Electricity and Control August 2025

Industry 4.0 + IIoT

WRITE @ THE BACK

Building a data foundation for AI in manufacturing In a recent Belden blog post, Sam Veng, Digital Automation Consultant, highlighted the importance of building a data foundation as essential to using AI in manufacturing. He outlined the steps to take to establish a unified and accessible data foundation. We share his guidelines here.

Sam Veng, Belden.

Connected, unified and accessible plant data unlocks actionable insights and more.

T alk of AI for manufacturing may be everywhere, but there’s one big problem to overcome when putting the technology to work in the real world: Before AI can deliver on its potential, organisations must be able to unlock the data AI needs in order to generate meaningful, actionable insights and drive real eˆiciency gains. Building a unified, accessible data foundation This is the first step to scaling AI for manufacturing. It means: ƒ Bringing together data that’s scattered across diˆerent systems and formats ƒ Breaking down silos between departments and tech nologies ƒ Ensuring data quality and consistency across all sourc es ƒ Labelling and organising it in a way that’s useful for AI ƒ Enabling real-time access for faster, more responsive applications. But these are not easy jobs in today’s world, where the volume of enterprise data is exploding. Most plants can no longer claim ‘data lakes’. Instead, they are dealing with ‘data oceans’. The scale and complexity of information is overwhelming. Data is being created faster than ever before, and it is coming from more sources than organisations and their teams can manage. This data is also o“en scattered across diˆerent systems, trapped in silos and moving through networks that were not designed to

support real-time analytics. This fragmentation and complexity make building a data foundation for AI in manufacturing more diˆicult. Steps to prepare your data for AI success To deploy and scale AI for manufacturing, you must be able to identi fy, connect and prepare the right data. Here’s how to make it happen. ƒ Determine what data matters: Assess your operations to decide which data sources have the greatest impact on eˆiciency, quality and business outcomes. That’s the infor mation you need if you want to optimise processes, drive informed decision-making and achieve measurable im provements in productivity, throughput and quality. ƒ Find and connect that data: Modern operations depend on seamless communication between people, devices and processes. But if you don’t know where your data is coming from, or if it’s managed independently in discon nected systems, then it’s useless to you and your AI models. To address this, you must identify and connect valuable data from every corner of your plant, breaking down silos between systems and departments. For example, con necting information from production lines, quality control, maintenance logs and IoT sensors can enable real-time monitoring, predictive maintenance and process optimi sation. ƒ Bring all your data together: Once you’ve identified and con

4 Electricity + Control AUGUST 2025