| Abstract | 第4-5页 |
| 摘要 | 第6-9页 |
| 1 Introduction | 第9-16页 |
| 1.1 Modern construction issue forecasting systems | 第10-11页 |
| 1.1.1 Russian and other foreign systems of electric load forecasting | 第10-11页 |
| 1.2 Current methods of forecasting the electrical load | 第11-13页 |
| 1.3 Research object statement and construction of an advanced EPC forecasting system | 第13-14页 |
| 1.4 Summary | 第14-16页 |
| 2 The development of the actual technology of construction of forecasting system | 第16-32页 |
| 2.1 Dynamics of the time series of electricity consumption and its view model | 第16-20页 |
| 2.2 Overview of forecasting methods for different frequency components of time seriesand selection of the most appropriate one | 第20-28页 |
| 2.2.1 Analysis of the low-frequency component | 第25-26页 |
| 2.2.2 Analysis of the mid-frequency component | 第26-27页 |
| 2.2.3 Analysis of the high-frequency component | 第27-28页 |
| 2.3 Technology of forecasting system development | 第28-31页 |
| 2.4 Summary | 第31-32页 |
| 3 Development of a system of STLF | 第32-50页 |
| 3.1 Forecasting electrical load | 第32-34页 |
| 3.2 Algorithm of formation of NFN structure | 第34-38页 |
| 3.3 Search algorithm for the optimal depth of the wavelet decomposition using the methodof standardized range | 第38-41页 |
| 3.4 Search for the optimal architecture of NFN using genetic algorithm | 第41-45页 |
| 3.5 Algorithm of formation of training samples and input variables | 第45-48页 |
| 3.6 Summary | 第48-50页 |
| 4 The experimental operation of the electric load forecasting system | 第50-64页 |
| 4.1 Preparation for the experiment | 第50页 |
| 4.2 Evaluation of prediction system accuracy | 第50-55页 |
| 4.3 Performance analysis of the built system | 第55-57页 |
| 4.4 Comparison of the developed prediction model and other models | 第57-61页 |
| 4.5 Graphic User Interface development | 第61-64页 |
| Conclusions | 第64-65页 |
| Acknowledgement | 第65-66页 |
| References | 第66-69页 |
| Appendix | 第69-72页 |
| Research achievement during working for the degree | 第72页 |
