A method and system for evaluating the hydrocarbon potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models.

Abstract

This invention discloses a method and system for evaluating the hydrocarbon-bearing properties of complex reservoirs based on time-domain electromagnetic inversion and rock physics model constraints. The method includes: acquiring the original time-domain electromagnetic signal, performing full-cycle superposition, positive and negative polarity superposition, and hybrid intelligent filtering to obtain a smooth signal with a high signal-to-noise ratio; performing electromagnetic inversion based on this signal to obtain electrical sensitive parameters such as resistivity and polarizability; establishing an evaluation model between electrical parameters and reservoir physical properties by combining rock physics test results; assigning weights to each single-parameter model to construct a multi-parameter fusion evaluation model, and verifying it using well logging data. Once the accuracy requirements are met, the hydrocarbon-bearing property evaluation result is output. This system is used to implement the above method. This invention reduces the ambiguity of hydrocarbon-bearing property interpretation and improves the accuracy, stability, and applicability of the evaluation by jointly constraining time-domain electromagnetic inversion and rock physics models.

Description

Technical Field

[0001] This invention relates to the field of energy exploration, specifically to a method and system for evaluating the hydrocarbon potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models. Background Technology

[0002] Complex reservoirs (such as tight sandstone, shale, and fracture-pore reservoirs) are highly heterogeneous, and seismic data has limited sensitivity to fluids, leading to significant uncertainties in hydrocarbon assessment. Electromagnetic methods are sensitive to formation fluid resistivity and have the potential to directly detect hydrocarbons; however, in complex reservoirs, the interpretation of a single electromagnetic response is highly ambiguous, resulting in insufficient quantitative assessment capabilities.

[0003] In existing technologies, several related solutions exist. For example, patent CN121410813A proposes a method for acquiring and processing time-frequency electromagnetic signals; CN121454636A involves constructing a geoelectric structure prediction model based on electromagnetic data. These technologies focus on electromagnetic data processing or geoelectric structure modeling, but lack rock physics constraints and quantitative evaluation chains for complex reservoirs with hydrocarbon content. To improve interpretation reliability, some solutions fuse multi-source data such as electromagnetic and seismic data, such as CN117233859A and CN114706124A. However, how to couple the fusion results with rock physics models to form a generalizable quantitative evaluation method still needs improvement. Furthermore, utilizing machine learning to assist in... oil and gas Predictive techniques (such as CN121763440A and CN114575834A) have advantages in pattern recognition, but under complex reservoir conditions, they lack constraints from rock physics mechanisms, resulting in weak generalization ability and interpretability. Cross-regional applications require retraining. Resistivity inversion and in-well electromagnetic detection technologies (such as CN105607147A, CN112327376A, and CN102121374A) provide means to obtain electrical parameters, but subsequent joint inversion and evaluation processes with parameters such as porosity and saturation are lacking. Other schemes include controlled-source electromagnetic detection (CN106646632A) and high-resistivity anomaly identification (CN109061741A), but in complex reservoirs, high-resistivity anomalies may be caused by multiple factors such as lithology, compaction, and dry layers, making it difficult to achieve stable quantitative evaluation of hydrocarbon content based solely on electrical anomalies.

[0004] In summary, existing technologies generally suffer from the following shortcomings: the relationship between electromagnetic response and hydrocarbon-bearing properties under complex reservoir conditions is complex, lacking a mechanism constraint and parameter mapping system centered on rock physics models; multi-source fusion or machine learning methods often focus on data-driven approaches, resulting in insufficient cross-regional generalization and interpretability; and after obtaining electrical parameters, there is a lack of a joint evaluation process and systematic implementation with key parameters such as porosity, saturation, permeability, and fluid properties. Therefore, it is necessary to propose a method and system for evaluating the hydrocarbon-bearing properties of complex reservoirs based on time-domain electromagnetic methods and rock physics models, in order to achieve quantitative conversion and systematic output of electromagnetic response to hydrocarbon-bearing property indicators, thereby improving the accuracy, stability, and scalability of the evaluation.

