Whether it’s the drive to decarbonisation, increasing decentralisation, or democratisation and digitalisation, the forces disrupting the utility industry are plain to see. Utilities must find ways to ensure they transform their businesses to step up to the challenge, drive efficiency and attract the latest industry talent.
Artificial intelligence (AI) has the power to change how the utility industry addresses these challenges – complex machine-learning algorithms can be implemented to support use cases from energy management to job site safety. But unless utility companies embrace and capitalise on this capability, identify areas to provide early value, and deploy early stage pilots to learn and evidence value, they will not keep pace with the disruption taking place in their industry.
When it comes to utilities, AI is attributed to computers or computer-controlled machines that, thanks to incredible processing speed and memory capacity, can reason, generalise and perform selected intellectual processes. In other words, AI makes it possible for intelligent devices to process and perform tasks commonly associated with humans – only faster and with more consistent quality.
There are a plethora of areas where AI can be used, and utility companies have much to gain by leveraging this massive computational power and exceptional speed. Here are three examples of where utilities should look to innovate.
Image recognition and drone technology
Image recognition technology is commonly used for everything from identifying faces in photos to detecting vehicles in the vicinity of driverless cars. Many utilities can take advantage of similar pattern recognition technologies in different environments and applications. Linking this new technology with the capabilities of AI will be extremely powerful.
Consider, for example, how helpful image recognition technology might be when assessing and responding to severe flood damage. If utilities were able to use camera-enabled drones to conduct aerial surveys of service areas prior to flooding, they could then later send the drones back into the skies to survey and assess the damage. By comparing images before and after, AI could identify everything from missing poles to chemical spills.
As part of a rapid response effort, instead of sending crews to investigate damage, AI could be enlisted to solve the logistics of the restoration by estimating needs for replacement poles and conductors, locating replacement equipment in storage or from supply chain, and prioritising repairs.
Looking forward, utilities should also consider ways to use drones to perform geographical surveys and use the resulting data sets to provide utility planners with better decision-making tools. We are already seeing drones used in maintenance. A newly announced project will see an autonomous vessel transport a fleet of self-piloting drones to repair offshore wind farms, for example.
Weather prediction and battery storage
Weather affects every part of the utility sector, whether its understanding wear on assets over their lifetime or better predicting and dispatching renewables.
The cost of solar panels and wind generation continues to decline, but the unpredictability of renewables too often constrains deployment. However, high resolution, high accuracy and hyper-local weather prediction capabilities may enable utilities to solve reinforcement challenges through flexible marketplaces.
If a utility knows that substantial cloud cover will be forming over a solar array on a hot summer day, for example, it could tap into battery storage or an overproducing wind farm to fill the gap – using machine learning to determine optimal system efficiency and network configuration in near real time. This would help utilities remove much of the uncertainty that surrounds the shift from fossil fuels to renewables, while creating a dynamic, sustainable energy supply.
Sensors and smart technologies
The decreasing size and increasing power of sensing technology provides more data for AI algorithms to process and draw insights. Sensing AI technologies hold great promise for a variety of utility applications as well. The safety benefits of outfitting manholes and other confined spaces with sensors that feed data to an AI device, testing air and water quality, could be significant. The sector now needs to work on extracting the value from the data by making it actionable.
Sensors can also alert utility maintenance workers to other potentially dangerous conditions. Before entering an ageing manhole, workers could use laser radar to create an image point cloud that illustrates cracks or water seepage. Sensors supported by machine vision or learning sensors can in addition serve as the foundation for early warning systems. With sensors shrinking in size, “smart” sensor-enabled paints can be developed to detect rising levels of carbon monoxide, gas seepage or seismic vibrations.
It may be tempting for utilities to deploy a wait-and-see approach to AI and machine learning, anticipating that others will do the heavy lifting. But we must remember that these are disruptive, industry-changing technologies that are readily available at economical prices – and are being deployed across other industries every day.
By combining the strengths of human insight and intuition with the raw processing power of computers, AI and machine learning have already demonstrated the ability to save time and money while improving safety and service. The key for utilities is to identify low-cost, low-risk opportunities to enter this daunting world. We should think big, start small, and move fast.