Magnetic Anomaly Evaluation for Air Cargo Employing Array TMR Sensors and Deep Learning Algorithm

The swift advancement of e-commerce has led to an increased transit of magnetic items via air freight, which may jeopardize airplane safety. It is essential to detect and assess the magnetic anomalies for maintaining flight safety. However, the industry still lacks online detection equipment for magnetic anomaly measurement. This article presents an automated magnetic anomaly detection system that employs array tunneling magnetoresistance (TMR) sensors and a deep learning calculation algorithm. The system has four sensor arrays that are located on the four sides of a cargo conveyor belt to continuously monitor the magnetic field. The magnetic abnormalities are detected and quantified as the cargo passes through the sensor arrays. A deep learning algorithm is developed to ascertain the position and magnetic moment of magnetic sources, enabling a quantitative evaluation of the risk associated with magnetic abnormalities. A prototype system including 64 sensor modules has been developed and tested on an airport cargo conveyor belt to evaluate the practicality of the technology. Experimental validation on airport cargo belts shows that, for single-source cases, the system attains a position RMSE of 3.22 cm and a dipole-angle RMSE of 1.07°. In double-source scenarios, the corresponding errors are 13.18 cm and 25.07°, confirming reliable performance across both simple and complex magnetic configurations. This automated technology significantly improves the efficiency and reliability of magnetic anomaly detection in air transportation operations compared to the traditional method of using a handheld magnetometer.