Fig. 1. Structure of planar-cross MGTS

Fig. 2. Variation curves of magnetization offset sensitivity of attributes

Fig. 3. MGTS heading-line surveys for pattern recognition training

Fig. 4. SLFNs model of the ELM

Fig. 5. Flow chart of MGTS heading-line surveys for pattern recognition based on SSA-KELM

Fig. 6. Posture characteristics attributes g_xx， g_xy， g_xz， g_yy， g_yz， ΔT， λ₃， I₂， cos θ measurement signal

Fig. 7. Dimensionless waveform feature fractal dimension of the posture characteristics attributes signal

Fig. 8. Target posture characteristics recognition accuracy and feature visualization in PCA dimensionality reduction process

Fig. 9. Target magnetization direction and shape recognition results

Fig. 10. Dimensional waveform features fractal dimension of shape characteristic attributes signal

Fig. 11. Target geometric shape recognition accuracy and feature visualization in PCA dimensionality reduction process

Fig. 12. Target single heading-line surveys training experiment and the magnets to be measured

Fig. 13. Measurement results of the posture characteristic attributes of the magnets three poses on the first route

Fig. 14. Measurement results of the shape characteristic attributes of the magnets three shapes on the first route

Fig. 15. 3D feature distribution of posture and shape characteristic attributes （the original 54D dataset was reduced to 3D by PCA）

Fig. 16. Recognition results of three labels for the pose and shape of the magnets in the experiment