Predictive emission monitoring system for gas turbine

在第一年，学生将致力于对管理燃气轮机运行的核心原则以及如何在这个问题的背景下使用使用机器学习方法的数据驱动模型有一个坚实的理解。在第二年，学生将考虑已建立的燃气轮机运行的物理模型，以及如何开发新的方法来利用大量的燃气轮机运行数据，这些数据结合在一起可以提供准确的排放预测。在第三年，研究将集中于开发新的、可推广的机器学习方法，这些方法可以提取良好的表示，与物理模型相结合，可以在不同的现实生活环境中提供准确的预测。在第四年，学生将在几个燃气轮机模型上进行广泛的验证实验，同时还将撰写博士论文(最后六个月)。

In the first year the student will work on developing a solid understanding of core principles governing the operations of gas turbines and how data-driven models that use machine learning approaches can be used in the context of this problem. In the second year, the student will consider established physical models of gas turbine operations and how novel approaches can be developed to leverage the vast amount of gas turbine operational data that combined can provide accurate predictions of emissions. In the third year the investigations will focus on developing novel and generalisable machine learning approaches that can extract good representations, which coupled with physical models can provide accurate predictions across different real-life settings. In the fourth year, the student will carry on extensive validation experiments across a few gas-turbine models, whilst also writing-up the PhD thesis (last six months).