Assess the utility company’s AI maturity curve to determine next steps
Elevate your technology and enterprise data strategy to Transform 2021.
This article was written by Kristin Ellerbe, VP of Technology, PrecisionHawk
While many utilities recognize the transformative power of data, harnessing it can be a different story. You could have millions of data points on everything from the condition of utility poles to hours spent in the field. But all that data is useless if you can’t synthesize it into something useful and actionable.
So how can you unlock the value of data, without spending endless hours and dollars?
Artificial intelligence (AI) has revolutionized everything from devices in our hands and homes to the way we drive, and now it’s even taking over. Drive-thru in fast food restaurants.
With widely distributed physical assets that can be difficult to access or scrutinize on an individual basis, the utilities industry is an optimal use case for AI. By integrating AI into your asset monitoring efforts, you can achieve significant efficiency gains in both your ability to quickly locate and categorize assets and determine their condition, ultimately leading to better business decisions.
But getting into AI can be daunting. Before adopting an AI-powered solution, it helps to understand where you fit on the AI maturity curve.
First steps: display and validate assets
For utilities that haven’t adopted AI, it’s not about using AI, it’s about preparing for it. It starts with how you view and post assets. Too many public services still rely on outdated data collection methods. This often leads to costly re-inspections as the data is subjective and unverifiable. Additionally, the data itself tends to be more traditional – checklists and form data siled across departments, with little value on its own.
Conversely, visual data, including RGB, thermal and LiDAR data, is reliable, provable, traceable and auditable. Whether collected by drones, helicopters, satellites, or ground crews, visual data is where many utilities begin their AI journeys.
Collecting visual data and combining it with traditional data in a centralized location provides utilities with a digital audit system. This offers many benefits, including reducing the time spent on non-contributory actions such as re-inspections and allowing you to consolidate the work that needs to be done, which improves operational efficiency.
More importantly, creating a digital auditing system puts in place utilities to analyze changes over time. With all of your asset data in one place, you can start using the predictive capabilities of AI to compare the past with the present to predict the future.
On the way: assess the condition of assets
Once you’ve implemented a digital auditing system for your asset data, the next step in AI maturity is to use AI to assess asset conditions. This is usually done through a three-step process that involves both humans and machines:
- Identify and sort assets based on their priority. The objective here is to discover the anomalies in the asset images according to the priority. Doing this manually can take a person hundreds of hours. However, by prioritizing issues and feeding them into the machine, humans can apply AI image detection technology to examine a huge library of images and instantly identify the most pressing issues to be found. solve.
- Correct, evaluate and teach the machine. Just because a template was created to accomplish a certain task on day one doesn’t mean it will change over time. As inspectors review anomalies and issue work orders, they will use the system to mark new conditions, modify conditions, and modify severities.
- Let the machine do its job. Training a machine learning algorithm is an evolving lifecycle of continuous improvement. As inspectors change their priorities, they recycle the machine and the machine changes its outputs to accommodate. Mid-term utilities use multiple machine learning algorithms to gain efficiency.
Operationalize AI: Make Better Decisions
While operationalizing AI may still seem conceptual to many, some utilities are doing it today to make better business decisions.
Inventory automation is one of those areas. The inventory is fairly static, so utilities go to great lengths to ensure it is correct at all times taking into account changes in the field. The latest generation utilities use AI to automatically update inventory systems when changes are detected.
This is particularly useful in the event of a disaster. When a hurricane hits, AI can compare post-disaster images with pre-disaster inventory data, producing a delta report that can show response teams the best places to focus their efforts.
Additionally, by connecting this technology to your work order systems, utilities can train the system to automatically create a work order for a fallen pole, for example. And for tradable items, the system can convey these issues to a human to decide if corrective action is needed.
Size matters, in a way
While the maturity of the AI often matches the size of the utility, there are exceptions. The important thing is to know where you are to guide your next steps.
When it comes to whether software or AI services are best for your organization, a good rule of thumb is that utilities early in their AI journey tend to benefit more from advisory AI services, while that late-stage utilities tend to prefer AI software. Take this with a grain of salt, however, as what’s best for your organization always depends on your unique goals, risks, and business model.
VentureBeat’s mission is to be a digital public place for technical decision-makers to learn about transformative technology and conduct transactions. Our site provides essential information on data technologies and strategies to guide you in managing your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the topics that interest you
- our newsletters
- Closed thought leader content and discounted access to our popular events, such as Transform 2021: Learn more
- networking features, and more
Become a member