Building a strong data foundation: Preparing utilities for AI integration

As utility operators across the globe contend with a rapidly evolving landscape, they are facing significant challenges, such as rising electricity demand, the decentralisation of energy production, and the transition to electrification. This shift is accompanied by an ever-growing grid complexity that requires more sophisticated management solutions, writes David Cottingham, CTO at IQGeo.

Artificial intelligence (AI) is emerging as a transformative tool to help navigate this complexity. From automating grid management to enhancing operational safety and improving customer service, AI is rapidly becoming a cornerstone of modern utility operations.

As a report from Moorhouse outlines: "Utility providers face challenges such as the growing complexity of the grid architecture, rising demand for electricity, increasingly onerous regulatory obligations, and decarbonisation. Leveraging AI can help address these challenges by automating grid management, improving safety, enhancing customer service, and supporting a sustainable transition."

However, the successful use of AI is directly tied to the quality of the network data upon which it operates. Without accurate, up-to-date information on network assets, configurations, and locations, even the most advanced AI systems are limited in their capacity to deliver meaningful results. Utilities therefore must first focus on building a solid data foundation to fully unlock the untapped potential of AI and machine learning (ML).

The Importance of network data quality

At the core of every utility company is its physical network infrastructure. Over time, these networks have become more complex as they grow, involving many different parts from substations and transformers to power lines and metres.

This complexity requires utility operators to have highly accurate and comprehensive data about their physical assets, including their location and how they interconnect. Incomplete or outdated network data can lead to operational inefficiencies, increased risks, poor decision-making, and unnecessary costs.

For AI and ML systems to provide utility operators with actionable insights, they require precise, high-quality data. This is especially true as these systems are asked to manage more advanced tasks, such as automating grid maintenance, predicting equipment failures, or optimising energy distribution. The principle of "garbage in, garbage out" applies here—AI cannot function effectively without a reliable data foundation.

Have you read?
How AI is ‘bending the curve’ on energy transition innovation
It’s time for utilities to evolve for the new energy customer says AGL’s Jo Egan

To ensure this strong foundation, utility operators should prioritise creating a complete and accurate network model, which involves comprehensive asset management, ongoing network documentation, and integrated systems.

With regards to comprehensive asset management, this is precisely documenting the location and architecture of assets through geospatial technologies. With ongoing documentation, this involves maintaining accurate network data as an ongoing process, requiring continuous updates and significant investment to ensure relevance and, transitioning from fragmented data management systems to integrated solutions that are specifically designed to support the full lifecycle of network operations.

Enhancing network optimisation efforts

Many utility operators are still in the early stages of their digital transformation journeys.

In some cases, this might involve moving from paper records to digital systems, while others may already be working with outdated geographic information systems (GIS) and a patchwork of siloed applications. These legacy systems often create significant roadblocks as utilities seek to meet increasing demand, navigate regulatory complexities, and modernise their networks.

An optimised network is not only essential for efficient operations but also critical for long-term profitability. To meet the demands of the future, utility operators need to streamline their processes for planning, designing, building, and maintaining their physical infrastructure. This means investing in the right technologies and making incremental improvements in network data management.

Steps to prepare for AI and ML

To successfully integrate AI into their operations, utility providers must first ensure that their network data is complete, accurate, and actionable. This involves adopting a series of best practices to build a strong data foundation:

• Conduct a data quality audit: Start by assessing the current state of network data to identify any gaps or inaccuracies. This helps utilities prioritise areas for improvement and ensure that AI systems are working with the best possible network data.
• Invest in integrated network management systems: Replace disparate data management tools with integrated solutions designed to manage the entire lifecycle of network operations. These systems offer a more holistic view of network data, reducing the likelihood of errors and inefficiencies.
• Implement rigorous network data management practices: Establish clear workflows for continuous data updating and verification to ensure that network documentation remains accurate over time.
• Select reliable technology partners: Work with vendors who specialise in utility solutions, particularly those focused on capturing and managing high-quality network data. These partners can provide software and tools that streamline data collection both in the office and in the field.
• Focus on incremental improvements: Instead of attempting a large-scale, costly overhaul, utilities should focus on manageable, systematic upgrades. For example:
  
  1. Digitising paper records in priority regions to reduce inaccuracies.
  2. Upgrading legacy GIS systems incrementally, starting with the most critical areas of the network.
  3. Implementing pilot programs that utilise drones, mobile GIS units, or other advanced technologies to gather more accurate data in the field.
  4. Gradually improving data integration by connecting siloed systems one department at a time.
  
  These smaller, targeted steps help utilities reduce the risks and costs associated with large-scale projects while ensuring steady improvements in data quality and network performance.

The long-term benefits of data quality for AI and ML

By prioritising data quality, utility operators are not just addressing immediate operational challenges—they are laying the groundwork for future technological advancements.

AI and ML technologies thrive on accurate, well-organised data, and as utilities improve their network data, they will unlock the full potential of these tools. This will not only improve operational efficiency and safety but also enhance the customer experience, drive sustainability efforts, and provide utilities with the insights they need to manage an increasingly complex grid.

In the coming years, as AI becomes even more integral to utility operations, those who have invested in high-quality data management practices today will be best positioned to capitalise on its benefits.

From automating grid management to enabling predictive maintenance, the possibilities for AI in utilities are vast—but they all depend on one thing: the quality of the data.

About the Author
David Cottingham has held senior Engineering and Product Management roles on product lines worth tens of millions of dollars, most recently as CPTO at an AIM-listed businesses, and previously at Citrix Systems for over a decade. He’s led large, multidisciplinary engineering teams distributed around the world, and worked on operating systems, SaaS, IoT hardware, developer ecosystems, and mobile applications. He holds a Ph.D. in Computer Science from the University of Cambridge.