Low-Frequency Environmental Magnetic Noise Elimination Based on a Neural Network Algorithm for TMR Sensor Arrays

Tunneling magnetoresistance (TMR) sensors have shown the capability of operating in weak magnetic fields. However, the environmental magnetic noise limits their applications in open field detection. This article proposes a novel background noise cancellation method based on a backpropagation (BP) neural network for TMR sensor arrays. According to simulation results, the BP-based noise reduction method can eliminate background noise more effectively than the traditional coherence coefficient method. The signal-to-noise ratio (SNR) of the sensor can, thus, be improved by over 20 dB, especially when detecting extremely low SNR signals. This algorithm is demonstrated using a TMR sensor array, which shows a capability of greatly enhancing the sensor array’s limit of detection in open field testing.