Testing water data for detecting leakages and burst pipelines in buildings. Testing and data analysis is based on this paper: Leakage Detection Using Smart Water System: Combination of Water Balance and Automated Minimum Night Flow
Methodology:
- Training data are retrieved from SparkWorks.
- Water Volume measured during morning hours is calculated (water consumption between 0200-0500) called Minimum Night Volume.
- A moving average of the last 15 day's Minimum Night Volume is calculated.
- Iterating on the data in chronological order we compare the MNV with the current day's NV to detect if it exceeds a predefined threshold.
- If the threshold is exceeded the value is considered an abnormal value, and the day is reported as a possible leakage/pipe burst.
Legend:
- mnv: moveing average of the Minimum Night Volume
- limit: the threshold to consider a value as a Leakage
- nv: the current day's NV
- full data in SparksDrive/Projects/LORA Pilot/Night Flows
Examples:
- 0
- 2
- 10
- 12
- 14
- 25
