Research On Aviation Transient Electromagnetic Noise Reduction Based On Kernel Function

Aviation transient electromagnetic method is an aircraft geophysical exploration method using an aircraft as a carrier and electromagnetic induction as an exploration mechanism.Compared to terrestrial transient electromagnetic method,aviation transient electromagnetic method has the advantages of rapidity,high efficiency,low cost,and less influence of terrain.Therefore,aviation transient electromagnetic method has been widely used in swamps,lakes,deserts,mountains,oceans,and seas.Mineral exploration in landfills and other areas,as well as hydrological,engineering,environmental investigations and pollution monitoring,are particularly suitable for the harsh geological environment in the central and western regions.However,due to the wide frequency band of aviation transient electromagnetic signals,especially the weak signals in the middle and late stages,they are vulnerable to external noise.If the noise in the electromagnetic signal cannot be effectively suppressed,the reliability and quality of the data will be seriously affected,and the inversion accuracy of the later data will be affected.This paper relies on the “Thirteenth Five-Year Plan” national key R&D project “Helicopter Aviation Electromagnetic Detection Data Processing Interpretation Software System R&D” to study the method of aviation transient electromagnetic data denoising.Firstly,the forward simulation of aviation transient electromagnetic data and noise is performed.Secondly,the linear principal component denoising method and the kernel principal component denoising method are studied.At the same time,the effectiveness of the simulated noisy electromagnetic data programming is verified to verify its validity.Finally,it is verified by the measured data.The applicability of the denoising method studied.Specifically,the main content of this paper is as follows:(1)Aviation transient electromagnetic de-noising method based on linear principal component analysis.The principal component analysis method mainly uses the first-order and second-order statistical characteristics of the data and obtains uncorrelated or nearly uncorrelated components through coordinate transformation,thereby achieving signal-to-noise separation.This paper first reads the noisy data,then normalizes the data and finds the zero-mean covariance matrix.Then it performs eigenvalue decomposition on the covariance matrix to obtain all the components of the data.Finally,the energy analysis of each component is used to extract the data.The main components of the signal are reconstructed and the filtered data is obtained.In order to test the performance of the linear principal component denoising method,electromagnetic signals with different signal-to-noise ratios were used in the experiment.The analysis of the experimental results shows the scope of application of the linear principal component denoising method.(2)Aviation transient electromagnetic de-noising method based on kernel principal component analysis.Because the linear principal component denoising method only uses the first-order and second-order statistical properties,this method can not describe non-Gaussian signals.Therefore,the kernel principal component analysis method is recommended for this purpose.The kernel principal component analysis method is a nonlinear principal component analysis method.The core of the method is to map the original data to the nuclear space through a nonlinear transformation function,so that the original data space can only be converted into data using nonlinear principal component analysis.Linear principal component analysis data can be used in nuclear space.Therefore,as a nonlinear principal component analysis method,the kernel principal component analysis can use high-order statistical properties to better describe non-Gaussian signals.This paper first reads in the noisy data,then normalizes the data and computes the data components in the kernel space through the kernel function.According to the analysis of each component in the nuclear space,finally obtains the main component in the original data space by the approximation method.The source image is reconstructed to achieve signal-to-noise separation.In order to test the performance of the kernel principal component denoising method,the same noisy electromagnetic data as the linear principal component denoising method was used in the experiment.The experimental results show that the kernel principal component denoising method has better denoising effect than the linear principal component denoising method.(3)In order to verify the applicability of different kernel principal component de-noising methods to measured data,this paper uses the above method to de-noise the field aviation flight survey data of a certain exploratory region in Hami,Xinjiang,and evaluate its de-noising performance.