主讲人简介   

王者，香港科技大学助理教授、博导，香港科技大学智慧城市研究院副院长、智能建筑与建造实验室主任，研究方向为建筑能源系统模拟与优化，获2023年国家自然科学基金优秀青年基金，2019 国家科技进步二等奖，2022国际建筑机电人工智能挑战赛第一名，NeurIPS CityLearn Challenge 2023 第二名，清华大学首届苏世民学者，布什总统基金会Fellow。从2021年起，连续四年入选斯坦福大学和爱斯维尔联合发布的“全球前2%顶尖科学家”榜单。曾以青年学者代表身份受邀在刘延东副总理、克里国务卿主持的第七届中美人文交流高层对话（2016）上发言。

  讲座简介  

Data is the oil of the 21st century that fuels many smart building applications. Therefore, modern buildings have collected increasing amounts of data. However, those collected data are largely underutilized, partly because the complicated topology between sensors, devices, and spaces makes data analysis challenging and time consuming. In this work, we proposed an innovative tool, BuildingGPT, which allows the site engineers, data analysts, and building owners to engage with building data using natural language. Our tool comprises two key components. First, we employ a semantic model known as Brick to represent the relationships between data, sensors, devices, and spaces in a standardized, machine-readable way. Next, we developed a Vector-Graph Retrieval-AugmentedGeneration (VG-RAG) that enables interaction with the semantic model, facilitating queries and responses using natural language. We believe this tool can greatly enhance the efficient utilization of building data and unlock itssubstantial potential in areas such as optimal control and automatic fault diagnostics and detection.

  讲座信息