An interview with Dr. Chen Song

AI is becoming integral to smart buildings and smart cities, autonomously balancing energy demand, comfort, and cost efficiency across entire networks. ABB is positioning itself as a provider of secure, industry-ready AI with competitive advantages. Dr. Chen Song, Senior Scientist at ABB's Corporate Research Center in Germany, explains how these developments drive energy optimization and grid stability, and how ABB customers benefit from this transformation.

Which developments are driving AI in energy and industrial applications?   Dr. Chen Song: AI has evolved far beyond general reasoning and language tasks. We are now entering a new era, often referred to as Scientific Machine Learning or AI for Science. In this field, AI not only answers questions but also learns from physical models and sensor data.   This enables us to work on challenging engineering tasks more effectively—such as load forecasts and stability analyses in the grid or energy optimization with physics-informed AI models. The great advantage of scientific machine learning is that it integrates domain knowledge, physics, boundary conditions, and other knowledge about the underlying system. This means that AI systems behave robustly and predictably, even in safety-critical applications such as energy grids or industrial automation.   However, data quality is still a major challenge and will remain so, at least in the near future.

What distinguishes ABB's AI from generative systems like ChatGPT?   Dr. Chen Song: ABB has very strong and unique foundations in electrification, automation, and drive technology—fields that require deep engineering knowledge and high reliability.   This gives us a competitive advantage when it comes to using AI in real industrial systems. In comparison, ChatGPT or other generative models are primarily developed for language understanding and reasoning. In an industrial environment, however, data quality remains a major challenge—and will continue to be so in the foreseeable future.   "What sets ABB apart is our more than one hundred years of experience and knowledge of the systems and devices that we develop, build, and operate."   This puts us in the ideal position to combine data with physical models, expert knowledge, and operational experience—something that pure generative AI providers cannot do in this form.

How does the combination of physics-based and data-driven AI actually work—and what exactly is the advantage for ABB?   Dr. Chen Song: The combination of physics-informed and data-driven AI can make our engineering work both more efficient and more accurate.   One example: OPTIMAX® is our software suite for energy management and optimization. The optimization modules are based on high-precision system models, for example of power plants, district heating networks, or water systems. The development of such models used to be very engineering-intensive and required in-depth domain knowledge.   To reduce this effort and increase accuracy at the same time, we have introduced Physics-Enhanced Neural Ordinary Differential Equations (PeNODEs) - a hybrid AI approach that combines mathematical physics models with machine learning, into the engineering workflow.- This approach is used within the framework of the ITEA OpenScaling project together with our research partners and industrial partners.   By closely linking physical principles with field data, PeNODEs gives us two major advantages: higher model accuracy, because the neural network learns the residual dynamics and supplements the physics. Secondly, less engineering effort, as fewer manual modeling steps are required.

AI is becoming an integral part of smart buildings and smart cities. What does the future look like?   Dr. Chen Song: In the future, systems will autonomously balance energy demand, comfort, and cost efficiency—not only within individual buildings but across entire networks. AI will facilitate real-time balancing and predictive maintenance at a city-wide scale.   ABB is positioning itself as a provider of secure, industry-ready AI, offering significant advantages over the competition. Our approach emphasizes security, reliability, and transparency—critical factors that distinguish us in the market.       How does ABB ensure data security and transparency in operations?   Dr. Chen Song: Data security and transparency are key principles according to which ABB designs and delivers digital and AI solutions. All our systems follow strict industrial cybersecurity standards and are developed according to ABB's secure-by-design philosophy. This includes encrypted communication and continuous monitoring.   At the same time, ABB ensures full transparency when handling data.   "Customers always retain ownership of their operating data. Our networked products provide clear information about what data is collected and how it is processed—in line with regulatory requirements such as the EU Data Act."   Wherever possible, AI is operated on-premises or at the edge¹, which minimizes data movements and keeps information in the customer's infrastructure.   In addition, we attach great importance to making our AI models comprehensible and explainable by integrating physics-based logic and explainability functions.

How do planners, operators, and installers actually benefit from the use of AI in everyday life?   Dr. Chen Song: Planners, operators, service technicians, and installers often work in environments with high stress levels. Decisions have to be made quickly, and mistakes can result in significant safety risks, downtime, or unnecessary costs.   This load can be significantly reduced with an industrial AI assistant. AI systems that are based on ABB's collective engineering knowledge and understand the physical behavior of the systems can provide support on several levels—for example with better planning and optimization or faster troubleshooting and more efficient maintenance.   One example: We have a system to support service technicians which is based on historical service experience. If a root cause analysis needs to be carried out on site, the system provides step-by-step instructions based on the faults or symptoms observed. This allows problems to be identified more quickly and with greater certainty.   Overall, AI helps to operate systems more safely, reliably, and sustainably. It gives people in the field the certainty that they can draw on the collective knowledge of many experts when they need to make decisions.   Dr. Chen Song, thank you for this conversation!

ABOUT DR. CHEN SONG

Senior Scientist ABB Corporate Research Center, Germany Dr. Chen Song is a Senior Scientist in Industrial AI at the ABB Corporate Research Center in Germany. He has outstanding interdisciplinary expertise in scientific computing, finite element methods, high-performance computing, uncertainty quantification, and modern machine learning methods. After graduating in mechanical engineering in Lyon, he completed his doctorate in mathematics at the University of Heidelberg.

Footnotes:

1. On-premises: in your own data center. Edge: directly at the system, close to the data source—sensitive data remains in-house.