GroupCore Acoustic AI-Based Urban Water Supply
Network Leak Monitoring Solution

I. Solution Background and Industry Challenges

China’s per capita water resource availability is only one quarter of the world average, and of the more than 600 cities nationwide, over 400 suffer from insufficient water supply. At the same time, the pipeline network leakage problem is extremely severe—according to statistics from the Ministry of Housing and Urban-Rural Development, national water supply losses reached 8.16 billion cubic meters in 2019, with direct economic losses exceeding RMB 20 billion, equivalent to the annual domestic water consumption of 120 million people. Most cities still rely on traditional manual leak-listening detection; in the face of increasingly large and complex pipeline networks, they suffer from bottlenecks such as slow response, narrow coverage, frequent misjudgment, and data that is difficult to accumulate.

At the policy level, the “14th Five-Year Plan” Water-Saving Society Construction Plan explicitly requires that by 2025 the leakage rate of urban public water supply networks be controlled within 9%. In 2022, the National Development and Reform Commission and the Ministry of Housing and Urban-Rural Development jointly issued a document setting a higher target of ≤7%–8% for pilot cities and providing central budget subsidies of up to RMB 200 million. In 2024, the ultra-long special treasury bonds also listed “public water supply network leakage control” as a key support area. Leakage control in water supply networks has risen from industry self-discipline to a national strategic action.

II. Solution Overview

The GroupCore Acoustic AI-based urban water supply network leak monitoring solution integrates high-sensitivity acoustic sensing, IoT wireless transmission, cloud-based acoustic signature big data, and deep learning algorithms to deliver all-weather online monitoring, second-level early warning, and remote assisted confirmation of water supply network leaks. Centered on low-cost, easily deployable, highly reliable hardware terminals, and driven by a domestically large-scale leak acoustic database and proprietary feature extraction algorithms, the solution provides water utilities with intelligent leakage management capability that shifts from “passive response” to “proactive prevention.”

III. System Architecture

1. Perception Layer — Magnetic-Mount Noise Recorder

• Employs a high-sensitivity piezoelectric ceramic sensor, with a sampling rate of 4/8 kHz that can be configured remotely;

• Installed by strong magnetic attachment on pipe walls, valves, or flanges; metal clamps can be fitted for non-metallic pipes; no excavation, no water supply shutdown, and no wiring required;

• Self-powered (lithium thionyl chloride battery, capacity 19,000 mAh) with a low-power design, achieving a battery life of up to 5 years;

• IP68 protection rating, suited to harsh environments such as damp underground chambers and high/low temperature extremes;

• Built-in edge computing unit performing preliminary feature determination directly at the probe, reducing power consumption and cloud load.

  
  

2. Network Layer — Multi-Mode Wireless Transmission

• Supports 4G, NB-IoT, LoRa and other communication methods, using existing public networks or self-built base stations;

• Full-process automation of data acquisition, transmission, storage, and analysis, completing the alarm closed loop within 30 seconds;

• The underground communication success rate has been verified in practice to perform excellently, ensuring stable data backhaul.

3. Platform Layer — AI Acoustic Big Data Cloud Platform

• Receives monitoring data from the entire network and performs deep learning comparison against a database of over 23 million sets of leak acoustic signatures;

• A proprietary wide-spectrum colored noise feature quantification model effectively filters out environmental interference such as traffic and machinery, delivering a relatively high accuracy and a low false-alarm rate;

• Provides auxiliary tools including remote recording playback, remote cross-correlation localization, and a sound intensity–frequency bandwidth dual-parameter criterion, supporting manual secondary confirmation;

• Modular GIS map display supporting pipeline information entry, device status viewing, alarm lists, and single-point data analysis;

• Open data interfaces for seamless integration with existing third-party monitoring platforms such as GIS/SCADA.

4. Application Side — PC Management Backend + Mobile APP

• Multi-dimensional alarm lists and map localization, supporting risk grading by pipe diameter, noise intensity, leak duration, and other factors;

• Remote listening, spectrum display, and historical trend queries, facilitating rapid assessment by O&M personnel;

• On-site deployment, parameter configuration, and device inspection can all be completed via the APP, improving frontline operational efficiency.

IV. Core Technical Advantages

V. Typical Application Workflow

1. Deployment: install recorders at accessible points on the target network such as valves, flanges, and fire hydrants (spacing 50–300 m), and enter GIS information.

2. Nighttime monitoring: the system automatically acquires acoustic signals synchronously between 2:00 and 4:00 a.m. (a period of low background noise).

3. AI assessment: the cloud platform performs deep learning comparison on feature quantities such as sound intensity, frequency bandwidth, and spectral distribution, automatically determining suspected leaks.

4. Alarm push: an alarm is generated within 30 seconds and pushed to O&M personnel via PC/APP, with the pipeline segment location marked.

5. Remote review: O&M personnel remotely play back the noise and review the correlation analysis results to confirm whether the leak is genuine.

6. On-site precise localization: for confirmed leak points, leak-listening technicians are dispatched to perform meter-level precise localization within a hundred-meter range (accuracy ±2 m), reducing line patrol workload.

7. Work-order closed loop: after the repair is completed, the system records the disposal result and continuously tracks changes in leakage.

VI. Core Value

• Reduce total leakage: greatly shorten leak discovery time from the traditional days or weeks to the minute level, effectively reducing water loss.

• Optimize O&M efficiency: rule out areas without leaks, rank leak points by risk level, prioritize urgent leaks, and avoid blind line patrol.

• Reduce secondary disasters: timely warning of derivative risks such as road collapse, loss of fire-fighting water pressure, and water quality contamination.

• Empower smart water management: long-term archiving of monitoring data to assist pipeline health assessment and provide data support for renovation and upgrading.

• Support policy compliance: help water utilities meet national leakage rate assessment requirements and secure central budget and ultra-long special treasury bond funding support.

VII. Applicable Scenarios

• Urban public water supply networks (metallic / non-metallic pipelines)

• Water supply systems within large parks and industrial and mining enterprises

• District heating networks (leak monitoring during the heating season)

• Water supply networks in rural areas and peri-urban areas

• Overall smart water management upgrade projects

VIII. Success Cases

• A certain water group: deployment across the pipeline network of the core urban area, achieving intelligent leakage monitoring and effectively reducing the intensity of manual line patrol.

• A certain heating company: leak monitoring of the district heating network, safeguarding the safety of winter heating supply.

• A certain reclaimed water company: leak monitoring of the reclaimed water network, improving the utilization efficiency of unconventional water resources.