METHOD FOR THE LOCATION AND ESTIMATING OF CHROMITE ORE BODYs BASED ON MULTIFREQUENCY ELECTROMAGNETIC RESPONSES

Description

2. This is attributable to surface disordered electrical structures. The same data set from field measurements can often lead to completely different interpretations of the location and morphology of the ore body. This ambiguity significantly impairs the reliability and accuracy of decisions for locating drilling targets. Therefore, the effective separation of near-surface noise and the true response of deeper, high-resistance ore bodies under strongly disturbed field conditions, as well as the representation of clearly distinguishable frequency-change characteristics of signals from ore bodies at different depths, represents a central problem that must be solved for the precise localization of chromite ore and the reliable estimation of resource size. 10. CONTENT OF THE PRESENT INVENTION To solve the aforementioned technical problems, the present invention proposes a method for locating and estimating chromite ore bodies on the Basis of multi-frequency electromagnetic responses,This enables the accurate localization and reliable estimation of chromite ore bodies under complex geological conditions. To achieve the above-mentioned goal, the present invention provides a method for the localization and estimation of chromite ore bodies based on multifrequency electromagnetic responses, comprising: obtaining mixed responses and an initial spectral distribution, containing both near-surface interference and deep high-impedance anomalies, from acquired electromagnetic signal sequences; performing a multiscale decomposition based on the initial spectral distribution to separate low-scale subbands dominated by near-surface interference and medium- to high-scale subbands associated with deep ore bodies; Suppression of interference in the sub-bands of low scale based on a signal attenuation degree of the sub-bands of medium to high scale,to obtain a cleaned frequency spectrum after interference removal; extracting frequency features from the cleaned frequency spectrum and identifying narrowband patterns of the high-resistance anomaly to obtain anomaly localization vectors; performing an inversion of the anomaly localization vectors while integrating an electrical model of ultrabasic rocks to estimate the depth of an ore body; generating a multi-depth transmit parameter matching scheme based on the depth of the ore body to obtain an optimized electromagnetic signal acquisition sequence; and separating ore body responses from the optimized electromagnetic signal acquisition sequence and validating their agreement with the frequency features to provide a basis for drilling decisions. Optionally, this includes obtaining mixed responses and an initial spectral distribution, which contains both near-surface interference and deep high-impedance anomalies.from acquired electromagnetic signal sequences, the following: Discretizing acquired raw electromagnetic signals to obtain a time series sequence; identifying features of a superposition of abrupt changes in high frequency and gradual changes in low frequency in the time series sequence and, based on this, extracting mixed signal segments; and performing a discrete frequency domain transformation of the mixed signal segments to obtain an amplitude vector in the frequency domain, and calculating a power spectral density based on the amplitude vector in the frequency domain to determine the initial spectral distribution. Optionally, performing a multiscale decomposition based on the initial spectral distribution includes the following: performing a discrete wavelet transformation of the initial spectral distribution.to obtain multiscale decomposition coefficients; delineating scale ranges of the near-surface high-frequency response and scale ranges of the deeper-lying low-frequency response based on the energy magnitudes of the multiscale decomposition coefficients; 25 separating detail coefficients of the scale ranges of the near-surface high-frequency response and mapping them to generate the low-scale subbands; and reconstructing coefficients of the scale ranges of the deeper-lying low-frequency response to obtain the mid- to high-scale subbands. 30 Optionally, suppressing interference in the low-scale subbands based on a signal attenuation rate of the mid- to high-scale subbands includes: calculating a signal attenuation rate of the mid- to high-scale subbands; BE2026 / 7030 4 Using lower scale subbands as reference signal and the initial spectral distribution as main input signal,when the signal attenuation rate exceeds a predetermined threshold; iterative adjustment of weighting coefficients of an adaptive filter to match a correlation between the reference signal and the main input signal; and convolving the reference signal with the converged weighting coefficients to obtain an estimated noise component, and subtracting the estimated noise component from the main input signal to determine the cleaned frequency spectrum. Optionally, extracting frequency features from the cleaned frequency spectrum and identifying narrowband patterns of the high-impedance anomaly includes: constructing a matrix of frequency features based on the cleaned frequency spectrum; performing a feature decomposition of the matrix of frequency features, selecting eigenvectors whose contribution rate meets predetermined requirements, to form a projection-transform matrix; Processing matrix of frequency characteristics using the projection-transformation matrix,to obtain spectral principal component features, and to identify data on narrowband patterns of high-resistance anomalies based on the spectral principal component features; and to generate anomaly localization vectors based on the narrowband pattern data and their spatial position information. Optionally, performing an inversion of the anomaly localization vectors under integration of an electrical model