This invention describes a system that detects water leaks in a property by analyzing water consumption data from a connected smart water meter. It works by identifying the unique water usage patterns, or "signatures," of individual appliances like washing machines or dishwashers. By comparing these expected patterns with the actual overall water consumption, the system can determine if a leak is occurring, what type it is, and then trigger an action to prevent further damage.
Why it matters: Filed in 2020, the core challenge of identifying specific appliance signatures from aggregated data has become significantly easier. Advances in machine learning and AI for time-series analysis and anomaly detection since then allow for more robust and automated pattern recognition, directly improving the accuracy of leak detection based on expected versus actual consumption.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI