| In recent years, fog and haze are now becoming common in China, which produces unfavorable influence on the production and life of the residents, and further influences the electricity load and its trend. The severe air condition presents a challenge to power system short-term load forecasting. Also, different type of power load suffers disparately from fog and haze. Therefore, a study of short-term load forecasting considering air condition and load classification is conducted aiming at aforementioned phenomena in this paper.The basic classification of power load and the corresponding characteristic are introduced firstly. Then the power system in Baoding city is taken as an example and the load data is concerned. Furthermore, the trait of different type of power load is analyzed while the correlation between various power loads and fog is explored. And based on that, two methods are taken to do load forecasting in different type of load. In the prediction of residential load and commercial load, the similar load days are extracted by using the gray correlation analysis method whilst the arithmetic of random forests is adopted in building forecasting model. The forecasting model is trained by using the sample set of similar days selected with air quality index(AQI) as input. In the prediction of industrial load which fluctuates in the smooth, the forecasting model based on wavelet neural network is built with the load data of a certain cement plant in Baoding city being chosen for preparation. Afterwards, the sample used for training and forecasting is set after data preprocessing, and then forecasting load is obtained by the usage of wavelet neural network with error analysis followed.Through calculation analysis, these two prediction methods are proved to possess good stability and precision. Therefore, the proposed research on short-term load forecasting considering air condition and load classification provides a new idea for load prediction in fog and haze. |
PDF Full Text Request