Research on CNN-BiLSTM-Attention-Based Geomagnetic Positioning Technology

Geomagnetic positioning is a technology that utilizes the Earth's magnetic field for location determination. It offers advantages such as autonomy, no accumulation of positional errors, absence of signal radiation, and strong concealment. However, its accuracy is influenced by factors such as the adaptability of the geomagnetic matching region and the performance of geomagnetic matching algorithms. To address the problems of poor accuracy and low convergence rate in existing geomagnetic positioning algorithms, a geomagnetic positioning algorithm based on CNN-BiLSTM-Attention is proposed. This algorithm begins by utilizing Convolutional Neural Network (CNN) to process geomagnetic sequences and extract spatial features from the geomagnetic signals. Subsequently, a Bidirectional Long Short-Term Memory (BiLSTM) network is introduced to fully learn the temporal dependency features of the geomagnetic sequence. By employing forward and backward propagation, BiLSTM effectively captures the complex temporal dynamics within geomagnetic signals, enabling a more comprehensive understanding of the variations in geomagnetic sequences. Furthermore, the model incorporates an attention mechanism to dynamically assign importance weights to different time steps, allowing the model to focus on time-step features that contain more significant information. The geomagnetic positioning algorithm based on CNN-BiLSTM-Attention is simulated in three different regions and compared with traditional algorithms as well as the latest algorithms. The experimental results show that the geomagnetic positioning algorithm based on CNN-BiLSTM-Attention achieves a minimum average positioning error of 24.91m. This represents a 32.24% improvement over the CNN-BiLSTM algorithm and a 44.64% improvement over the Gray Wolf Optimization(GWO) algorithm, demonstrating superior geomagnetic positioning accuracy.