Smart Monitoring for Power Theft using WinML and OpenVino
- 0 Collaborators
Power Theft is a major problem faced by power utilities today. In India, this accounts for roughly 50% of the total power loss every year. Power Theft can be caused by infidelity at the consumer end, unethical tapping of transmission lines and hacking or tampering of energy meters. In this work, we use LSTMs to shortlist customers who are stealing power. We use OpenVino and WinML to see people who are tapping power lines or have tampered meters and then use LSTMs to check whether power is actually being stolen or not. ...learn more
Project status: Under Development
RealSense™, Internet of Things, Artificial Intelligence
Intel Technologies
OpenVINO
Overview / Usage
In India, Power Theft accounts for nearly $4.5 Billion of losses every year. In the UK and US, this loss ranges from $1-6 Billion. This is a big issue for power utilities.
Unfortunately, it is very difficult to figure out who is stealing power. More importantly, thieves are now adapting to steal power such that it is difficult to spot by analysing power data.
This project has two parts: In the first part, users can take a picture of a person they think is stealing power or a tampered smart meter. Using Convolutional Neural Networks, we can then figure out whether someone is stealing power there or not. Secondly, using this information, we can run an LSTM on smart meters and substations near the location where the picture was taken and see whether someone was stealing power in that area or not.
This process will significantly reduce the number of false claims of power theft that power utilities have to investigate. It will also allow citizens to take an active role in reducing power theft.
Methodology / Approach
For the first part, using an Intel Real Sense camera, we can capture images. These images will then be fed into a convolutional neural network and using OpenVino we can predict whether someone was stealing power or tampering with a smart meter.
Then using WinML and LSTM networks, we can predict whether someone was stealing power in that area or not. The algorithm used for doing this is shown in this research paper that I co-authored here: https://ieeexplore.ieee.org/document/7977665
Technologies Used
Intel Real Sense D415
WinML
OpenVino
Python3