Study On The Model And Experiment Of Spectrum Characteristics Of Rock Mineral In Natural Field IP

The previous studies put forward 12 kinds of models to describe the induced polarization (IP effect) of natural rock ore through laboratory and field measurement.These models are widely applied in mineral resources exploration, geological structure survey, etc. In this article,the performance of seven existing models proposed to describe IP has been studied systematically based on large amounts of simulation calculation and comparison analysis, the spectrum test of natural rock ore and inversion fitting experiment. The purpose is to determine the most effective models for describing the IP effects of various rock ore.First,a program "IP equivalent model of the rock ore simulate computing system" was constructed for the seven models using a Delphi7 programming language,in order to calculate the spectrum of phase, amplitude, real and imaginary parts. The influences of the parameters on the spectrum and the spectral characterization scope have been studied also systematically. More than 50 pieces of various natural rock ore samples from 16 regions of China were collected for the sweep-frequency measurement using SolartIon SI-1260 spectrometer in the 10-3?105 Hz range. In this test, we have obtained the spectrum of natural rock ore and summarized their characteristic rules.In order to study the performance of the seven models, another Matlab program written based on least squares and differential evolution algorithm (DE) was used for inversion fitting the experimental data of samples with present unimodal and bimodal spectrum characteristics, respectively. The program was able to inversion fit the spectrum of the samples with the spectra calculated by the models. The fitting precision are evaluated by percent root mean square error (RMSE). The fitting precision of seven models can be comparable intuitively to determine all kinds of natural rock (ore) corresponding to the most effective model.The results suggest that the best fit for characterizing the unimodal and bimodal phase spectra of the selected metal minerals is Dias and Multi-Cole-Cole model. It is first time that bimodal phase spectrum data of a variety of metal minerals are fitted directly to get high precision using the two models. Warburg, Cole-Cole and Generalized Cole-Cole model are suitable for characterizing the spectrum of the unimodal characteristics. Madden and Cantwell model is suitable to characterize the spectrum of silver ore and gold bearing pyrite, but the characterization ability is poor for other ore samples. Debye model has the low fitting precision for all the samples above. In conclusion, the ability to describe IP of the classical models is different for the peak position and peak width of phase curve. When analyze the electromagnetic data collected from the field, the use of the improper IP model for forward or inverse analyses under some conditions will lead to major errors.Therefore, it is especially important in choosing the appropriate model against different types of minerals. The conclusion can provide theoretical basis for choosing the appropriate model with the frequency domain IP method applied to mineral exploration.In this paper, a new IP parameter, phase offset ?P is put forward, it with relative video dispersion rate PPs constitute a set of effective parameter are used to characterize the IP effect of the metal minerals. It provides a new thought for practical applications.