of ultrabasic rocks includes: constructing an initial electrical structure model with three-dimensional geometric boundary conditions based on the anomaly localization vectors; performing a numerical forward simulation of the initial electrical structure model to obtain forward simulation response data; calculating a fitting residual between the forward simulation response data and the anomaly localization vectors.as well as iterative optimization of the fitting residue by means of an inversion algorithm to generate a depth inversion matrix; and calculating a coefficient of agreement between the depth inversion matrix and predefined typical electrical patterns of ultrabasic rocks, and determining the positions of the upper and lower interfaces of a high-resistance anomaly body based on the coefficient of agreement to obtain an estimation result of the ore body depth. Optionally, generating a fitting scheme for multi-depth transmit parameters based on the depth of the ore body includes the following: calculating a set of transmit frequencies for the corresponding penetration depth based on the positions of the upper and lower interfaces in the estimation result of the ore body depth; Configuring groups of physical quantities of the transmitting sources that have different pulse widths and current intensities,according to the set of transmit frequencies; 10 Subdividing the groups of physical quantities of the transmit sources into multi-stage signal acquisition time windows to form a multi-stage time-domain sampling configuration; and encoding and packaging the multi-stage time-domain sampling configuration into the optimized electromagnetic signal acquisition sequence. 15 Optionally, separating ore body responses from the optimized electromagnetic signal acquisition sequence and validating their conformance includes: removing background field responses from the optimized electromagnetic signal acquisition sequence to isolate pure ore body secondary field response data; 20 Performing a time-frequency transformation and a petrophysical mapping of the pure ore body secondary field response data to obtain values ​​for the frequency dispersion rate and differences in polarizability; Check whether the values ​​of the frequency dispersion rate and the difference in polarizability satisfy linear correlation conditions.and based on that25 calculating an anomaly consistency confidence value; and determining a target area validation vector based on the anomaly consistency confidence value to form the basis for decision-making for drilling operations. Technical Effects of the Invention: The present invention discloses a method for locating and evaluating chromite ore bodies based on multi-frequency electromagnetic responses. By means of multi-scale decomposition using wavelet transformation in combination with adaptive filtering, near-surface disturbances are effectively separated and suppressed, thereby significantly improving the signal-to-noise ratio of weak responses from deeper, high-resistance ore bodies. Based on a cleaned frequency spectrum, principal component analysis enables the precise identification of narrowband patterns of high-resistance anomalies. In conjunction with an electrical model of ultrabasic rocks, an inversion is performed.This significantly increases the accuracy and reliability of the depth estimation of the ore body and effectively reduces the ambiguity of the inversion. By dynamically optimizing the transmission parameters depending on the estimated depth, the present invention achieves targeted amplification and acquisition of signals at different depths, thus overcoming the depth limitations of conventional single-transmission methods. Finally, consistency validation provides a quantified basis for decision-making regarding drilling measures, enabling precise localization and reliable size estimation of chromite ore bodies under complex geological conditions. DESCRIPTION OF THE DRAWING The attached drawings, which are part of this application,for the purpose of a better understanding of this application. The exemplary embodiments and descriptions of this application serve to explain this application and do not constitute an inadmissible limitation of this application. In the drawing: Fig. 1 is a flow diagram of a method for locating and estimating chromite ore bodies based on multi-frequency electromagnetic responses according to an embodiment of the present invention. 20 DETAILED DESCRIPTION It should be noted that the embodiments and features described in this application are combinable with one another unless otherwise stated. The application is now described in detail with reference to the attached 25 drawing and embodiments. It should be noted that the steps shown in the flowchart in the attached drawing can be executed in a computer system like a series of computer-executable instructions, and although the flowchart shows a logical sequence,The steps shown or described may, in some cases, be carried out in a different order than shown here. As illustrated in Figure 1, this form of execution provides a method for locating and estimating chromite ore bodies based on multi-frequency electromagnetic responses, comprising: BE2026 / 7030 7 Obtaining mixed responses and an initial spectral distribution containing both near-surface interference and deep high-impedance anomalies from acquired electromagnetic signal sequences; Performing a multi-scale decomposition based on the initial spectral distribution to separate low-scale subbands dominated by near-surface interference and mid- to high-scale subbands associated with deep ore bodies; Suppression of interference in the sub-bands of low scale based on a signal attenuation degree of the sub-bands of medium to high scale,to obtain a cleaned frequency spectrum after interference removal; extracting frequency features from the cleaned frequency spectrum and identifying narrowband patterns of the high-resistance anomaly to obtain