Artificial Intelligence Methods For Electricity Market Imbalance Prediction

该博士的核心目标是识别英国电力不平衡的驱动因素，创建预测工具以更好地预测它们，然后为平衡市场形成概率价格预测模型。新的人工智能方法，特别是贝叶斯深度学习，将用于预测模型的开发。这些预报将在不同的空间和时间范围内进行，以确定最佳的预报分辨率，为系统运营商提供最大的价值。这将需要使用来自英国和欧洲的多个现有和新的数据集。拟议的博士学位主题涉及人工智能方法的开发，该方法可以利用与电力市场相关的各种不同的数据，如天气预报，可再生能源发电生产，历史需求和许多其他因素来开发月，每半小时周期的市场不平衡量/价格的周前和日前预测。研究还将涉及估计这些预测的不确定性水平，并对电力市场结构进行理解，并解释研究结果，以促进更好的市场设计。在你的博士学位，你将有望开发从强化学习和神经网络到贝叶斯深度学习的人工智能方法的专业知识。该项目是一个高度跨学科的项目，需要了解电力系统，市场设计和先进的人工智能方法，理想的候选人应该是舒适的，并有兴趣与这些领域的同事合作。

The core objective of the PhD is to identify drivers of UK electrical imbalance, create forecasting tools to better anticipate them, and then form a probabilistic price forecasting model for the balancing market.Novel artificial intelligence methods, specifically Bayesian deep learning, will be employed for the forecast model development. The forecasts will be carried out across varying spatial and temporal horizons to identify the optimum forecast resolution which provides the most value to system operators. This will require the use of multiple existing and new datasets sourced from across the UK and Europe.The proposed PhD topic involved the development of Artificial Intelligence approaches that can make use of diverse and disparate data relevant to electricity markets such as weather forecasts, renewable generation production, historical demand and many other factors to develop month, week and day ahead forecasts of the imbalance volume/price in the market per half hourly period. The research will also involve estimating the level of uncertainty around these predictions and developing an understanding of electricity market structure and interpreting findings to promote better market design. In your PhD you will be expected to develop expertise in Artificial intelligence methods from reinforcement learning and neural networks to Bayesian deep learning. The project is highly interdisciplinary one requiring an understanding of power systems, market design and advanced artificial intelligence methods and the ideal candidate is expected to be comfortable and interested in working with colleagues across these domains.