Magnetic Anomaly Signal Detection And Source Locating Method Research

We live in an environment where magnetic field distribution everywhere due to the presence of the great earth magnets, so any object on the earth is affected by geomagnetic field. Additional magnetic fields produced by local ferromagnetic objects or electric currents causes a perturbation to the geomagnetic fields to form the magnetic anomaly signals. Detection and localization of magnetic target based on the magnetic anomaly phenomena is called Magnetic Anomaly Detection(MAD). MAD is a passive detection method based on the basic physical phenomena. It has good stability and versatility. MAD technique has been widely used in military scenario, geological resources exploration, archeology, earthquake early warning, space exploration, medical diagnostics, and biological magnetics. This thesis is mainly concerned with the description and processing of magnetic anomaly signals for aero geophysical survey.First, based on the fact that the magnetic anomaly signal is very weak compared with the ambient magnetic noise, we apply the orthonormal basis function decomposition(OBF) algorithm that can effectively improve the SNR. The algorithm calculates the energy function of the magnetic anomaly signal in the selected basis function space. Simulation results confirm that the algorithm can obtain an improved SNR and give correct estimation of the characteristic time. Further analysis show that the energy function is not optimal because the autocorrelation and cross-correlation functions of the three basis functions do not have good shapes. An improved scheme that combines the three basis functions to form a new basis function is developed, which makes the energy function to look like a much sharper impulse.Secondly, considering the circumstances that the processing effects by OBF algorithm is optimal for magnetic anomaly signal contaminated by Gaussian white noise but is not satisfactory in case of non-Gaussian white noise. We select a whitening filter method based on the autoregressive(AR) model to suppress the geomagnetic noise with a power spectral density(PSD) of 1/ f?0 2???? ?, which is the major component of the environment magnetic noise. For marine background magnetic noise that is in general non-Gaussian magnetic noise, we present a method based on wavelet transform. It is verified by simulations with Rayleigh distribution and exponential distribution that the effectiveness of whitening is remarkable.Third, the energy function is used as the test statistics for target detection by comparison with a threshold function. If the energy function is bigger than the threshold within a continuous period of a time, a target is considered to have been detected. In order to achieve the effect of constant false alarm rate(CFAR) detection based on the distribution characteristic of energy function, adaptive thresholds are determined by using both sides of the unit average lager selection of the CFAR detection algorithm.Finally, the OBF algorithm is extended from its classical case that either the platform or the target but not both is moving, to more general case that both the platform and target can be moving simultaneously at any speeds and relative track angle. Accordingly, a “&-route” searching strategy is suggested for tracking of the magnetic target.