Core Technology
NMD Non-Contact Magnetic Memory Detection — Pioneered in China, Leading Globally
Technical Principle
Based on the magnetic memory detection method first proposed by Russian scholar Professor Dubov at the 1997 American International Welding Conference. When ferromagnetic components are subjected to external loads, under geomagnetic field excitation, irreversible reorientation of magnetic domain structures occurs in stress and deformation concentration zones. The magnetic state on the metal surface memorizes the location of defects or stress concentrations. By detecting pipeline magnetic field data from ground level, the stress distribution of pipelines can be analyzed.
Technical Advantages
Speed Leadership
Real-time capture and analysis of magnetic signal anomalies during inspection, results in one day — 100x efficiency improvement
AI Intelligence
Multimodal AI automated analysis, automatic anomaly determination, eliminating extensive manual analysis
High Precision
Detection rate improved from 80% to over 90%, world's largest magnetic force dataset powering first-ever AI fitting model
Low Cost
Non-excavation detection, no impact on pipeline operations, significantly lower cost than traditional excavation methods
Fully Domestic
BeiDou positioning + domestic weak magnetic sensors + AI technology, completely autonomous and controllable
Data Security
All data processed locally, no overseas transmission, ensuring national energy data security
Technology Comparison
| Comparison | NMD Magnetic | UK SCT | Ultrasonic | MFL Internal |
|---|---|---|---|---|
| Excavation Required | No | No | Yes | No (Internal) |
| Analysis Speed | 1 Day | 3 Months | On-site | Weeks |
| AI Analysis | ✓ Automated | ✗ Manual | ✗ Manual | ✗ Manual |
| Domestic | ✓ Fully | ✗ UK | Partial | Partial |
| Data Security | ✓ Local | ✗ Sent to UK | ✓ Local | ✓ Local |
| Detection Rate | ≥90% | ~80% | High | High |
| Impacts Operations | No | No | Yes | Yes |
AI Intelligent Analysis System
Proprietary multimodal AI model integrating structured data models, large language models, and image models. The system processes images, structured data, and linguistic data to automatically output anomaly information. Built on the world's largest magnetic force dataset to create the first-ever magnetic signal-to-stress signal fitting AI model.