Development of machine learning approaches to geotechnical design of marine renewable energy foundations

该项目将开发一种优化工具，将岩土和基础设计考虑因素纳入风电场布局优化，以及文献中通常关注的风力和能源生产条件。在开发基于神经网络的自动化设计工具，快速模拟尖端的单桩设计技术后，优化和/或遗传编程将用于基于各种设计约束进行整个站点设计。该项目直接适用于正在进行的向可再生能源过渡的努力。自动化的桩设计方法将会通过使用土工离心机和/或数值模拟进行的实验室测试进行验证和测试。将使用风电场开发的现场数据演示风电场布局优化方法。该项目将利用南安普顿广泛的岩土工程实验室设施以及海事和机器学习优势。

This project will develop an optimisation tool that allows for geotechnical and foundation design considerations to be incorporated into windfarm layout optimisation, along with the wind and energy generation conditions that are typically focused on in literature. After developing a Neural Network based automated design tool that rapidly emulates cutting edge single monopile design techniques, optimisation and/or genetic programming will be used to carry out whole site design based on various design constraints. The project is directly applicable to the ongoing efforts to transition to renewable energy.The automated pile design methodology will be validated and tested using laboratory tests carried out with the geotechnical centrifuge and/or numerical modelling. The windfarm layout optimisation methodology will be demonstrated using site data from windfarm developments.The project will take advantage of the extensive geotechnical laboratory facilities alongside maritime and machine learning strengths at Southampton.