Smart Water Grid Anomaly Detection via Trend Change Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection methods for smart water grids, such as prediction-classification, clustering algorithms, model calibration, and statistical process control, face challenges like requiring vast historical data, producing false alarms, and being limited to stationary time series data, which hinder near-real-time anomaly detection and classification.
Innovation Solution
A technique that incorporates trend change detection in two phases: establishing historical trend patterns for sensors and using a valid event evaluation time window to classify anomaly events as true or false alarms by analyzing deviations in new sensor data compared to historical patterns, leveraging statistical process control methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If prediction-classification approaches are used to detect anomaly events, then anomaly detection capability is improved, but vast amounts of historical data are required which increases system complexity and data storage requirements
Solution Approach 1:
The patent extracts only the essential trend pattern from historical data rather than using the entire historical dataset. The trend change detection module identifies and stores only the characteristic trend patterns (e.g., typical daily consumption patterns, seasonal variations) that are necessary for anomaly detection, discarding redundant historical data points while maintaining detection effectiveness.
Solution Approach 2:
The system performs preliminary trend pattern establishment before anomaly detection. Historical data is pre-processed to establish baseline trend patterns that capture normal consumption behavior. This preliminary action allows the system to detect anomalies by comparing current data against pre-established trends rather than requiring access to vast amounts of raw historical data during operation.
2Measurement precision
If model calibration approach is used to detect leakages, then detection accuracy is improved, but frequent model calibration consumes processing resources and delays near real-time detection
Solution Approach 1:
The trend change detection module operates autonomously without requiring external model calibration. It automatically establishes trend patterns from historical data and continuously monitors for deviations. The system serves itself by maintaining its own detection capabilities through automatic trend pattern updates rather than requiring periodic manual or automated model recalibration, enabling near real-time detection without resource-intensive calibration processes.
3Reliability
If statistical process control approach is used for anomaly detection, then false alarm rate is reduced, but the system is limited to stationary time series data which reduces adaptability to varying consumption patterns
Solution Approach 1:
The patent implements dynamic trend patterns that can adapt to changing consumption patterns over time. The trend change detection module continuously updates trend patterns to reflect seasonal variations, daily patterns, and long-term changes in water consumption behavior. This dynamic approach allows the system to handle non-stationary time series data while maintaining the false alarm reduction benefits of statistical process control by comparing current data against adaptive rather than fixed baselines.
Data Source
AI summary
Techniques are provided for near real-time anomaly event detection and classification with trend change detection for smart water grid operation management. In the first phase, a trend change is detected in each of one or more sensors by comparing new sensor data of a sensor with a historical trend pattern of the same sensor. In the second phase, and after the trend changes are detected, a valid event evaluation time window can be determined based on combining and analyzing the detected trend changes for flow and pressure sensors, e.g., at least one flow sensor and at least one pressure sensor from the same supply zone of the smart water grid. The valid event evaluation time window can be used with anomaly events that are detected in near-real time to classify the anomaly events in near-real time as valid, e.g., true anomaly events, or invalid, e.g., false alarms.