Research And Application On Hybrid Models Based On Seasonal Exponential Adjustment Method

Wind energy, which is intermittent by nature, can have a significant impact on power grid security, power system operation, and market economics, especially in areas with a high level of wind power penetration. Wind speed forecasting has been a vital part of wind farm planning and the operational planning of power grids with the aim of reducing greenhouse gas emissions. Improving the accuracy of wind speed forecasting algorithms has significant technological and economic impacts on these activities, and significant research efforts have addressed this aim recently. However, there is no single best forecasting algorithm that can be applied to any wind farm due to the fact that wind speed patterns can be very different between wind farms and are usually influenced by many factors that are location-specific and difficult to control. In this paper, we propose a new hybrid wind speed forecasting method based on a back-propagation neural network and the idea of eliminating seasonal effects from actual wind speed datasets using seasonal exponential adjustment. A case study conducted using a wind speed dataset collected from the Minqin area in China demonstrates that this method can forecast the daily average wind speed one year ahead with lower mean absolute errors compared to figures obtained without adjustment. Besides, combined models based on the seasonal exponential adjustment method and regression approaches are introduced to forecast the load demand. Load forecasting results on VIC in Australia indicates that seasonal exponential adjustment method can really improve the forecasting accuracy when the seasonal item exists in the datasets.