[0005] To address the aforementioned issues, the applicant proposes a method and system for evaluating the hydrocarbon potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for evaluating the hydrocarbon potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models, so as to solve the problems in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for evaluating the hydrocarbon-bearing capacity of complex reservoirs based on time-domain electromagnetic methods and rock physics models, comprising the following steps:

[0008] Step S1: Acquire the original time-domain electromagnetic signal, and sequentially perform full-cycle superposition, positive and negative polarity superposition, and hybrid intelligent filtering on the original time-domain electromagnetic signal to obtain smooth signal data with high signal-to-noise ratio;

[0009] Step S2: Perform electromagnetic inversion based on the smoothed signal data to extract electrically sensitive parameters such as resistivity and polarizability;

[0010] Step S3: Based on the results of rock physics testing and analysis, establish an evaluation model for electrical sensitive parameters and reservoir hydrocarbon content;

[0011] Step S4: Assign weights to each single-parameter response relationship model, construct a multi-parameter fusion hydrocarbon evaluation model, use well logging data to verify and iterate the initial model, and output the hydrocarbon evaluation results of the target complex reservoir based on the evaluation model parameters locked after verification.

[0012] Optionally, the hybrid intelligent filtering in step S1 further includes: sequentially performing notch filtering, median filtering, empirical mode decomposition, long short-term memory neural network prediction filtering, and recursive filtering on the signal data after the positive and negative polarities are superimposed.

[0013] Optionally, the construction of the single-parameter relationship model in step S3 further includes: using a modified generalized equivalent medium-induced polarization model to establish a quantitative physical mapping relationship between the rock micropore structure, multiphase medium, complex resistivity, and polarizability.

[0014] Optionally, the verification and iteration of the initial model using well logging data in step S4 specifically involves comparing and verifying the evaluation calculation results of the initial model with the actual well logging data of the work area. If the verification error does not meet the preset accuracy threshold, it is fed back to step S3 to readjust the single-parameter oil and gas relationship model or update the weight coefficients until the error meets the preset accuracy threshold.

[0015] Optionally, the oil and gas content evaluation results in step S4 include an oil and gas saturation distribution map and / or a sweet spot prediction map.

[0016] Optionally, the acquisition of the time-domain electromagnetic signal in step S1 may employ a multi-waveform and / or high-power transmission method.

[0017] Optionally, the method further includes applying the validated evaluation model to the combined seismic and electromagnetic interpretation of the work area to achieve combined seismic and electromagnetic detection of hydrocarbon potential.

[0018] A system for evaluating the hydrocarbon-bearing potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models, used to implement the methods described above, the system comprising:

[0019] Electromagnetic signal combination noise reduction module: used to receive the original time-domain electromagnetic signal and obtain smooth signal data by sequentially performing full-cycle superposition, positive and negative polarity superposition and hybrid intelligent filtering;

[0020] Electrical parameter inversion and extraction module: used to process the smoothed signal data and invert to obtain the electrical sensitive parameters of the formation, wherein the electrical sensitive parameters include at least resistivity and polarizability;

[0021] Rock physics modeling and analysis module: used to construct a model of the relationship between various electrical sensitive parameters and reservoir hydrocarbon-bearing physical properties based on rock physics experiments;

[0022] Multi-parameter verification and joint evaluation module: used to assign weights to each single-parameter relationship model, verify and iterate by comparing with actual well logging data, and finally output the comprehensive oil and gas content evaluation results of the target reservoir.

[0023] Optionally, the hybrid intelligent filtering in the electromagnetic signal combination noise reduction module specifically includes a notch filter unit, a median filter unit, an empirical mode decomposition unit, a long short-term memory neural network filter unit, and a recursive filter unit.

[0024] Optionally, the rock physics modeling and analysis module specifically uses a modified generalized equivalent medium-induced polarization model to construct the single-parameter relationship model.

[0025] Beneficial effects:

[0026] 1. The physical mechanism is well-defined, and the quantitative evaluation is highly accurate;

[0027] 2. Significantly reduces the ambiguity in interpreting the hydrocarbon content of complex reservoirs.

