| “Ore-prospecting nearby the existing deposits” is an effective wayto solve energy crisis. However, the working environment of mine areais complex, so it puts forwards higher request to the detection equipmentand the data processing method. From the view of ground and deviceplaced, a loop antenna which can be approximated to a horizontalmagnetic dipole has been designed and it has been demonstrated intheory; In order to obtain a higher sensitivity of magnetic sensor, theparameter analysis and correction has been done; In the last, for theelectromagnetic noise question, the FastICA algorithm has been putforward to separate signal from data with noise, and then thehigher-quality electromagnetic data will be gotten.The main contents of this paper can be summarized as followings:1) The theories argument of transmitting antenna used inelectromagnetic detection has been done. In order to make up for the lack of natural magnetic field signal, the high-frequency electromagneticwave has been produced by artificial method. The transmitting device isdesigned as a couple of loop antennas which are mutually perpendicular,and they are perpendicular to the ground while working. The signalswith14different frequency components which is between1kHz and80kHz were generated by selecting different combination of capacitors. Insome certain condition, the transmitting antenna can be approximated toa horizontal magnetic dipole. Through theoretical deduction andexperimental simulation, the electromagnetic wave generated by thehorizontal magnetic dipole is deemed to satisfy the plane wave in somesituations. The simulation result of resistivity under different conditionsgives the the relationship between distance, frequency and targetresistivity, and it provides a theoretic guide for the device placed. Thereceiver surface has been designed using LabVIEW graphicprogramming platform, and it provides a friendly man-machine interface.Using the interface, the experimental data can be observed at any time,and the function such as data pre-processing can be realized togetherwith some simple algorithms. 2) The non-uniformity of calibration magnetic field which is used inthe sensor sensitivity test is demonstrated in theory, and thenon-uniformity has been processed using average method. Theenvironment for magnetic sensor calibration requires no electromagneticinterference, and the calibration magnetic field must be stationary anduniform. However, the ideal situation cannot be achieved in practice.This paper analyzed the distribution of the magnetic field generated bydifferent equipment and decided to use long straight solenoid as themagnetic field generating equipment. The magnetic field generated bythe long straight solenoid can be deemed as stationary and uniform. Inthe past calibration experiments, the inner magnetic field of the solenoidis regarded as uniformly distributed, and the magnetic field around itscentral axis was used as the estimation of its inner magnetic field. Thisestimation caused the sensitivity of magnetic sensor to be lower than itsactual value so that the deviation is introduced. There is no numericalsolution of the non-axis magnetic formula, so the high-precisionGaussian quadrature formula is proposed and the result is consideredaccurate. The Simulation results show that the inner magnetic field of solenoid has little variation in landscape orientation. However, in thevertical orientation, the magnetic field is homogenous in the middle butvaries a lot in both ends. The intensity of magnetic field at both endsdrops to50%of that in the middle. In order to reduce the impact of thenon-uniformity and increase the accuracy, the average method isproposed to deal with this question, and the length of step is minimizedas possible to improve the accuracy. The experimental sensitivityprocessed by this method has improved than before.3) Various factors influencing sensor theoretical sensitivity has beenanalysed in detail, and the impact of the magnetic core’s averagemagnetic permeability is mainly discussed. Moreover, a correctionprogram for parameters in theoretical sensitivity formula of magneticsensor is also put forward. Since the magnetic core of magnetic sensorwas made from permalloy, and permalloy has a highmagnetic permeability under low frequency but themagnetic permeability drops under high frequency due to the increase inmagnetic hysteresis loss. In the broadband measurment, the effective magnetic permeability should be a function to the frequencybut not a fixed value. Unfortunately, the magnetic permeability ofmagnetic core material has a complicated relationship to the frequency,and it is also influenced by its shape. Due to the complexity, there is noquantitative function to calculate the relationship between them.Through the comparison with the experiment, the functionμ app (f)=1050.005f2+0.0005f+105μapp低is put forward to calculate therelationship between the magnetic permeability of magnetic core and itsoperating frequency. The compare between theoretical sensitivity andexperimental sensitivity showed that the corrected theoretical sensitivityis coincide with the experimental sensitivity.4) After analyzing the feature of the electromagnetic signal andnoise, the noisy electromagnetic data model has been built. The workingenvironment of mine area is complicated and there are various noises.The correct method to deal with the noise and gain high quality sensingdata were