Study On Nonlinear Inversion For Complex Resistivity Based On Parallel Electrical Method

CR(Complex resistivity)method,also known as SIP(Spectrum Induced Polarization)method,is one of the geophysical prospecting methods that based on the different induced polarization characteristics of rocks and ores in the frequency domain.Multi-parameter is the main feature of complex resistivity inversion.Therefore,it is necessary to study the multi-parameter nonlinear inversion of the complex resistivity method.The theory of FEM(Finite Element Method)and system control theory as the theoretical basis of this paper.A variety of methods are used for study,such as FEM numerical simulation,ANN(artificial neural network),nonlinear inversion algorithms,circuit simulation,etc.were used to study the multi-parameter inversion of resistivity,Excitation current source signal analysis,and finite element modeling of 2.5D CR based on parallel electrical method.The study work of the 2.5D forward modeling of complex resistivity uses the parallel electrical method observation system as the basis of the measurement mode,and the DC resistivity method as the theoretical calculation basis.On the premise of neglecting the electromagnetic coupling,taking the Cole-Cole complex resistivity model into a complex resistivity forward modeling system based on parallel electrical method.At the same time,in the process of forward calculation,the pseudo-delta source is used to reduce the singular feature at the source point and improve the iterative calculation accuracy of the ILU-BICGSTAB algorithm.The relative error between the numerical potential value and the theoretical potential value near the source point is reduced to less than 0.15%,and the relative error of the node potential after leaving the source point is less than 0.08% or even smaller.In the forward calculation process,the parallel electric method observation mode is fully absorbed,and the approximate boundary conditions and the parallel calculation framework are used to further improve the efficiency of the forward calculation.The circuit simulation experiment research found that the single frequency sinusoidal signal can only obtain the response of exciting a single frequency,the measurement accuracy is high but the measurement efficiency is low,and the time cost is high;Rectangular pulse square wave type excitation signal,with strong antiinterference ability in observation stable field,suitable for direct current exploration,at the same time,it can obtain harmonic frequency response close to the main frequency,and the measurement efficiency is improved compared with single frequency signal;the M-sequence pseudo-random excitation signal has rich first main lobe frequency,high correlation identification accuracy,and strong anti-random interference ability.It is an ideal excitation current source signal for complex resistivity exploration.To solve the multi-parameter nonlinear inversion problem of complex resistivity,an inversion algorithm based on the combination of Quantum Particle Swarm Optimization(QPSO)and BP neural network was constructed.The inversion calculation process avoids the difficulty of solving the multi-parameter partial derivative matrix.The output of the algorithm is the inversion result,and the logical structure is simple and clear.The inversion results show that the QPSO-BP algorithm inversion clearly characterizes the anomalous boundaries and is closer to the values of the parameters.By establishing the geoelectric model of typical polarized anomalous bodies in coal seam floor,the forward response characteristics of apparent complex resistivity amplitude,phase and polarizability of different structural forms and different groundwater migration times are studied.It is found that the change in the spatial extent of the polarization anomalies in the coal floor will cause corresponding changes in the amplitude,phase and polarization rate of apparent complex resistivity in the observation area.Aiming at the problem of inversion of polarized anomalous bodies in coal seam floor,combined with the spatial structure characteristics of anomalous bodies to be inverted,training and test samples are reasonably set.The inversion calculation results show the good association ability of QPSO-BP algorithm for forward simulation data with similar characteristics.The inversion results are obvious in characterizing the boundary of the anomalous body,and the spatial location of the water-rich polarization anomaly area is consistent.It has a good inversion effect for the more complicated coal floor hidden faults and the unconventional form of water guide channels.