Publications

Integrating symbolic neural networks with building physics: A study and proposal

Journal of Building Engineering, 2025

Explores Kolmogorov–Arnold Networks (KAN) for symbolic learning in building physics, demonstrating improved interpretability, robustness under data scarcity, and potential for physics-informed knowledge discovery.

Chen, X., Lv, G., Zhuang, X., Duarte, C., Schiavon, S., & Geyer, P. (2025). Integrating symbolic neural networks with building physics: A study and proposal. Journal of Building Engineering, 113033.

Explainable AI for engineering design: A unified approach of systems engineering and component-based deep learning

Advanced Engineering Informatics, 2024

Introducing a groundbreaking intersection of systems engineering and deep learning: our component-based, data-driven model. This unique approach not only adapts to the inherent structure of systems engineering but also prioritizes domain knowledge, ensuring that our models are interpretable and aligned with engineering principles. Dive deep into our innovative methodology that bridges the gap between the vast potential of AI and the meticulous precision of engineering design.

Geyer, P., Singh, M. M., & Chen, X. (2024). Explainable AI for engineering design: A unified approach of systems engineering and component-based deep learning demonstrated by energy-efficient building design. Advanced Engineering Informatics, 62, 102843.

Utilizing domain knowledge: robust machine learning for building energy performance prediction with small, inconsistent datasets

Knowledge-Based Systems, 2024

In the engineering domain, most systems are composed of the same basic components with diverse compositions. By disentangling the compositionality from the system and embedding it into the model organization, we actually can construct a set of finite “Lego-Block” to predict infinite combinations. We proved that Component-based Machine Learning (CBML) turns the extrapolation problem at the system level into the interpolation problem at the component level in a more flexible and less data-reliance manner.

Chen, X., Singh, M.M. and Geyer, P., 2024. Utilizing domain knowledge: robust machine learning for building energy performance prediction with small, inconsistent datasets. Knowledge-Based Systems, p.111774

Machine learning in proton exchange membrane water electrolysis — A knowledge-integrated framework

Applied Energy, 2024

This paper demonstrates the general applicability of the knowledge-integration machine learning framework. It is showcased through real-world PEMWE case studies, highlighting its potential to drive sustainable energy solutions and pave the way for prior knowledge alignment in specialist AI systems.

Chen, X., Rex, A., Woelke, J., Eckert, C., Bensmann, B., Hanke-Rauschenbach, R., & Geyer, P. (2024). Machine learning in proton exchange membrane water electrolysis — A knowledge-integrated framework. Applied Energy, 371, 123550.

Using causal inference to avoid fallouts in data-driven parametric analysis: A case study in the architecture, engineering, and construction industry

Developments in the Built Environment, 2023

This research proposes a symbiosis pattern between human intelligence and machine capabilities with the importance of causal analysis in preventing biases in data-driven models and prior knowledge: the necessity of understanding the underlying connections – the why and how – behind each element to accurately solve the engineering puzzle.

Chen, X., Sun, R., Saluz, U., Schiavon, S. and Geyer, P., 2023. Using causal inference to avoid fallouts in data-driven parametric analysis: A case study in the architecture, engineering, and construction industry. Developments in the Built Environment, p.100296.

Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX

Biomedical Signal Processing and Control, 2023

Are the limits of our language the limits of our world? Just as we use words to share ideas, our brains have their own unique language, spoken through electrical signals. To better understand this language, this research focuses on developing cutting-edge deep learning methods: EEGNeX, a tool for decoding brain signals to explore how a blend of technology and neuroscience might just change the way we communicate.

Chen, X., Teng, X., Chen, H., Pan, Y., & Geyer, P. (2024). Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX. Biomedical Signal Processing and Control, 87, 105475.

Pathway toward prior knowledge-integrated machine learning in engineering

In Proceedings of 18th International IBPSA conference and Exhibition, Building Simulation 2023, 2023

Proposing a paradigm to systematically embed prior knowledge and domain insights into data-driven models, addressing domain-specific challenges through three levels of knowledge integration: modeling knowledge for system description, inductive logic and disentanglement for extrapolation, and abstract reasoning and deductive logic for representation. We demonstrated this paradigm in the context of building engineering.

Chen, X., & Geyer, P. (2023). Pathway toward prior knowledge-integrated machine learning in engineering. arXiv preprint arXiv:2307.06950.

Sustainability recommendation system for process-oriented building design alternatives under multi-objective scenarios

In Proceedings of 30th International Workshop on Intelligent Computing in Engineering, EG-ICE 2023, 2023

This manuscript introduces a recommendation system as an application extension practice for machine assistance in the sustainable building design domain. By involving multi-objective optimization processes, this system aims to be the bridge between intricate evaluations and tangible, sustainable outcomes in building engineering for achieving holistic sustainability objectives.

Chen, X., & Geyer, P. (2023). Sustainability recommendation system for building design alternatives under multi-objective scenarios, accepted by 30th International Workshop on Intelligent Computing in Engineering, EG-ICE 2023, London, UK.
Online Demo

Introducing causal inference in the energy-efficient building design process

Energy and Buildings, 2022

In the design and decision-making process, distinguishing between correlation and causation is not just a philosophical mindset but a crucial methodological consideration. As we transition from intuition-driven decisions to data-driven methodologies, it’s necessary to discover potential patterns directly from the raw data rather than rely heavily on potentially biased or incorrect prior knowledge. This manuscript underscores the significance of enabling machine learning models to conduct causal reasoning, ensuring that answers to “What-if” questions are unbiased.

Chen, X., Abualdenien, J., Singh, M. M., Borrmann, A., & Geyer, P. (2022). Introducing causal inference in the energy-efficient building design process. Energy and Buildings, 277, 112583.

