Research On Magnetic Target Localization And Identification Method Based On Magnetic Anomaly

As an important detection method,magnetic anomaly detection has been widely used in magnetic target detection.Compared with the total geomagnetic field and vector field,the magnetic gradient tensor data contains richer magnetic target information and has important advantages in the application of magnetic target positioning and recognition.With the continuous improvement of the technical level of magnetic sensors,the fine measurement of magnetic targets has become a research hotspot.It has important practical application value for detecting the position and shape of small ferromagnetic objects such as underground ferromagnetic pipelines,cable lines,and unexploded ordnance.Therefore,improving the positioning and recognition accuracy of magnetic targets in small spaces has become one of the important tasks in the development of magnetic anomaly detection.At present,the positioning of multiple magnetic targets in a small space often has problems of ambiguity and solution instability,and the boundary recognition method of magnetic targets has problems such as inaccuracy and poor robustness.This dissertation focuses on the magnetic target location and recognition methods.Firstly,aiming at the problems of environmental noise interference and sparse measuring points in the data measured by magnetic gradient tensor system,the moving average filter,and optimized Kriging interpolation method are used to preprocess the measured data to improve the quality of the measured data.For the positioning of magnetic targets,the Euler deconvolution method is extended to the compound Euler formula,combined with the K-means clustering algorithm to solve the problem of low positioning accuracy of multi-magnetic targets in a small space.The influence of the depth difference between the magnetic sources and the distance between the measuring points on the positioning accuracy is analyzed to provide a theoretical reference for the experiment.The simulation results show that the method proposed in this paper can effectively locate multiple magnetic targets and successfully applied the method to actual measurement experiments to accurately locate the spatial positions of multiple magnetic sources.For magnetic target recognition,to improve the effect of magnetic target recognition,the target detection method combined with the magnetic gradient tensor data analysis method is used to realize the rapid and accurate recognition of magnetic targets.First,through comparative experiments,it is proved that the tensor data analysis method of normalized magnetic source intensity is less affected by magnetic declination and inclination,and the boundary information presented is clearer.Using COMSOL software to establish a variety of magnetic target models,use the normalized magnetic source intensity method to process the data to highlight the target boundary,and establish a related sample set.Using this sample set,the Mask-RCNN network model is trained to recognize the category and shape of the magnetic target.The detection accuracy of the model is 97.5%,and the sphere target and the cuboid target are recognized,which verifies that the method in this paper has certain practicability.