anomaly localization vectors; performing an inversion of the anomaly localization vectors while integrating an electrical model of ultrabasic rocks to estimate the depth of an ore body; generating a multi-depth transmit parameter matching scheme based on the depth of the ore body to obtain an optimized electromagnetic signal acquisition sequence; and separating ore body responses from the optimized electromagnetic signal acquisition sequence and validating their agreement with the frequency features to provide a basis for drilling decisions. Furthermore, obtaining mixed responses and an initial spectral distribution includes both near-surface interference and deep high-impedance anomalies.from acquired electromagnetic signal sequences, the following: Discretizing acquired raw electromagnetic signals to obtain a 25-second time series sequence; Identifying features of a superposition of abrupt changes in high frequency and gradual changes in low frequency in the time series sequence and, based on this, extracting mixed signal segments; and performing a discrete frequency domain transformation of the mixed 30 signal segments to obtain an amplitude vector in the frequency domain, and calculating a power spectral density based on the amplitude vector in the frequency domain to determine the initial spectral distribution. Specifically, the implementation process of this execution form comprises the following: To detect the characteristics of high-frequency abrupt changes and low-frequency gradual changes in time series, an analysis of the local rate of change and the periodicity of the signal can be performed. It is assumed thatthat in the aforementioned sequence, a specific data segment exhibits strong amplitude changes within a short time, which may indicate high-frequency abrupt changes due to near-surface interference; while another data segment shows slow, periodic fluctuations, which may be due to low-frequency, gradual changes due to deeper-lying high-impedance anomalies. Based on these features, a mixed signal segment is extracted that contains both types of features superimposed, for example, the data points from point 2000 to point 3000, corresponding to a signal segment with a duration of 1 second. This process helps to focus on the key signals, avoid interference from irrelevant data, and increase analysis efficiency. After extraction of the mixed signal segment, the mixed response data can be further recovered and will be subjected to a discrete transformation into the frequency domain. It is assumed thatthat a fast Fourier transform (FFT) is applied to the aforementioned 1-second signal segment to convert the time signal into a frequency signal and to obtain a corresponding amplitude vector in the frequency domain. This vector reflects the intensity distribution of the signal over different frequencies; for example, a higher amplitude at 10 Hz may correspond to a low-frequency gradual change, while a pronounced amplitude at 500 Hz may indicate a high-frequency abrupt change. Such a transformation facilitates the separation of different frequency components and provides a clear perspective in the frequency domain for subsequent analysis. Furthermore, performing a multiscale decomposition based on the initial spectral distribution includes the following: performing a discrete wavelet transformation of the initial spectral distribution,to obtain 25 multiscale decomposition coefficients; delimiting scale ranges of the near-surface high-frequency response and scale ranges of the deeper-lying low-frequency response based on the energy magnitudes of the multiscale decomposition coefficients; separating detail coefficients of the scale ranges of the near-surface 30 high-frequency response and mapping them to generate the low-scale subbands; and reconstructing coefficients of the scale ranges of the deeper-lying low-frequency response to obtain the mid- to high-scale subbands. Specifically, the implementation process of this execution form comprises the following: 35 BE2026 / 7030 9 Through an energetic analysis of the multiscale decomposition coefficients, the scale intervals for near-surface high-frequency responses and deeper-lying low-frequency responses can be clearly delineated. Surface disturbances typically manifest themselves through high-frequency, short-term abrupt features, while deeper-lying high-impedance anomalies predominantly manifest themselves through low-frequency,long-period fluctuations are characterized.5 It is assumed that in the aforementioned decomposition, a high energy concentration exists in scale levels 1 to 2, which accounts for more than 50% of the total energy, corresponding to the high-frequency responses of near-surface disturbances; while in scale levels 4 to 5, a comparatively uniform, but temporally sustained 10 energy distribution exists, accounting for approximately 30% of the total energy and thus potentially related to deeper-lying anomalies. Such an analysis of the energy distribution allows the signal sources of different scale ranges to be uniquely determined. By separating the detail coefficients of the scale ranges with near-surface high-frequency responses and mapping them back into the time-frequency domain, a lower-scale subband can be generated, which is predominantly dominated by near-surface disturbances. Specifically, it is assumed thatthat the detail coefficients of scale levels 1 to 2 are extracted; these coefficients reflect the rapidly varying components of the signal, such as surface clutter or short-term electromagnetic disturbances. By mapping them back into the time-frequency domain, the temporal and frequency distribution of these disturbance signals can be clearly represented, for example, the concentrated occurrence of high-frequency components in a specific time interval. This separation facilitates subsequent targeted processing of the disturbance signals. By reconstructing the approximation coefficients and the remaining