Integrated data-driven and knowledge-based performance evaluation for machine assistance in building design decision support

In Proceedings of 29th International Workshop on Intelligent Computing in Engineering, EG-ICE 2022, special issue, 2022

This research implements a practice of the integration between a data-driven model and knowledge-based methods within the same context in the early phases of building design. The objective is to enhance decision-making to achieve augmented intelligence, ensuring both energy efficiency and minimal environmental impact in the given scenario.

Chen X., Saluz U., Staudt J., Margesin M., Lang W., & Geyer P. (2022). Integrated data-driven and knowledge-based performance evaluation for machine assistance in building design decision support, In the 29th International Workshop on Intelligent Computing in Engineering, EG-ICE 2022. Aarhus, Denmark.

A hybrid-model forecasting framework for reducing the building energy performance gap

Advanced Engineering Informatics, 2022

An intuitive way to combine knowledge-based simulation for data enrichment while introducing extra information to machine learning models. This hybrid-model framework ensures that both domain knowledge and data are well utilized, reducing the effort in detail modeling, and minimizing the performance gap in prediction.

Chen, X., Guo, T., Kriegel, M., & Geyer, P. (2022). A hybrid-model forecasting framework for reducing the building energy performance gap. Advanced Engineering Informatics, 52, 101627.

A Dynamic Feedforward Control Strategy for Energy-efficient Building System Operation

In Proceedings of Passive and Low Energy Architecture, PLEA 2022, 2022

Leveraging the confluence of simulation, first-principles models, and machine learning, we introduce a pioneering approach that marries the strengths of feedforward and feedback loops. Think of it as a reinforcement learning paradigm: we construct a dynamic environment, allowing the model to adapt and perform optimally. This synergy crafts a dynamic and highly efficient building system control strategy, all conceptualized within a ‘gray-box’ framework. By embedding dynamic prior knowledge of building system characteristics, we not only enhance the model’s behavior but also tap into previously unexplored energy-saving potentials, paving the way for the future of optimized built environments.

Chen X., Cai X., Kümpel A., Müller D., & Geyer P., (2022). Dynamic Feedforward Strategy Development for Building Heating System based on AI Forecasting and Simulation. In Passive and Low Energy Architecture, PLEA 2022, Santiago de Chile, Chile.

Machine assistance in energy-efficient building design: A predictive framework toward dynamic interaction with human decision-making under uncertainty

Applied Energy, 2022

Drawing inspiration from the human nervous system’s estimation process, we present a key framework designed to revolutionize how we approach complex decision-making challenges across various domains. This framework seamlessly integrates uncertainty quantification, cutting-edge machine learning techniques, and first-principles models, fostering a dynamic and ongoing interaction with users. We elaborated this framework into the building design scenario. More than just a decision aid, it lays the foundational principles for integrating domain knowledge with machine learning and propelling the frontier of intelligence augmentation. It’s not just about making informed decisions, but about enhancing our capability to interact with, interpret, and innovate from the vast data-driven landscapes of the modern world.

Chen, X., & Geyer, P. (2022). Machine assistance in energy-efficient building design: A predictive framework toward dynamic interaction with human decision-making under uncertainty. Applied Energy, 307, 118240.
Online Demo

欧盟-德国建筑碳中和前沿; Frontiers of carbon neutrality in EU-German building sector

暖通空调; Heating Ventilating & Air Conditioning, 2022

A thorough overview of carbon neutrality in the EU regarding policy framework, technology support, economics foundation, and recycling strategy.

陈夏，张怡卓，蔡晓烨．欧盟-德国建筑碳中和前沿［Ｊ］．暖通空调，２０２２，５２（３）：２５-３８． Chen X., Zhang Y., & Cai X. (2022). Frontiers of carbon neutrality in EU-German building sector, Heating Ventilating & Air Conditioning, TU-023; X322.

A hybrid-model forecasting framework for reducing the building energy performance gap

In Proceedings of 28th International Workshop on Intelligent Computing in Engineering, EG-ICE 2021, special issue, 2021

Introducing a cutting-edge approach to bridging the persistent energy performance gap in the building industry: By synergistically merging domain knowledge from building simulations with the prowess of machine learning, our hybrid-model forecasting framework not only delivers enhanced prediction accuracy but also provides a deeper understanding of the underlying reasons for discrepancies in building energy consumption. Drawing on real-world data from an office building in Shanghai, this manuscript showcases how the integration of first-principles with data-driven methods can pioneer a new path to minimizing energy disparities.

Chen, X., Guo, T., & Geyer, P. (2021). A hybrid-model forecasting framework for reducing the building energy performance gap. In 28th International Workshop on Intelligent Computing in Engineering, EG-ICE 2021. Berlin, Germany, 2021, special issue on Advanced Engineering Informatics.

Component-based machine learning for predicting representative time-series of energy performance in building design

In Proceedings of 28th International Workshop on Intelligent Computing in Engineering, EG-ICE 2021, 2021

Imagine building performance prediction as a dynamic Lego set, where each block represents a distinct component of the design. Just as Lego pieces can be assembled in countless ways to create varied structures, our component-based machine learning (CBML) approach offers a modular and scalable solution for predicting time-series energy performance in building designs. Dive into our manuscript to uncover how we’ve redefined building performance simulation with CBML, ensuring accurate hourly energy predictions that pave the way for optimized equipment sizing, efficient energy demand control, and balanced network loads.

Chen, X., Singh, M.M. & Geyer, P. (2021). Component-based machine learning for predicting representative time-series of energy performance in building design. In 28th International Workshop on Intelligent Computing in Engineering, EG-ICE 2021. Berlin, Germany.

Xia Chen

Publications