Longitudinal energy waste detection with visualization

Abstract

Leaky windows and doors, open refrigerators, unattended appliances, left-on lights, and other sources subtly leak energy accounting for a large portion of waste. Formal energy audits are expensive and time consuming and do not capture many sources of leakage and waste. In this short paper, we present a hybrid IR/RGB imaging system for an end-user to deploy to perform longitudinal detection of energy waste. The system uses a low resolution, 16 × 4 IR camera and a low cost digital camera mounted on a steerable platform to automatically scan a room, periodically taking low resolution IR and RGB images. The system uses image stitching to create an IR/RGB hybrid panoramic image and segmentation to determine temperature extrema in the scanned room. Finally, this data is combined with thermostat set-point information to highlight hot-spots or cold-spots which likely indicate energy leakage or wastage. The system obviates the need for expensive, time-consuming waste detection methods, for professional setup, and for more intrusive instrumentation of the home.

Detecting Common Insulation Problems in Built Environments using Thermal Images

“…For example, using a thermal camera that cost upward of $ 20000 ( [18], [19]) will incur far higher startup cost and than saving power consumption. Motivated by these challenges we proposed IRLeak [20] in our previous research, a low cost low power thermal imaging unit that is capable of continuously capturing thermal data and conducted some preliminary analysis to detect energy leakage in homes. Due to the extreme low resolution of thermal camera (16 × 4), the data collection becomes particularly challenging.…”

“…Longitudinal thermal image acquisition with smartphone is a big challenges due to deployment, privacy and battery charge issues. Therefore, we propose to construct a new thermal imaging edge device that improves our prior IRLeak [20] unit where we propose to use the FlIR devkit that consists of an 80 × 60 thermal camera controlled by a Raspberry Pi Zero. In the previous unit we constructed the unit using an MLX90621 IR camera unit that has a very low resolution of 16 × 4 thermal imaging pixels.…”