Research On Target Localization Methods Based On Magnetic Anomaly Gradient Signal Detection

Airborne magnetic detection technology is now widely applied to both military and civil use fields, and shows promising prospects. Generally, geomagnetic field,which is considerably larger than the magnetic field intensity generated by ferromagnetic materials at some distance, is the major natural magnetic field on the surface of the earth. Nevertheless, geomagnetic field would also produce a negative effect on magnetic characteristics of other ferromagnets, that possess not only intrinsic magnetic moment but also induce magnetic moment, which jointly determine the magnetic field of ferromagnetic materials. As a necessary condition under which magnetic detection could proceed, magnetic anomaly field would always be motivated by the disturbance generated by ferromagnets on the earth magnetic field. Magnetic anomaly detection(MAD), with no need to emit signals on its initiative, is classed as one of the so-called passive detection technologies. Being strong in penetrability, MAD signals can be in existence steadily and consecutively. This thesis mainly presents the detection methods of magnetic anomaly gradient signals.First, we need to analyze the characteristics of magnetic anomaly gradient signals and model them for detection. Evolved from the traditional signal detection technology,gradient MAD signal detection regards the ferromagnet as a dipole, hence obtaining its magnetic field model in the space. Then, the gradient signal is decomposed as 4independent orthogonal basis functions(OBF) from the gradient MAD signals before operating the correlation detection to them. When the aircraft moves non-inertially and changes its postures, we will record the changing signal characteristics as detection features for recognition, since the magnetic measurement device is fixed on the detection aircraft.Second, airborne magnetic detection is conducted in the open and natural environment, which will inevitably introduce various magnetic noises, so we can do nothing before analyzing the magnetic noises in nature. Furthermore, magnetic disturbance due to the airborne detection platform will also be recorded, making airborne magnetic compensation a significant part of the detection process. As for the magnetic noise caused by the geological nonuniformity, we place top priority on geomagnetic modeling to deal with it. Regional geomagnetic modeling is carried out byapplying the cubic B-spline interpolation to the collected geomagnetic data improving both the resolution of spatial magnetic field and accuracy of the model. Next, we gather the data of the environmental magnetic field statically and dynamically, and analyze the spectrum and time-frequency characteristics respectively, which provides certain guidance for the subsequent signal processing.Next, we conduct the gradient orthogonal basis function detection directly for the gradient MAD signals with only Gaussian noise in them, contributing largely to increase the signal to noise ratio(SNR). For colored noises whose power spectral density function equals ? 201 ?? ???f, we whiten and filter the noise, largely improving the effectiveness of detection on the whitened and filtered basis functions compared with original basis functions for ? single-valued case. With regard to nonstationary colored noises, we improve the unified whitening filter with modulation factor added as a condition for reconstitution of the whitening filter. By this means, the detection effectiveness of the signal passing the modulation whitening filter is improved in contrast to the unified whitening filter.Finally, we operate the target detection using with the signal energy combined with the constant false alarm rate(CFAR) detection with OSVI-CFAR detector. Meanwhile synthesizing characteristics of the radar CFAR detection and the gradient MAD detection, with fixed threshold is introduced to recognize the target signal more accurately. We have acquired mass data of the geomagnetic field by performing a multitude of outdoor experiments, after which we conduct correlation processing to the experimental data is conducted to verify the method of signal processing. In the end, we synthesize each module of the procedure of the gradient MAD signal detection, and complete the software design using the MATLAB GUI detection.Through several detection tracks, we can accomplished the target location.