孙科 · Sun Ke

AI × Energy Systems

EN

Research

RESEARCH

Turning complex urban and building data into explainable, actionable evidence

My work sits at the intersection of artificial intelligence, urban energy systems, and structured semantics. Rather than listing titles alone, this page explains the problem behind each study, the approach taken, and the insight that emerged.

Journal articles

Problem

Conventional urban energy analysis usually stops at final energy. It cannot show how much useful service remains after conversion into lighting, heat, mobility, or industrial processes, nor explain where efficiency differences between cities come from.

Approach

We developed a city-wide useful-work model linking more than 100 energy-conversion processes across 48 end-use sectors to economic structure, development level, and energy mix. It covers 63 cities in the Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta regions for 2010, 2015, and 2020.

Findings

City-wide useful-work efficiency ranged from 17.2% to 47.4%, indicating substantial room for improvement. Industry and power generally dominated demand, northern cities used more useful work per unit of economic output, and a sublinear relationship between useful work and output appeared in 2015 and 2020.

Why it matters

The study moves the question from “how much energy was consumed?” to “how much service did that energy ultimately provide?” This helps identify conversion losses and design differentiated strategies for cities with different industries, climates, and development stages.

Problem

Urban energy-carbon studies often focus on one infrastructure system or direct emissions within administrative boundaries. This can miss supply-chain impacts from mobility, materials, food, and other systems, while assuming that energy savings and carbon mitigation always move together.

Approach

We built a multi-infrastructure accounting framework for direct and embodied energy and carbon footprints across 25 cities in Jiangsu, Zhejiang, and Shanghai from 2010 to 2020, then used spatial autocorrelation to identify regional clustering and change.

Findings

Energy supply, mobility, and construction materials were the major contributors. Food and waste together represented about 4% of energy footprints but 8%–9% of carbon footprints. Transboundary footprints generally increased, energy and carbon did not always change proportionally, and spatial correlation between cities weakened.

Why it matters

Cities cannot rely only on local emissions or a single energy indicator. Effective intervention must consider infrastructure type, supply-chain position, and possible decoupling between energy and carbon.

Conference presentation

Developing process-based modelling to quantify urban energy service: Using two cities as cases City+2023 · Perth International Conference

Selected projects

Smart-city GIS platform

Engineering enterprise

Integrated multi-source GIS data and developed an interactive Web GIS platform with Django and Vue. Dijkstra and A* path algorithms were used to analyse access to facilities and support urban-planning decisions.

ESG report automation and data verification

EY Shanghai

Combined OCR and translation APIs in a batch-processing workflow for bilingual ESG reports, while building reusable corpora for internal AI model training.

Carbon accounting for large substations

State Grid Yingda Carbon Asset Management

Calculated construction-stage carbon emissions, cleaned large datasets with Pandas, and used Matplotlib to analyse and communicate mitigation benefits.

0%