I-Corps: Short-term Wind Forecasting Engine

该项目的目的是演示短期风预测引擎的概念验证。在过去的十年里，全国的风力发电装机容量急剧增加。然而，由于间歇性的风，公用事业公司，电网运营商和风电场业主面临着重大困难。对于位于复杂地形区域的风电场，风的变化性进一步加剧。在短期风力发电预测和电网整合方面存在巨大的技术差距。目前的实践依赖于天气预报和历史风力分析，但使用这些方法进行电力预测容易出现较大的误差。该研究团队开发的预测引擎旨在通过现代图形处理单元以多尺度方式加速并行计算集群的快速执行，其中微尺度计算将由中尺度天气预测模型同时驱动。预测引擎内部采用的数值方法和湍流模型为复杂地形提供了极大的灵活性，并显著改善了对涡轮机轮毂高度的预测。为公用事业公司、电网运营商和风电场所有者提供准确的风力预测引擎对能源的高效生产和经济性具有广泛的影响。可靠的风力预测引擎将有助于增加整体能源组合中的风力发电量。准确的预测还有望帮助降低风能的电网整合成本，并使这种形式的可再生能源在智能电网技术推动的放松管制的电力市场中具有竞争力。预测引擎有可能提高电网的可靠性，因为当风力突然增加而没有任何关于风力的预先信息时，电网很容易过载。除了预测现有风电场的功率外，该技术还可用于增加现有输电线路容量，优化风电场布局，以实现最大发电量。

The purpose of this project is to demonstrate a proof-of-concept for a short-term wind forecasting engine. Installed wind power capacity across the nation has increased dramatically over the last decade. However, utility companies, grid operators and wind farm owners are facing major difficulties because of intermittent winds. The variability of wind is further exacerbated for wind farms that are located on complex terrain regions. Substantial technology gaps exist in short-term wind power forecasting and grid integration. Current practice relies on weather forecasting and historical wind analysis, but power forecasting using these approaches is prone to large errors. The forecasting engine developed by the research team is designed for fast execution on parallel computing clusters accelerated by modern graphics processing units in a multi-scale fashion, where micro-scale computations will be concurrently driven by meso-scale weather prediction models. The numerical methods and the turbulence model adopted inside the forecasting engine offer great flexibility for complex terrain and enable significantly improved predictions at the turbine hub height. The availability of an accurate wind forecasting engine for utility companies, grid operators and wind farm owners has a broad impact on the efficient production and the economics of energy. A reliable wind forecasting engine will help increase the amount of wind power in the overall energy portfolio. Accurate forecasting is also expected to help lower the grid integration cost of wind energy and make this form of renewable energy competitive in a deregulated electricity market that is expected to be driven by smart-grid technologies. A forecasting engine has the potential to enhance the reliability of the electricity grid, because the grid is vulnerable to overload when winds suddenly ramp up without any advance information on the wind. In addition to forecasting power from existing wind farms, the proposed technology can be used to increase existing transmission line capacity and optimize wind farm layouts for maximum power generation.