[0028] 3. The combination of seismic and electromagnetic fields has broad application prospects.

[0029] This invention overcomes the limitations of traditional electromagnetic methods that rely solely on resistivity anomalies for qualitative prediction. By introducing rock physics models (such as a modified generalized equivalent medium-induced polarization model) as quantitative constraints, it establishes a physical mapping relationship between the actual resistivity and polarizability of the formation and reservoir parameters such as porosity, hydrocarbon saturation, and permeability. This mechanism effectively isolates the interference of lithological and physical property changes on the electromagnetic response, decomposing a single electrical anomaly into a comprehensive contribution from multiple factors such as fluid, lithology, and pore structure. This significantly reduces the probability of misjudgment caused by densification or lithological changes, enabling accurate identification of fluid targets in complex geological settings.

[0030] 2. Improve the stability and physical interpretability of cross-work area forecasts.

[0031] Unlike purely data-driven machine learning methods, this invention deeply integrates the physical forward and inverse modeling mechanism of time-domain electromagnetic methods with the laws of rock physics, forming a dual-constraint evaluation system of "physical mechanism as the main driver, rock physics as the constraint, and data-driven assistance." This system ensures that the hydrocarbon prediction results not only statistically match the training data but also fundamentally conform to geological and geophysical laws, significantly enhancing the model's generalization ability and cross-regional migration ability under different complex reservoir conditions. Simultaneously, each intermediate parameter (resistivity, polarizability, porosity, saturation, etc.) has a clear physical meaning, and the evaluation process is transparent and traceable, overcoming the poor interpretability of "black box" models.

[0032] 3. Construct a complete closed-loop system from data collection to quantitative evaluation.

[0033] This invention establishes a complete technical chain encompassing "multi-waveform / high-power time-domain electromagnetic data acquisition—multi-level combined noise reduction processing—multi-parameter high-precision inversion—rock physical parameter conversion—quantitative evaluation of hydrocarbon potential," forming a modular and highly engineered evaluation system. This system can directly output quantitative results such as hydrocarbon saturation distribution, reservoir effectiveness indicators, and sweet spot prediction maps, providing direct and reliable decision-making basis for oil and gas exploration deployment, well location selection, and reserve assessment. Furthermore, the system incorporates a built-in well logging verification and feedback iteration mechanism, which can dynamically optimize model parameters based on actual data, ensuring that the evaluation results continuously converge to the true geological conditions.

[0034] 4. Compatible with multi-source data, expanding application scenarios

[0035] This invention, based on time-domain electromagnetic data, can conveniently integrate information from seismic and well logging data to achieve joint seismic and electromagnetic interpretation. Because the rock physics model provides a unified parameter conversion scale, multi-source data can work collaboratively within the same physical framework, further improving the reliability and resolution of hydrocarbon detection. It is applicable to various complex reservoir types, including tight sandstone, shale, deep-ultra-deep reservoirs, and fracture-pore reservoirs.

[0036] In summary, this invention effectively overcomes the shortcomings of existing technologies, such as difficulty in quantitative evaluation, weak generalization ability, and lack of systematic closed-loop process. It achieves high-precision, high-stability, and high-interpretability evaluation of the hydrocarbon content of complex reservoirs, and has significant technological advancements and broad industrial application value. Attached Figure Description

[0037] Figure 1 This is a flowchart of the time-domain electromagnetic data processing of this invention;

[0038] Figure 2 This is a flowchart of the well seismic-constrained time-domain electromagnetic three-dimensional inversion technology of the present invention;

[0039] Figure 3 This is a flowchart illustrating the implementation of the present invention of constructing a single-parameter relationship model of reservoir oil and gas based on rock physics testing and analysis. Detailed Implementation

[0040] The preferred embodiments of the present invention are described below with reference to the accompanying drawings to make the technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0041] This invention provides a method and system for evaluating the hydrocarbon potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models. This method is particularly suitable for hydrocarbon exploration under complex geological conditions such as tight sandstone, shale, deep and ultra-deep reservoirs, and fractured-porosity reservoirs. The core idea of ​​this invention is to obtain the true electrical response of the formation through high signal-to-noise ratio time-domain electromagnetic data acquisition and processing. Then, combined with rock physics models, such as the modified generalized equivalent medium-induced polarization model, a quantitative mapping relationship is established between electrical parameters and reservoir properties and hydrocarbon potential. Finally, through multi-parameter weighting and well logging verification iterations, stable and reliable hydrocarbon potential evaluation results are output.

