Utilities & AI: The Time Has Come
image credit: Dreamstime.com
- Nov 29, 2019 4:00 am GMTNov 26, 2019 7:36 pm GMT
- 432 views
This item is part of the Artificial Intelligence - Special Issue - 12/2019, click here for more
By Robert Ansah & Mike Smith
Think of an industry that has been humming along for over a century that has arrived at a point where its traditional needs and goals have run right into a massive re-set as to how the industry operates, engages with customers, and meets a myriad of financial and regulatory goals. This industry, of course, is our beloved utility industry, which in a way has been its own worst enemy in that it has provided for a quality of life that is so good that people take it for granted and only think about it when there is a problem.
The “massive re-set” noted above is the infusion of new energy resources that are changing, first, the operating model and nature of customer engagement, but ultimately all aspects of how the utility exists and functions. It is this re-set, with its new, massive, and complex data sets, that is driving the call for artificial intelligence (“AI”) to create opportunities from these challenges.
While there are many notions of what AI is in the marketplace of ideas, let’s think of AI as being a tool that helps us create experiences that will be smart, intuitive and informed by analytics that are not seen but felt via new business, personal and operational engagement models. Enabling this interaction requires AI applications that can sense, analyze and respond to their environment in an intelligent and interactive manner. And if we can deploy AI without requiring the end user to write, understand or interpret code, all the better.
This might sound abstract but it is becoming commonplace, particularly as machine learning models – a core component of AI - become more pervasive and their application become more core to utility operations. Machine learning is a method of data analysis that enables systems to learn from data, identify patterns and make decisions with minimal human intervention. Add in AI’s abilities to automate these processes and, in effect, sense, and you have some powerful capabilities that will enable utilities to be viable in a changing energy landscape.
And, interestingly, these ideas are not new. Alan Turing, who became a familiar name via the movie “The Imitation Game,” said in the 1940s that “what we want is a machine that can learn from experience.” A more recent popular example is when Google’s artificial intelligence algorithm beat a professional player at the Chinese board game Go, which is considered the world’s most complex board game in 2016.
So, what does this look like in a utility? The examples are many, and we explore a few of these below, but at the core is an influx of new data sources that are typically massive, usually complex, and often mixed with other data sources to help provide insights that were not possible just a few years ago. Think in terms of the influx of sensors and intelligent devices in an IoT environment and all of the data coming off of those devices ranging from energy usage from a smart meter, to indicators like temperature, current and voltage from a myriad of devices and sensors that are now starting to proliferate the grid.
Among the utility examples that we see in the industry is some of the work being done at SoCalGas where their analytics group is leveraging hourly smart meter data from 5.8 million customers and other internal and external data sources to do things like improve collections operations (with its very real and tangible bottom line impact) and consumption load profiling (enabling deeper customer insights for new program targeting like paperless billing and special payment programs). Other utility use cases that are leveraging AI/ML include:
- Using AI image detection and classification algorithms to analyze drone images for damage assessment, such as downed poles and wires, after a storm. The intelligence gathered allows utilities to more quickly review damages and optimize restoration activities. Compared to the traditional truck rolls which are costly and sometimes inhibited because of floodwaters or crew unavailability, AI, and in this case drone technology, is allowing utilities to make quicker and more reliable decisions to restore normal grid operations.
- Free-form text, referred to as unstructured data, has been the least leveraged data source because of the amount of human interaction needed to perform analysis. Using natural language processing (NLP), utilities are now able to mine the massive amounts of unstructured call center data to optimize call routing and use the findings to train call center agents to resolve calls quicker to reduce call center operation costs as well as provide a better customer experience. NLP is a branch of AI concerned with how well computers can understand, interpret and manipulate human language. NLP is also being used to analyze social media and survey data to gauge customer sentiment towards a utility.
Today’s utility leaders are grasping these new opportunities in light of the many challenges ahead. Much like the telecommunications and retails industries, utilities will emerge from this transition with a more customer-focused approach to their business that will ultimately benefit customers, the environment, and the bottom line.