Apparent Resistivity Inversion Research Of Electrical-source Of Semi-airborne Transient Electromagnetic Based On Neural Network Metyiod

Electrical-source of Semi-airborne transient electromagnetic method is a part ofTime-domain electromagnetic methods(TEM).It is consists of a long grounding wireas transmit source, when sending electromagnetic pulse into the ground, receivingsystem placed on airship or unmanned plane collects the responsesignal. Its presencesolved the low efficiency and terrain restricted problems in ground transientelectromagnetic system, small exploration depth and man-machine safety problemsin airborne transient electromagnetic system. Semi-airborne transient electromagneticmethod has already became the important research objects of electromagneticexploration field in recent years.Resistivity calculations of Electrical-source of Semi-airborne transientelectromagnetic at the present time mostly are approximated by using long offsettransient electromagnetic method on the ground. on the one hand, ignoring the altitudeimpact leads to the big deflection between apparent resistivity and the true resistivity.Withal, forward modeling calculation speed is slow, the multiple solution problems ofapparent resistivity calculation results in low inversion efficiency and poorapplicability.Combined with the Double Ten Project “development of Time domainSemi-airborne transient electromagnetic system” of JiLin province, This articleresearched these problems based on the theory of Semi-airborne transientelectromagnetic and Artificial Neural Network. The main research content is asfollows:(1) the author further studied the electromagnetic field theory of Semi-airbornetransient electromagnetic, deduced the expressions of the vertical magnetic fieldresponse of grounded long wire electrical source. Realized fast forward calculation of the electrical-source Semi-airborne transient electromagnetic response in time domainby using MATLAB. Took forward calculation for typical earth model, verified theeffectiveness of the algorithm.(2) Discussed the characteristic of Semi-airborne transient electromagnetic response.analysis the influence of altitude through the comparative study with the groundtransient electromagnetic response. Then presented the basic method of datainterpretation. research the structure, principle and realization method of BP artificialneural network, as well as the process of the establishment of the artificial neuralnetwork which is suitable for the Semi-airborne transient electromagnetic dataexplain.(3) we conduct an equivalent conversion of the induced voltage based on fast forwardmodeling and build single sample set. Then, choose three-layer BP neural networkand Levenberg-Marquardt algorithm for training sample set, optimizing the networkparameter with convergence speed and small error. The feasibility and effectiveness ofneural network inversion method is verified by simulation model.(4) Involved in the electrical-source of Semi-airborne transient electromagneticsystem field test and the data processing and interpretation. Presented the process andresult of the test data processing. The result is in line with results of other surveymethods and local geological conditions.This paper completed the fast theoretical calculation of electrical-sourceSemi-airborne transient electromagnetic response, laid a theoretical foundation forSemi-airborne transient electromagnetic data interpretation. Mainly studied theelectrical-source of Semi-airborne apparent resistivity interpretation method based onartificial neural network, ensure fast data interpretation for wild flight experiments.