Data Mining-Based Upscaling Approach for Regional Wind Power Forecasting: Regional Statistical Hybrid Wind Power Forecast Technique (RegionalSHWIP)

Abstract

With the increasing need for the energy, the importance of renewable energy sources has also been increasing. In order to include the power produced by the wind into electricity grid in a controlled manner, power prediction has an important role. To produce a reliable wind power forecast, obtaining Wind Power Plants' (WPP) power generation data in real time and constructing the power forecast model with historical production values is a desirable action plan. However, this situation may not be applicable for all WPPs in the country due to difficulties in obtaining such data from WPP in real time. Therefore, there is a need for upscaling algorithm for generating the power forecast of such WPPs and producing a regional power forecast for a given region or the whole country. In this work it is aimed to construct an upscaling wind power forecast model to answer these needs. Many models in the literature propose techniques for the estimation of the entire zone rather than an offline plant. Offline plants are the plants such that their production data is not available in the system. In this work, we propose a method for generating power forecasting for offline plants, and for a region at the same time. The technique is based on firstly power forecasting for offline plants, and then upscaling to region by using forecasts for both online and offline wind power plants. The performance of the method is experimentally evaluated with baseline and previous techniques and it is shown to provide higher accuracy for power prediction.

Reducing the cost of wind resource assessment: using a regional wind power forecasting method for assessment

“…As stated in Figure 4, by using the NWP and power data inputs, SHWIP and RegionalSHWIP models are run daily for every online and offline WPPs in the system in order to produce power forecasts for these plants. 16,35 For the online WPPs in the system, a data mining based wind power forecast model, namely Statistical Hybrid Wind Power Forecast Model (SHWIP), is developed. 16 This model dynamically clusters the weather events according to the most important weather forecast parameters including wind speed, wind direction, temperature, and pressure.…”

“…16 For the offline WPPs in the system, another data mining based model, namely Regional Statistical Hybrid Wind Power Forecast Model (RegionalSHWIP), has been developed and is operational in the center. 35 Unlike the other upscaling methods in the literature, RegionalSHWIP follows a bottom-up approach and can produce power forecasts for offline WPPs. RegionalSHWIP initially determines the online WPPs with most similar characteristics to the target offline WPP in terms of historical weather behavior.…”

“…In this work, in order to reduce the cost of assessment, we investigate the use of a regional forecast model namely Regional Statistical Hybrid Wind Power Forecast Technique (RegionalSHWIP) for Wind Resource Assessment problem. 16,34,35 RegionalSHWIP forecast model is proposed specifically for generating wind power forecasts for plants with no historical production data. 35 In constructing the model, two main inputs are used: the meteorological weather forecasts in the plant area and wind power forecasts of the other neighboring plants.…”

“…16,34,35 RegionalSHWIP forecast model is proposed specifically for generating wind power forecasts for plants with no historical production data. 35 In constructing the model, two main inputs are used: the meteorological weather forecasts in the plant area and wind power forecasts of the other neighboring plants. Regional power prediction models in the literature generally select some online power plants in a region and generate power estimates by modeling the region over these plants.…”

Multi-Step Short-Term Wind Power Prediction Based on Three-level Decomposition and Improved Grey Wolf Optimization

“…Medium-term wind power prediction ranging from one day to several days is mainly used for wind farm maintenance and scheduling plan. The existing wind power prediction methods can be divided into three main types: Physical models [2], Statistical models [3] and Hybrid models [4]. Physical models mainly focused on using mathematical model to describe the terrain and obtain more accurate Numerical Weather Prediction (NWP).…”