detail coefficients in the scale range of the lower-lying low-frequency response, subbands of medium to high scale can be generated.which are associated with the deep-lying ore body. Furthermore, the suppression of interference in the low-scale sub-bands based on a signal attenuation rate of the medium- to high-scale sub-bands comprises the following: calculating a signal attenuation rate of the medium- to high-scale sub-bands; using the low-scale sub-bands as the reference signal and the initial spectral distribution as the main input signal when the signal attenuation rate exceeds a predetermined threshold; iteratively adjusting weighting coefficients of an adaptive filter to match any correlation between the reference signal and the main input signal; and convolving the reference signal with the converged weighting coefficients to obtain an estimated interference component, and subtracting the estimated interference component from the main input signal.to determine the corrected frequency spectrum.5 Specifically, the implementation process of this implementation form comprises the following: The filter's weighting coefficients are iteratively adjusted based on the statistical correlation features between the reference signal and the main input signal. The specific process can be carried out using the Least Mean Square (LMS) criterion, whereby the weights are stepwise optimized to maximize the degree of agreement between the two signals. In one possible implementation, the initial correlation coefficient is 0.42 and converges to 0.89 after approximately 15 iterations, at which point the weighting coefficients reach a stable state. Using the converged weighting coefficients, a convolution operation is performed on the reference signal,15 to obtain an estimated disturbance component. In the aforementioned exploration data, the folding result shows,that the estimated interference component reaches a peak value of 85 microvolts in the near-surface time domain and is concentrated predominantly in higher frequency ranges, which largely corresponds to the characteristics of the low-scale subband. 20 Subsequently, a differential is calculated between the main input signal and the estimated interference component, resulting in an error signal. The verified error signal represents the cleaned frequency spectrum obtained after interference suppression. In practical applications, interference suppression is considered effective if the energy component of the error signal obtained after differential calculation is less than 8% of the total energy and the low-frequency signal components are fully preserved. The technical advantage is that this adaptive filtering method selectively removes the masking of lower-lying signals by near-surface interference.so that the low-frequency features of deeper-lying high-resistance anomalies stand out clearly in the cleaned frequency spectrum. In the processed frequency spectrum, pronounced peaks appear in the frequency range of 8 to 15 Hz, whereas these peaks were masked by high-frequency noise in the intra-frequency spectrum. In this way, exploration professionals can more accurately identify the location of underground high-resistance ore bodies, which significantly improves the reliability and interpretability of resource exploration. 35 BE2026 / 7030 11 Furthermore, extracting frequency features from the cleaned frequency spectrum and identifying narrowband patterns of the high-resistance anomaly includes: constructing a matrix of frequency features based on the cleaned frequency spectrum; 5 Performing a feature decomposition of the frequency feature matrix, selecting eigenvectors whose contribution rate meets specified requirements,for the formation of a projection-transformation matrix; processing the matrix of frequency features using the projection-transformation matrix to obtain spectral principal component features, and 10 identifying data on narrowband patterns of the high-impedance anomalies based on the spectral principal component features; and generating anomaly localization vectors based on the data on narrowband patterns and their spatial position information. Specifically, the implementation process of this execution form includes the following: 15 A principal component analysis, i.e., an eigenvalue decomposition, is applied to the matrix of frequency features to calculate the variance contribution rates of the individual eigenvectors. Usually, the first several eigenvectors are selected whose cumulative variance contribution rate reaches at least 85%.and assembled into a projection-transformation matrix. In a given measurement area, the original matrix of 20 frequency features comprises 30 frequency points. After decomposition, the variance contribution rates of the first four principal components are 52%, 21%, 9%, and 6%, resulting in a cumulative variance of 88%; therefore, these four eigenvectors are used to construct the projection matrix. Using this projection-transformation matrix, a linear transformation of the original matrix of 25 frequency features is performed, yielding dimension-reduced spectral principal component features. These principal component features preserve the essential energy information while simultaneously filtering out some of the random noise as well as minor fluctuations. Each measurement point is represented after the transformation by a four-dimensional vector.which appropriately summarizes the overall shape of the adjusted frequency spectrum at this point. Based on the spectral principal component features, the narrowband pattern data of high-impedance anomalies can be identified. High-impedance bodies typically show a concentrated energy gain in a specific narrowband frequency range in the adjusted frequency spectrum, while the remaining frequency ranges remain relatively flat. By setting a threshold for the energy concentration and limiting the frequency bandwidth, for example, if the energy contribution of a frequency