[0042] In its implementation, this invention first performs the acquisition of the original time-domain electromagnetic signal. To adapt to the characteristics of complex reservoirs—deep burial and rapid signal attenuation—this invention prioritizes a multi-waveform, high-power transmission method. For example, a bipolar square wave is used as the transmission waveform, and a high-power transmitter supplies tens to hundreds of amperes of current to the underground. The fundamental frequency can be selected between 4s and 128s depending on the target depth. The receiving end uses non-polarized electrodes to acquire the electric field signal and employs a wideband, low-noise electromagnetic sensor, typically an induction coil, to record the secondary magnetic field attenuation curve after the primary field is turned off. During the acquisition process, a sufficiently dense measurement network needs to be deployed, typically with a measurement point spacing of 50-200m, which can be increased to 25m for key target areas. m Each measuring point records time-domain electromagnetic signals for multiple cycles (e.g., more than 100 cycles), providing a data foundation for subsequent superposition processing.

[0043] After obtaining the original time-domain electromagnetic signal, this invention proceeds to step S1, a multi-stage combined noise reduction process. Due to electromagnetic interference from complex surface environments (such as high-voltage lines, railways, towns, etc. near the work area) and the thermal noise of the instrument itself, the signal-to-noise ratio of the original signal is often low. Therefore, this invention designs a progressive noise reduction process. First, full-cycle superposition is performed: the signals from all cycles collected at each measuring point are aligned by time and then superimposed and averaged. Since the mean of random noise is zero, the improvement in the signal-to-noise ratio after superposition is proportional to the square root of the number of superpositions. Taking 64 cycles of superposition as an example, random noise can be suppressed to one-eighth of its original value. Next, positive and negative polarity superposition is performed: the time-domain electromagnetic method typically uses alternating positive and negative current waveforms, and the corresponding received signals also exhibit alternating positive and negative characteristics. Subtracting the response corresponding to the positive polarity transmission from the response corresponding to the negative polarity transmission and superimposing them effectively eliminates baseline drift and low-frequency background interference. This is because baseline drift and slow changes in the geomagnetic field manifest as common-mode components in the positive and negative responses, which are suppressed after subtraction.

