Leakage Detection in Water Supply Networks Using Robust Regression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for leakage detection in water supply networks are inefficient, leading to high false alarm rates and excessive costs due to their inability to continuously monitor and accurately locate leaks, especially in large networks.
Innovation Solution
The method involves dividing the supply network into areas with comparable consumption profiles, using robust regression to determine a baseline inflow curve and confidence region, measuring minimal inflow, and generating time series to identify leaks outside this region, allowing for automated detection and localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional night flow analysis is used to detect leakages, then leakage detection is possible, but false alarms occur very frequently leading to high operational costs
Solution Approach 1:
The supply network is divided into multiple areas with comparable consumption profiles. By segmenting the network and comparing relative changes across areas rather than using absolute threshold values, the system reduces false alarms while maintaining reliable leakage detection, thereby lowering operational costs associated with false positives.
Solution Approach 2:
The invention changes the detection parameter from absolute consumption thresholds to relative changes in consumption patterns across segmented areas. This parameter transformation allows the system to distinguish actual leakages from normal consumption variations, improving detection reliability and reducing false alarm-related operational costs.
2Measurement precision
If step tests are performed for leakage localization, then leakage location can be determined, but considerable expense is incurred due to household notifications and replacement supply
Solution Approach 1:
The invention replaces the mechanical/physical step test method (which requires physical disconnection of regions and replacement supply) with a computational approach using regression analysis and time series comparison. This substitution maintains leakage location accuracy while eliminating the high implementation costs associated with customer notifications and replacement supply arrangements.
Solution Approach 2:
Instead of physically disconnecting regions for step tests, the system creates virtual models of expected consumption patterns using regression curves. By comparing actual measurements against these copied/modeled patterns, the system achieves leakage localization without the expensive physical interventions required by conventional step tests.
3Reliability
If noise meters are used for local leakage monitoring, then leakage detection is possible, but measurements can only be performed by specialists on site at times of low consumption
Solution Approach 1:
The invention creates a universal monitoring system that processes data from multiple areas simultaneously using standardized regression analysis. This multi-functional approach replaces specialized on-site noise measurements with a centralized system that can continuously analyze consumption patterns across the entire network, eliminating the need for specialist intervention and enabling operation at any time.
Solution Approach 2:
The system enables self-service leakage detection by using automated regression analysis and time series comparison. Instead of requiring specialists to perform manual noise measurements, the system automatically processes consumption data, generates regression curves, and identifies leakages through computational methods, making the process independent of expert intervention.
4Ease of operation
If conventional methods are used for leakage detection, then simple measurement is possible, but continuous monitoring is not suitable and high false alarm rates occur
Solution Approach 1:
The invention enables continuous monitoring by continuously generating regression curves from historical data and comparing real-time measurements against these curves. This continuous application of the regression method maintains measurement simplicity while achieving sustained monitoring capability, unlike conventional threshold-based methods that generate high false alarms and cannot operate continuously effectively.
Data Source
AI summary
A method and device are provided for leakage detection and leakage location in an area of a supply network (e.g. water supply, gas supply or district heating network), wherein measurement values of sensors of the supply network are statistically analyzed for the presence of leakages using robust regression methods. The false alarm rate (type 2 error) may be reduced or minimized.