The Application Of Regression Analysis In Separation Of Superimposed Magnetic Anomalies

This paper is based on the theory of optimal linear filtering,make a comparative analysis of which the frequency response characteristics two filtering: Matched filtering and Wiener filtering in the separation of regional and local fields in magnetic data processing,and points out that the matched filtering is only a general case of Wiener filtering.In the 21 st century,where information technology is highly developed and the rapid development of electronic computing technology.At home and abroad,the application of frequency domain filtering technology is increasing.As we all know,whether it is in the aeromagnetic or in our ground magnetic test work,the magnetic anomaly data which we measured are a comprehensive reflection of the object being detected.That is a comprehensive reflection of different buried depths superimposed magnetic geological anomalies.While the spectrum of deep source field is mainly concentrated in the low frequency band,the shallow source field has strong medium and high frequency components.According to the characteristic difference of the spectrum,we can separate the regional field reflecting the deep source from the local field reflecting the shallow source.First,according to the application example of regression analysis to establish the corresponding regression analysis model.Through the experimental analysis of the model,it can be concluded that the nonlinear regression problem in the filtering process of superimposed anomaly can be transformed into a linear regression problem,and the established regression analysis model is effective and feasible.Then,through the analysis of which the plane characteristics and profile characteristics of the magnetic field of the sphere and the second-level cylinder forward model.The plane contours and the main profile curves of the single rule magnetic anomalies are obtained,and analyzing the relationship between the feature points of the main section curve(Such as extreme point,inflection point,zero point,etc.)and the depth of the center of the magnetic anomaly body,the corresponding relation of effective magnetization dip.On this basis,the characteristics of the plane contours and the main profile curves of the magnetic sphere superposition model,the second-level cylinder superposition model and the sphere--second-level cylinder superposition model with different burial depth are further discussed.Comparing and analyzing the characteristics of the abnormal curves of a single model and a combined model.Two different filtering methods are used to deal with the abnormal curve of the main section of ?T the combined model by forwarding.First,According to main section of ?T superimposed on the abnormal curve directly to the use of Wiener filtering or matching filtering method for filtering;The second filtering method is to introduce the regression analysis model in the process of filtering.Through the comparative analysis of the two methods,it is known that the filtering method introduced by the regression analysis method is better for the separation of the superposition field source.Finally,this method is applied to the measured data in the work area,and the actual section of the work area was subjected to abnormal separationwhichalso geting a very good effect.The results show that the regression analysis method has certain assisting effect on the filtering separation of superimposed magnetic anomalies.