[0044] After the initial processing described above, the main residual noise in the signal includes power frequency interference, pulse spikes, non-stationary noise, and broadband noise overlapping with the effective signal spectrum. This invention introduces a hybrid intelligent filtering module to address these complex noises. This module performs the following operations sequentially: First, notch filtering: a narrowband notch filter or adaptive notch filter is designed based on the precise frequencies of the power frequency and its harmonics (e.g., 50Hz, 100Hz, 150Hz, etc.) to filter out the power frequency interference component. Since notch filtering may cause minor damage to the target signal, subsequent steps will compensate for this. Second, median filtering: a sliding window (typically with a window length of 5 to 11 sampling points) is used to perform nonlinear median filtering on the signal sequence. All sampling points within the window are sorted by amplitude, and the median value is used to replace the center point, thereby effectively eliminating pulse-type spike noise caused by occasional events such as lightning and mechanical switches, while preserving the signal's edge characteristics relatively well. Finally, empirical mode decomposition: the processed signal is adaptively decomposed into a series of intrinsic mode functions. Empirical Mode Decomposition (EMD) does not require pre-defined basis functions and is entirely data-driven, capable of separating fluctuations at different time scales into different intrinsic mode functions (EMFs). Effective signals are typically concentrated in low- to mid-frequency EMFs, while high-frequency noise and random oscillations appear in higher-order EMFs. By analyzing the energy distribution and autocorrelation of each EMF, noise-dominated modes can be identified and removed, and the remaining EMFs can be reconstructed to obtain the denoised signal. Since EMD may leave some noise overlapping with the effective signal spectrum, this invention further introduces a Long Short-Term Memory (LSTM) neural network for deep filtering. A LSM model is pre-trained using high-quality time-domain electromagnetic data from the work area or similar geological conditions. This network can learn the dependencies and trend changes of time-series signals. The signal reconstructed from EMD is input into the trained LSM, and the network outputs a clean signal after nonlinear prediction and filtering, effectively removing residual noise in the same frequency band as the effective signal. Finally, to avoid phase delay and waveform distortion caused by the aforementioned filtering steps, this invention employs recursive filtering for correction. Recursive filtering recovers the original phase characteristics and amplitude ratio of the signal through reverse recursive operations, ensuring the authenticity of the output signal. After the above full-cycle superposition, positive and negative superposition, and hybrid intelligent filtering processes, a high signal-to-noise ratio, smooth, and waveform-fidelity secondary field attenuation curve signal is finally obtained, providing high-quality input for subsequent inversion. Figure 1 The time-domain electromagnetic data processing flow is demonstrated.

[0045] After obtaining the smoothed signal data, the present invention proceeds to step S2, where electromagnetic inversion is performed based on the smoothed signal data to extract the electrically sensitive parameters for reservoir fluid identification. Since the secondary field decay curve of the time-domain electromagnetic method is sensitive to the conductivity distribution of the subsurface medium, classic inversion methods include Occam inversion, Marquardt inversion, and conjugate gradient inversion. For complex reservoirs, simply inverting resistivity is often insufficient to distinguish between oil and gas-bearing layers and high-resistivity dry or tight layers. Therefore, the present invention not only inverts conventional resistivity parameters but also extracts parameters reflecting the induced polarization effect, including polarizability, time constant, and frequency correlation coefficient. The induced polarization effect is significant in formations containing clay or electronically conductive minerals (such as pyrite), while oil and gas themselves do not produce the induced polarization effect. However, oil and gas injection alters the distribution and connectivity of pore water, thereby changing the induced polarization response of the rock. By analyzing the late decay characteristics of the time-domain electromagnetic signal, the formation polarizability (equivalent to the charging rate in the frequency domain) can be inverted. In the specific inversion process, this invention employs a joint inversion strategy using resistivity and polarizability. First, a geoelectric grid is established based on the initial geological model, with resistivity and polarizability participating as independent parameters in the inversion. Forward modeling uses the finite-difference time-domain method or the finite-volume method to solve Maxwell's equations, obtaining the theoretical response at each measuring point. Then, the fitting difference between the theoretical response and the measured smoothed signal is calculated, along with the partial derivatives (sensitivity matrix) of the fitting difference relative to the model parameters. The model parameters are iteratively updated using the conjugate gradient method or the quasi-Newton method until the fitting difference converges to below a preset threshold. To improve the stability of the inversion, this invention introduces well-seismic constraint technology: if known well logging data or seismic interpretation results are available in the work area, they can be converted into prior information, constructing a priori model covariance matrix. This allows the inversion results to revert to the prior model in areas without data constraints, while finely characterizing anomalies in areas with data constraints. Figure 2 The technical process of well seismic-constrained time-domain electromagnetic three-dimensional inversion is demonstrated. After the inversion is completed, the resistivity and polarizability of each grid node are output, and the time constant τ and frequency correlation coefficient c are also output when necessary. These parameters constitute the input for subsequent rock physics modeling.

