State-of-the-art solar PV generation forecast for individual PV systems

在SBRI的支持下，Open Climate Fix(OCF)将为单个光伏系统开发一个开源的最先进的太阳能光伏发电预测。该预测将帮助我们的第一批用户优化和更好地管理他们的太阳能、存储或分布式能源资产，从而为所有能源用户减少二氧化碳排放和成本。此外，随着太阳能的扩大，在不久的将来，它可以在个人家庭层面上使用。正如我们的市场研究所表明的那样，显然需要更准确的光伏太阳能发电量预测。目前可用的预报大多仅依赖数值天气预报(NWP)，没有利用所有可用数据，难以预测未来的短期水平，这在做出资产运营决策时非常重要。我们的预测有何不同？我们的短期预测解决方案结合了尖端的机器学习(ML)研究，结合了一系列可用的数据源，例如来自数千个独立光伏系统的五分钟卫星图像、数值预报和光伏数据。卫星数据提供的天气信息只有五分钟，而不是几个小时前的信息，就像核子预报一样。目前，气象学家分析卫星图像，但大多数小公司没有资源雇用这些技能。卫星图像尚未被广泛应用于传统的太阳能光伏预测方法，因此我们特别有兴趣将其纳入到我们的模型中来改进我们的模型。OCF是一个非营利性产品实验室，完全专注于减少二氧化碳排放。该组织的每个部分都旨在最大化对气候的影响，例如我们的开放和协作方法、我们的快速原型制作以及我们对寻找可扩展和实用的解决方案的关注。OCF将开放源代码倡议(OSI)管理的开源原则和麻省理工学院的许可证应用于我们的所有解决方案。我们将把同样的开源方法应用于SBRI项目。我们将通过GitHub公开所有代码，供其他研究人员贡献和学习。我们正在致力于实现来自Google Research和DeepMind的机器学习前沿的当前文献。到目前为止，我们所有的工作都可以在[GitHub][0]上找到。凭借在领先技术公司和机器学习方面拥有丰富经验的团队，我们带来了最新的协作方法，并将它们与能源行业多年的经验相结合。通过这种使命驱动的方法，我们已经吸引了巨大的兴趣，并围绕我们的工作建立了一个贡献者社区。[0]：https://github.com/openclimatefix

With SBRI support Open Climate Fix (OCF) will develop an open-source state-of-the-art solar PV generation forecast for individual PV systems. The forecast will help our first users to optimise and better manage their solar energy, storage or distributed energy assets, which reduces CO2 emissions and costs for all energy users. Additionally, with the expansion of solar, it could be used at the individual household level in the near future.As shown in our market research, there is a clear need for more accurate PV solar generation forecasts. Most of the forecasts currently available rely solely on numerical weather predictions (NWPs), do not make use of all the data available and struggle to forecast short time horizons ahead, which are important when making operational decisions on assets.How is our forecast different? Our near-term forecasting solutions incorporate cutting-edge machine learning (ML) research combining a range of available data sources such as five-minutely satellite imagery, NWPs, and PV data from thousands of individual PV systems. Satellite data provides weather information that is only five minutes old as opposed to hours, as in the case of NWPs. Currently, meteorologists analyse satellite imagery, but most small companies do not have the resources to hire these skills. Satellite imagery has not yet been widely used in traditional solar PV forecasting methodologies, thus we are particularly interested in incorporating it to improve our models.OCF is a non-profit product lab, fully focused on reducing CO2 emissions. Every part of the organisation is designed to maximise climate impact, such as our open and collaborative approach, our rapid prototyping, and our attention on finding scalable & practical solutions.OCF applies open-source principles stewarded by the Open Source Initiative (OSI) and an MIT licence to all our solutions. We will apply the same open-source approach to the SBRI project. We will make all the code openly available for other researchers to contribute to and learn from via GitHub. We are working on implementing current literature at the cutting edge of machine learning from Google Research and DeepMind. All our work to date is available on [GitHub][0]. With a team with deep experience in leading technology firms and machine learning, we bring the latest collaborative approaches and combine them with many years' experience in the energy industry. Through this mission-driven approach, we have already attracted significant interest and established a community of contributors around our work.[0]: https://github.com/openclimatefix