From Grid Operations to Asset Management: Deploying AI in Industrial Electrification

In this talk, Amandeep Singh Rana, Global Service Manager for Electrification and Automation at Siemens AG, makes the case that AI for industrial electrical grids is fundamentally different from general-purpose AI — and must be treated as such. This talk explores what it means to deploy AI responsibly and effectively within electrical infrastructure. Rana opens with a direct observation: physical laws do not bend for technology. Ohm's law and protection logic operate on fixed principles regardless of what any AI model recommends. In industrial electrical systems, AI cannot hallucinate. It cannot invent conditions or generate recommendations that are not grounded in system reality. This constraint defines how AI must be built and deployed in this domain. The session addresses the real pressures facing customers across multiple sectors. Distribution utilities face growing demands to connect new substations and renewable energy sources. Data centers operate in environments where even a single second of downtime carries significant cost. EV infrastructure is creating new load peaks and network congestion. Heavy industry is navigating the challenge of growing while decarbonizing, often with aging and complex asset bases. Across all these sectors, the structural challenges are similar, and adding new physical infrastructure alone will not resolve them. Rana describes three foundations required for AI to work reliably in this context: deep knowledge of physical assets such as transformers, switchgear, and substations; contextual data that combines operational records, maintenance logs, and event histories; and strict adherence to the laws of physics and domain engineering principles. He then presents a five-layer model through which AI creates value: sensing, interpreting, predicting, optimizing, and augmenting human decision-making. Value is built not in a single application but across planning, operations, outage management, and asset management. Two practical examples are discussed. In grid operations, AI can process large datasets rapidly and suggest switching actions during demand peaks — such as a surge in EV charging — while keeping human operators in control. In asset management, AI can support analysis of millions of electrical assets, including those not connected to real-time IoT systems, through multimodal and visual inspection approaches, capturing details that manual processes may miss. The talk, recorded at Hannover Messe 2026, closes with a reflection on the growing skills shortage in the electrification sector and how AI tools can support and augment the engineers working within it.