adopt only when the feature and classification of the noisewere recognized. Some common noise source in mine area has been analyzed, and time-domain waveform, bandwidth and impact on datainterpretation of the noise has also been introduced in detailsrespectively. The simulated data of different noise is produced after theanalysis. The simulated noise data were added to the original signal sothat the target wave for verifying different noise-cancellation methodwas generated.5) The FastICA algorithm is used to process the simulated data andthe experimental data of the electromagnetic detection. After analyzingthe feature of signal and noise of mine area, it was decided to adoptFastICA algorithm to process the data with noise. In order to evaluatethe denoising effectiveness of the algorithm objectively, two qualityevaluate parameters such as SNR and coherency were introduced. Firstly,the noisy electromagnetic data model was processed using FastICAalgorithm, and we found from the comparison of wave in bothtime-domain and frequency-domain before and after denoising that thesignal can be clearly distinguish in the denoised data while it is totallysubmerged in the data before denoising. The SNR of processed signal was much higher than the original data, so that it is proved this methodis very practical in processing the simulated data of mine area. In orderto test the method’s application effectivity for the experimental data, theFastICA algorithm is used to denoise the received pseudo-randomthree-frequency signal and the pseudo-random five-frequency signal.The processing result shows that this method is capable for extractingmeaningful signal from noisey data, and the coherency of electric signaland magnetic signal is improved obviously.The main innovation works of this paper is as followings:1) The circuits model and physical model of the loop antenna whichis perpendicular have been built. The spatial distribution ofelectromagnetic field produced by antenna under the different geologicalmodel has been gained while the antenna is approximated to a horizontalmagnetic dipole. Combined with the plane wave detection theory, therelationship between the resistivity of conductor and the distance as wellas the working frequency was obtained, which provided a theoreticalfoundation for antenna design. 2) The average method for processing the calibration magnetic fieldof long straight solenoid is introduced, so the precision of inductivemagnetic sensor’s sensitivity has been improved. Before this paper,almost every magnetic sensor was calibrated with the inner magneticfield of long straight solenoid deeming as homogenous distributed. Andthe intensity of solenoid’s central axis was deemed as the intensity of thewhole magnetic field. This approximation causes the experimentalsensitivity to be lower than its actual value. This paper took thenon-homogenization into consideration and proved the existence of thisnon-homogenization through theoretical deduction and numericalcalculation. Using averaging method to deal with it made the resultsmore close to the actual value.3) A function model of the relationship between themagnetic permeability of magnetic core and its working frequency wasbuilt, and this model resolved the problem that there is no formulabetween the magnetic permeability of magnetic core and its workingfrequency, and then the theoretical model of the magnetic sensor has been corrected. The effective magnetic permeability of magnetic corewhich is made from magnetic materials is not only related to the shapeof core but is also affected by the working frequency. Theeffective magnetic permeability of magnetic cores made from permalloymaterial is high while working under the low frequency, and it dropsalong with the increase of working frequency. So far, there is no specificformula to characterize the law of the change. This paper reviewedrelated documents and considered the comparison between magneticsensor’s actual measured sensitivity and its theoretical sensitivity. Thispaper refered to related documents and considered the comparisonbetween experimental sensitivity and its theoretical sensitivity. And thena low-pass filter’s transfer function was utilized as the initial mathematicmodel for effective magnetic permeability of magnetic core and itsworking frequency. The parameters of the initial mathematic model weremodified to achieve a more precise function expression after manyexperiments. The result of the research has modified the formulation tothe magnetic sensor’s sensitivity. 4) The noisy electromagnetic data model of mine area has beenbuilt, and an electromagnetic processing method based on FastICAalgorithm for mine area detection is so also proposed.that a standardelectromagnetic database of mine area for signal processing is provided.Using this method, the quality of data has improved as well as theutilization of the data has increased. The working environment of minearea is complicated and there are various noises. It will bring moreconvenience for the electromagnetic data processing of mine area tobuild a standard noisy electromagnetic data model of mine area afteranalyzing the features of the noise and simulating the noise wave.The production of this paper has provided theory foundation for the“Ore-prospecting nearby the existing deposits” of mine areaelectromagnetic detection, and and it also provided a reference forinstrument design. |
PDF Full Text Request