June EnergyIncrease the quality of the
energy consumption data
in an automated way

To keep track of their customers’ energy consumption, June Energy provides ‘smart cameras’ which are installed on top of electricity and gas meters, converting them into ‘smart connected meters’. These cameras capture analog meter readings and send them to the June Energy cloud for processing. Those images are then converted into digits, representing energy consumption data and thus providing insight into energy usage. However, the image quality might be low, making the image unreadable. There are multiple reasons for low quality images:

• Dust in front of the camera.
• Too little or too much light influx.
• Misalignment of the camera frame.
• Spiders in front of the camera.

The low quality of the images had a significantly impact on the quality of the energy consumption data, causing the need for manual corrections afterwards.

Sentigrate was consulted to increase the quality of the energy consumption data in an automated way, with the goal of reducing human interventions.

The solution Sentigrate worked out consisted of 3 steps:

1. Predicting the energy consumption at the moment an image is taken. The prediction takes into account the customer’s usage pattern, historical energy consumption data and the image itself.
2. Comparing the prediction to the raw conversion of the image into digits. Therefore, a custom distance metric (based on the Levenshtein distance) was developed.
3. Based on the difference found in step two, the energy consumption data is corrected to be closer to the predicted value of step one.