An application of neural networks on campus electricity consumption forecast
| Autor: | Jui-Yi Hsia, 夏瑞毅 |
|---|---|
| Rok vydání: | 2018 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 106 Exhausted by energy all over the world, how to effectively save electricity has been planned by all countries in the world. The purpose of this thesis is to apply neural networks to predict changes in campus electricity consumption. This method will assist the management and decision-making of the power supply of the campus general affairs unit and avoid the maximum demand for electricity exceeding the capacity of the regular contract. This study uses the characteristics of neural networks with strong adaptability and can construct non-linear models. It actually collects six different electricity meters on the campus and takes five days of data. Each meter''s one-day data is ten minutes, for 30 data sets. The 24 trials were performed to evaluate the effectiveness of the method and compared with common prediction methods including gray prediction, exponential smoothing, moving average, and linear regression analysis. Finally, several methods are used to test the effectiveness of various forecasting methods in the prediction of campus electricity consumption. From the experimental results, it can be known that the average error of the neural network prediction results is smaller than those of other prediction methods. It show not only good prediction accuracy in various performance evaluations, but also the neural network has nonlinear output characteristics and parallel processing capabilities. The campus where the power consumption changes drastically and is unstable is quite suitable |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|