Vik Chaudhry
Modernizing Power Infrastructure: The Backbone of our Economy
Most people don’t give much thought to electrical infrastructure. While it is the very backbone of our digital age, the viability of the infrastructure that enables so much of what we depend on today is usually assumed and forgotten… at least until the power goes out or an electrically sparked wildfire threatens a community.
The task of understanding the changing condition of our shared electrical infrastructure is difficult work. Transmission towers might look like unstoppable metal robots along the highway, but the vast majority of these steel lattice structures are ~30 years old and most of the hardware was built for a 40-year useful life.

Historically, understanding the condition of electrical infrastructure has been the responsibility of men walking the line. When they’re lucky (and there’s an access road), line workers use bucket trucks. But when electrical structures are in a backyard easement, or on the side of a mountain, or otherwise out of reach for a mechanical lift, today’s line workers still belt-up their tools and start climbing.
This lends to several “What If” questions about both opportunities for innovation and improvements in this space:
What if a line worker could stand at the base of an electrical pole and see all of the important features up-top without having to climb?
What if a line worker could take another step back and understand all of the same information about features on top of the electrical pole, but from the safety of their truck?
What if the line worker could skip both the truck and the drive to the job site and see all of the component level detail on top of the electrical pole without leaving the office?
What if a line worker could understand all of the critical power infrastructure components without ever leaving their home?
With rapidly changing climatic conditions, we are witnessing rapid aging of our power grid infrastructure, therefore requiring more mandates for inspections of this infrastructure to keeping it up and running. Power utilities have been capturing vital data around power lines and grid infrastructure for years now, but with recent advances in technologies such as UAVs, drones, camera sensors, and IoT devices, they are capturing much more data with higher frequencies.

Historically, utilities would send out helicopter crews to capture visual data around the infrastructure. Now, however, with advanced drone and UAV technologies, most of the major power utilities are implementing drone aviation programs to carry out visual data collection operations. In fact, several major utilities are capturing even more than a million images every year around their grid infrastructure. This low-risk method of gathering crucial data for early identification and mitigation of risks in power infrastructure becomes cheaper and faster every day. Typical UAV data collection services for transmission & distribution infrastructure save over 30% in expense and over 50% over time by using drones for analysis and work planning.
With cameras that allow for >30x optical zoom, granular details can be captured from safe distances with the UAV flying 50+ feet away from the object of interest and the pilot/operator 1,500+ feet away from the drone. Similarly, recent advances in UAV-based thermal sensors and aerial LiDAR mean that electrical faults and looming issues can be identified, understood and located in real-time without any line workers ever leaving the ground.
After the data has been collected using aerial and ground vehicle technologies, the problem of managing, processing and analyzing this vast volume of data still remains. When all this visual data comes back to the utility data centers and storages, field technicians, engineers and linemen spend months to analyze it when they could be out in the field performing condition-based maintenance. On an average, analyzing vast volumes of data can take 6-8 months of time per inspection cycle during which time frame the in-field data becomes outdated and a power line can go down or a wildfire can start.
It’s time for AI to step in.

There has been a lot of buzz around the words AI and Machine Learning in the past decade. So, what is AI? In order to understand AI, first we need to understand the computer. Computer is something that can “compute”. Computers carry out calculations and perform operations as per our instructions. Humans can do all these things, but computers do it faster. A computer can do more than 2-3 billion transactions per second while humans can hardly do one. Therefore, computers are very good at calculating, storing and sorting through data, and finding relevant information among huge amounts of data. But there are still some tasks that the computer cannot do.
Computers are great for following our instructions to achieve a task. But what if a problem requires making decisions, complex thinking, or intelligence? This is where AI comes into play. You do not have to tell AI what to do. Instead, you provide many examples of what is right and what is wrong, and the machine becomes capable of learning from it, just like a human.
AI and Machine Learning has transformed our world in the last few decades. AI already plays a critical role in most businesses, industries, and even societies.
In the last decade, Artificial Intelligence and Machine Learning technologies have disrupted and revolutionized both industry and society. In fact, the global AI software market is expected to grow over 150% in the next year alone, to approximately $23 Billion USD.
Today, power utilities are looking for technologies to help accelerate their infrastructure inspections. With aging of power infrastructure, these utilities are mandated to perform frequent inspections which results in collection of massive amounts of data from the field. We have seen AI successfully use massive amounts of data to learn and make efficient decisions.
Hence, AI can play a vital role in this process. AI and Machine Learning technologies are already being deployed to detect faults and breakages in power lines.
Computer Vision is a specific field in AI that relates to machines that learn and understand from visual data such as images and videos. It basically provides the capability for the machine to understand what it is seeing through its eyes (cameras). Let’s take an example of teaching an infant what an apple looks like. We show the infant various examples of apples over and over until the child is able to learn specific characteristics for an apple. Characteristics, such as color, shape, smell, and taste of an apple are registered into the mind of the child. Therefore, the next time the child sees an apple, the child is able to recognize that the object is an apple and not an orange. Computer vision models work on similar principles thus providing capabilities of recognition, detection, classification and understanding various objects in images shown to them. Computer vision technologies are being heavily used in the self-driving automotive industry.
Neural Networks are a framework in Machine Learning that has been widely adopted in the recent years. This framework mimics the functioning of neurons in our brains. The basic unit for this network, just like a human brain, is a neuron. Information enters the neuron; the neuron performs some mathematical operations on it and then outputs it to another neuron. All this happens thousands of times within microseconds. In the end, just like a human brain learns various things such as singing, dancing, sketching, etc., using this neural network framework, the machine also learns to perform specific tasks. One class of neural networks is called Convolutional Neural Networks, that is particularly great at working with visual data. We use Convolutional Neural Networks for laying the foundations of our AI algorithms.
Through the use of Convolutional Neural Networks, showing the machine large amounts of right and wrong images of electrical components, the machine is now trained and understands the differences between various electrical components as well as recognizing the ones that are faulty.
Our AI can now process through tens of thousands of images to detect faults and issues in a matter of hours and days, compared to the current method which takes months taken by utilities to manually analyze these images (in which time a line can go down or a wildfire can spark). Not only does our AI save power utilities months of time to identify areas where faults exist, but also it saves them on an average 50% of manual analysis costs. Another area where AI and machines can greatly help humans.
All this information and results generated from our AI systems needs to be easy to comprehend. We bring in the power of software into the picture where all these results are displayed on our software platform, PowerAI, for utilities to take actions based on them. For instance, the geographic regions where our AI is highlighting a lot of issues, is where utilities can schedule repair or replace based maintenance. Again, our aim is to protect the energy infrastructure and keep it maintained and running so that our society has electricity and wildfires do not start or spread due to power lines.
Where do we go from here?
The key to modernizing our aging energy infrastructure lies in leveraging innovative technologies such as UAVs, software, AI, and analytics to map our physical infrastructure to the digital world. This digital transformation within our utilities sector will not only help make more informed, data-driven decisions, but also effectively forecast and predict the future health of our infrastructure.
Leveraging the software and AI/ML based approaches will create even greater systemic opportunities. Every day the power sector becomes less reactive and more proactive towards both chronic and acute problems on the grid. Here’s to the future of technology and to enabling predictive support of our electrical infrastructure. From data collection, to analysis and reporting, ensuring a safe and reliable power grid is indeed critically important, even when we do not think about it.
- Vik Chaudhry Co-Founder, COO and CTO Buzz Solutions
Enjoyed reading this article? Share this with your colleagues.