Xincong YANG

Xincong YANG

Assistant Professor @HITSZ

Research on Key Technologies for Digital Twin of Dual-Carbon Goal Driven by Knowledge and Data, RMB 300,000 (On-going)

Highlights

  • Knowledge-driven small-scale space carbon emission distribution and short-term carbon footprint tracer simulation research—through a series of geometric, physical and behavioral models, empirical knowledge of rules, and domain knowledge for carbon emissions and carbon footprints, in the digital twin In the carbon system, simulate the dual carbon effects of existing buildings and communities, explore the distribution of carbon emissions in small-scale spaces and its efficient visual representation method, and clarify the trace accounting of unilateral short-term carbon footprint and its visual traceability method.
  • Experimental research on large-scale spatial carbon emission distribution and long-term carbon footprint tracer based on data-driven — Based on long-term sampling of multi-modal big data at the urban regional level, an artificial intelligence deep learning graph network is established in the digital twin carbon system The model considers multi-factor coupling and establishes adjustable hyperparameters, explores the distribution of large-scale spatial carbon emissions and its efficient visual representation method, and clarifies the tracing and accounting of multi-sided long-term carbon footprints and its visual traceability method.

  • Knowledge and data collaboratively drive the research and development of the digital twin carbon system—combining the knowledge-driven method of mechanism model, prior knowledge or rules with data-driven methods such as deep learning and reinforcement learning, the exploration is carried out at the architecture level and algorithm level The collaborative integration of knowledge and data-driven models enables accurate simulation of carbon emissions and carbon footprints in time and space, and builds a dual carbon observation and analysis system that is “measurable, predictable, reportable, and verifiable”.

  • 基于知识驱动的小尺度空间碳排放分布与短时碳足迹示踪模拟研究——通过一系列几何、物理和行为模型、规则经验知识以及面向碳排放与碳足迹的领域知识，在数字孪生双碳系统中模拟既有建筑、社区的双碳效应，探索小尺度空间碳排放分布规律及其高效可视化表征方法，明确单侧短时碳足迹的示踪核算及其可视化追溯方法。
  • 基于数据驱动的大尺度空间碳排放分布与长时碳足迹示踪试验研究——根据长期采样的城市区域级别的多模态大数据，在数字孪生双碳系统中建立人工智能深度学习图网络模型，考虑多因素耦合并设立可调节超参数，探索大尺度空间碳排放分布规律及其高效可视化表征方法，明确多侧长时碳足迹的示踪核算及其可视化追溯方法。
  • 知识与数据协同驱动数字孪生双碳系统的研究与开发——结合机理模型、先验知识或规则的知识驱动方法与深度学习、强化学习等数据驱动方法，探究在架构级和算法级上进行知识与数据驱动模型的协同融合，实现碳排放和碳足迹在时间和空间上的精准模拟，构建“可测量、可预估、可报告、可核查”的双碳观测与分析系统。