AI-Driven Automated Meter Reading for Improved Efficiency
Author: Zhuo Wu, AI Software Evangelist
Maintenance of industrial plants and infrastructures depend on many manual, very time-intensive, and often risky procedures. Take industrial meter reading for example, a process where utilities and businesses alike collect data to measure the amount of water or electricity used by a customer so they can accurately charge them. Meters can sometimes be on top of powerline poles or attached to tanks storing hazardous substances, making it dangerous to collect results. If an operator makes a simple mistake, it can result in a costly miscalculation or a serious accident. On top of it, the task can be very labor-intensive and error-prone. And that’s not even accounting for the lack of manpower willing to perform these regular meter readings.
Thankfully, it doesn’t have to be that way. With recent advancements in machine learning (ML), AI, and computer vision, we can not only develop a solution that keeps operators out of harm’s way but also provide greater value for all businesses in ways not possible until very few years ago.
The Solution: Automated Meter Reading
Up until recently, only a few companies and developers had the knowledge, skills, and processing power to tackle AI-based challenges. For example, existing infrastructure is very expensive to replace decades-old analog or custom-made devices, and even if a business can afford to do so, it often can’t without avoiding costly disruptions of service.
Luckily, things have changed a lot in the past few years.
Today, hardware is much more affordable, and AI software has become more accessible, to use that very little remaining distinction, if any, between “AI developers” and “regular” software developers. The practical consequence, for every business that uses meters, is that today it is much easier and very cost-effective to both read meters automatically and process all the resulting data in the most effective way.
By deploying AI-based automated meter reading, managers can receive in-the-moment information on how to best maximize the performance of their equipment, or which parts should be replaced before they break to reduce downtime and repair costs. To be able to obtain all that information, quickly spot transient anomalies, and make reliable statistical analyses in real time would be an impossible feat for even the best of workers.
Today, ML-enabled maintenance operations (MLOps) can use AI techniques both to “read” large quantities of meters automatically, and to analyze the results directly in the facilities that contain said meters, or very close to them. Besides costs, AI-powered automatic metering can greatly reduce the time needed to refurbish an old plant or adapt to new business requirements. Additionally, adopting automatic metering automatically preserves much more evidence of operations that may be useful for legal or insurance reasons.
Finally, and maybe the best part of AI-based metering, is that it does not force companies to replace field-proven devices that may have many more years of life left.
Companies can get real-time monitoring, predictive maintenance, and all the other benefits of AI-based metering by simply placing in front of their existing meters cameras that send the pictures to AI image recognition software. This is a much more flexible solution than it appears at first sight.
To begin with, any ordinary wall-mounted cameras may handle several meters if their positions allow it, by focusing automatically on each of them at regular intervals. Additionally, if technicians must be on-site anyway, they also have the option of just taking the pictures themselves with a smartphone, avoiding all the risks and slowness of manual data entry. Drones and robots may do the same, both regularly, or whenever some disaster makes urgent inspections mandatory, but also impossible or too dangerous for humans to do.
How to Build an Automated Meter Reading Solution
Using Intel hardware and AI software products like the OpenVINO™ toolkit, it is possible to create fully integrated solutions for AI-based automatic metering directly on edge devices and for all scenarios, including rugged environments or very low power consumption without relying on heavy infrastructure.
More specifically, with OpenVINO — which just celebrated its five-year anniversary with the release of OpenVINO 2023.0 — developers get the tools, resources, and capabilities to build, study, implement, and optimize AI software applications much more easily on various hardware and for various business solutions.
Developers can leverage Intel’s edge AI reference kits, enabling them to easily access code from various solutions, such as the industrial meter reading application. By following a few simple steps, they can swiftly establish the entire environment required for installation. Moreover, this resource furnishes them with the capability to apply models to their own images, enabling rapid visualization of tangible outcomes from real-world images. For developers tasked with adapting code and configurations for diverse meter types, the industrial meter reading kit offers specialized functions that facilitate the customization of their applications.
For example, they have access to Intel’s edge AI reference kits, where they can download code from the provided solutions like the industrial meter reading one and, with just a few steps, set up the overall environment to install everything. It also provides them with a way to try out models on their own images so they can see real results from real images very quickly. For developers who need to adjust code and configurations for different types of meters, the industrial meter reading kit provides functions that helps them customize their solution.
OpenVINO™ for Intelligent Edge Applications
In fact, a lot of the algorithms and AI models you can develop using the AI reference kits can be customizable to fit other business use cases. The object detection algorithms needed to perform automated meter reading can be directly reusable in other scenarios like quality control of assembly lines or anomaly detection in manufacturing products.
The other great thing about OpenVINO is that it’s an open solution, so all these resources are free for developers to use, and you can join our GitHub discussion if you get stuck or have any questions.
Whether you are considering an automated meter reading solution to improve accuracy, reduce costs, and increase efficiency, or looking to apply AI to other parts of your business, please check out all our edge AI reference kits and see how else you can transform your business’ operations!
Notices & Disclaimers
Intel technologies may require enabled hardware, software, or service activation.
No product or component can be absolutely secure.
Your costs and results may vary.
Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy.
Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.
No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.
