Reducing Fraudulent Activity For Critical Infrastructures
With fraud costing utility companies $80-100 billion USD per year, it’s vital to detect fraudulent consumption behaviour to maintain the grid for paying customers. This missing revenue would usually finance regular maintenance, investments in new assets and services, as well as drive market innovations.
At Cuculus, we’ve developed a way to use smart meter data from the ZONOS™ IoT Platform to identify fraudulent consumption. Our newly published white paper explains how we aim to incorporate fraud detection into our solutions, using ZONOS™ FraudDetect.
Why is fraudulent activity a problem?
As well as costing utility companies a significant amount in lost revenue, the disruption of utilities due to fraud has other serious consequences, including:
- Reduced supply quality
- Safety risks
- Increased load on generating stations
- Potential damage to the supplying grid
All of these factors combined mean it’s essential to identify and stop fraud. Traditionally, this involved on-site inspections which are often time-consuming, expensive, and not always effective.
Now that many utility companies are transitioning to smart meters and the Internet of Things (IoT), there is a new solution. Analysing data from smart meters offers a completely new approach to fraud detection.
Challenges for utility companies
The benefits of analysing data from smart meters fall into two main categories:
- Quantify the losses. It’s difficult to quantify fraudulent consumption, which is often simply categorised as “non-technical losses”. Once you know how many of your customers are fraudulent, it’s much easier to calculate the overall financial loss to a higher degree of accuracy. These insights can then be used to create targets for loss reduction.
- Detect specific fraudulent consumption. Once you know where exactly the fraudulent consumption is coming from, you can take concrete countermeasures to stop this. This could include on-site assessments or the use of specific technical solutions like cables designed to resist fraud. Identifying patterns in fraud can help you decide on the best solution for specific scenarios.
What is Cuculus doing to support utility companies?
We’ve developed a new module for identifying fraudulent consumption behaviour, ZONOS™ FraudDetect. Its goal is to protect critical infrastructure from fraudulent customer behaviour and theft. Traditionally, the analysis of energy data involves mixing large groups of customers, resulting in a blend that doesn’t provide much information about each individual. We’ve developed machine learning models which can accurately distinguish between fraudulent and honest customers. These models provide high accuracy through various testing and believe that support vector machines are the best model for the effective detection of fraud.
ZONOS™ FraudDetect: a solution for critical infrastructure
Using smart meter data from individual customers, it’s possible to precisely identify fraudulent sources. Our research is based on electrical energy grids but can just as easily be applied to any critical infrastructure that uses smart metering.
The pilot study helped in the development of ZONOS™ FraudDetect, a new module designed to protect critical infrastructure from fraudulent customers. You’ll also be able to access detailed reports and interactive dashboards on the ZONOS™ IoT Platform of data analysis.