range is more than 45% of the total energy at this measurement point and the bandwidth is less than 10 Hz, this is interpreted as a narrowband pattern of a high-impedance anomaly. In the actual data, the principal component features of a measurement point show a pronounced peak. in the range of 11 to 14 Hz, whose energy content is 48% of the total spectrum and whose bandwidth covers approximately 8 Hz,which corresponds to the typical response characteristics of a high-resistance anomaly. Furthermore, performing an inversion of the anomaly localization vectors while integrating an electrical model of ultrabasic rocks includes the following: 10. Constructing an initial electrical structure model with three-dimensional geometric boundary conditions based on the anomaly localization vectors; performing a numerical forward simulation of the initial electrical structure model to obtain forward simulation response data; Calculating a fitting residue between the response data of the 15 forward simulation and the anomaly localization vectors, as well as iteratively optimizing the fitting residue using an inversion algorithm to generate a depth inversion matrix; and calculating a coefficient of agreement between the depth inversion matrix and predefined typical electrical patterns of ultrabasic rocks,and determining the positions of the upper and lower boundary surfaces of a high-resistance anomaly body based on the coefficient of agreement to obtain an estimation result of the depth of the ore body. Specifically, the implementation process of this implementation includes the following: The high-grade regions in the anomaly location vector field are projected onto a three-dimensional grid. In conjunction with prior geological information such as stratigraphic dip and fault directions, geometric boundary conditions are applied to the high-resistance anomaly body, for example by defining a near-ellipsoidal cylindrical or plate-shaped body. This creates an initial resistivity pattern. In the northeast of a particular exploration area, the anomaly location vector shows a high-grade region in elliptical shape,whose longitudinal axis direction coincides with the regional structure lines. Therefore, in the initial model, the upper boundary of the high-resistance body is set to a depth of 150 meters and the lower boundary to a depth of 450 meters, whereby the electrical resistivity of the body is set to 5000 ohm-meters and the resistivity of the surrounding 35 host rocks to 800 ohm-meters. BE2026 / 7030 13 To carry out the forward modeling of the initial electrical structure model, the finite element method is used. The simulated impedance and phase response data are calculated. A three-dimensional finite element lattice decomposition is applied, whereby the simulated frequency spectrum from 0,1 to 100 Hz is sufficient. The forward response curve is calculated for each measurement point. Subsequently, the function 5 of the fitting residual between the response data of the forward simulation and the vectors of the anomaly localization is calculated. Furthermore, generating a fitting scheme for multi-depth transmit parameters based on the depth of the ore body includes the following: calculating a set of transmit frequencies for the corresponding penetration depth based on the positions of the upper and lower interfaces in the estimation result of the ore body depth; configuring groups of physical quantities of the transmit sources that have different pulse widths and current intensities according to the set of transmit frequencies; and subdividing the groups of physical quantities of the transmit sources into multi-stage signal acquisition time windows.to create a multi-stage time-domain sampling configuration; and encoding and packaging the multi-stage time-domain sampling configuration into the optimized electromagnetic signal acquisition sequence.20 Specifically, the implementation process of this implementation form comprises the following: After determining the depth estimation of the ore body, the upper boundary of the high-resistance anomaly body is at 210 meters and the lower boundary at 365 meters. Based on the attenuation law of electromagnetic waves in a medium where the penetration depth is inversely proportional to the frequency, the25 principal frequency set that covers this depth range can be calculated. In the case that the resistivity of the surrounding rock is 800 ohm-meters and the resistivity of the target body is 5000 ohm-meters, the frequencies 0.5 Hz, 1 Hz, 2 Hz, 5 Hz, and 10 Hz are considered to be Main frequency selected to ensure,that the electromagnetic waves effectively penetrate to a depth of 365 meters and simultaneously maintain a sufficient signal-to-noise ratio. Based on this frequency set, the groups of physical quantities of the transmitting sources are created. For the low frequency range of 0.5 Hz and 1 Hz, the pulse width is set to 2 seconds, and the transmit current is 30 amperes to amplify the energy of the lower-lying signals. For the medium to high frequency range of 2 Hz to 10 Hz, the pulse width is reduced to 0.5 seconds, and the current is adjusted to 15 amperes to avoid overexcitation and simultaneously increase the resolution. This frequency-range-specific parameter configuration enables an effective balance between signal strength and resolution. The aforementioned physical quantities of the Transmitting sources are then mapped to a multi-stage time-domain sampling configuration. For the low-frequency range, the signal acquisition time window is extended to 85 seconds.with a sampling interval of 0.1 seconds to create a capture window of deep information. For the mid-frequency range, the window is set to 3 seconds, with a sampling interval of 0.02 seconds. For the high-frequency range, the window is compressed to 1 second, with a sampling interval of 0.005 seconds, to enable targeted capture of information at different depths. This multi-stage time-domain sampling,