast of the Dispatching Unit of the electricity network as a base model. We are going to answer two important questions: firstly, does combining the models give a better prediction or not, and secondly, how accurate a univariate model could be for load forecasting. The first question will be answered by comparing the combined model by the two original models. For the second question, we will compare the results of combined model with the Dispatching Unit’s prediction in which the weather parameters is exploited in a multivariate model. The results show that combining the models can enhance the accuracy in prediction. In addition, our univariate model predicts as good as the Dispatching Unit’s load forecast in some hours and even better in some (peak load hours)
  9. Keywords:
  10. Forecast Combination ; Time Series Decompositon ; Time Series ; Short Term Load Forecasting

 Digital Object List