[0046] Next, in step S3, a single-parameter relationship model of reservoir hydrocarbon content is constructed based on rock physics testing and analysis. This invention argues that quantitative evaluation can only be achieved by linking the electrical parameters obtained from geophysical inversion with the reservoir's physical and fluid properties through a physical mechanism. Therefore, it is first necessary to obtain representative core samples from the target reservoir and conduct rock physics tests under simulated in-situ temperature and pressure environments in the laboratory (e.g., temperature 80-150℃, confining pressure 30-100MPa, with corresponding adjustments to pore pressure). The tests include: complex resistivity spectroscopy measurements (frequency range 10^-2 Hz to 10^4 Hz), measuring the resistivity and polarizability responses of the core samples under different water saturation, mineralization, and porosity conditions; nuclear magnetic resonance (NMR) measurements of pore structure; X-ray diffraction analysis of clay mineral types and contents; and scanning electron microscopy observation of pore micromorphology. Based on abundant experimental data, this invention preferably introduces a modified generalized equivalent medium-induced polarization model (MGEMTIP model) as the theoretical framework. This model treats rocks as a multiphase medium composed of conductive mineral particles, clay particles, insulating framework particles, and water-filled pores, taking into account particle shape (aspect ratio), size distribution, and double-layer polarization mechanism. The MGEMTIP model can quantitatively describe the characteristics of complex resistivity changing with frequency and establish functional relationships between resistivity, polarizability, porosity, water saturation, clay content, and permeability. Specifically, for a given reservoir section, key parameters in the model, such as the cementation index m, saturation index n, and aspect ratio distribution parameters of polarized particles, can be determined by fitting experimental data through nonlinear regression. Based on this, this invention establishes single-parameter hydrocarbon-bearing relationship models for resistivity and polarizability, respectively. For resistivity, the classic Archie formula gives the relationship between resistivity, porosity, and water saturation: ρ = a φ^{-m} S_w^{-n} ρ_w, where ρ is the formation resistivity, φ is the porosity, S_w is the water saturation, ρ_w is the formation water resistivity, a is the lithology coefficient, m is the cementation index, and n is the saturation index. For hydrocarbon reservoirs, the hydrocarbon saturation S_o = 1 - S_w. Therefore, given ρ and φ, S_w can be calculated, and thus S_o ​​can be obtained. However, in complex reservoirs, the exponents in the Archie formula often deviate from the classic values. This invention calibrates a, m, and n based on measured data. Regarding polarizability, this invention establishes an empirical or semi-empirical relationship between polarizability and oil and gas saturation, clay content, and porosity. For example, it adopts a power function form: m = A S_w^BC^D φ^E, where C is the clay content, and A, B, D, and E are fitting coefficients, or it adopts a theoretical relationship derived based on the MGEMTIP model. Figure 3This invention demonstrates the process of constructing single-parameter relationship models of reservoir hydrocarbon bearing capacity based on rock physics testing and analysis. Through this step, the invention obtains multiple single-parameter relationship models that can be independently used for hydrocarbon bearing capacity evaluation, each model mapping reservoir hydrocarbon information from different electrical perspectives.

[0047] Because individual electrical parameters have varying degrees of ambiguity in fluid identification—for example, high resistivity could be caused by hydrocarbons or by calcareous cementation or densification; high polarizability might indicate clay development rather than hydrocarbon accumulation—this invention proceeds to step S4 to address this issue through multi-parameter weighting, well logging verification, and joint hydrocarbon assessment. The specific operation is as follows: First, initial weights are assigned to each single-parameter hydrocarbon relationship model established in step S3. The weights can be determined based on the sensitivity and stability of each parameter in fluid identification, as well as its accuracy in well logging calibration within the work area. For example, resistivity is sensitive to changes in pore fluid but is greatly affected by lithology; polarizability is sensitive to clay and pore connectivity but has a weaker response in some pure sandstone reservoirs. Typically, the weight of the resistivity model can be set to 0.5, the polarizability model to 0.3, and the time constant model to 0.2, or it can be dynamically adjusted based on the signal-to-noise ratio and inversion reliability of the measured data. Then, a multi-parameter fusion-based initial model for comprehensive evaluation of hydrocarbon potential is constructed. The output of this model can be expressed as a comprehensive hydrocarbon potential evaluation model. oil and gasThe index G = Σ (w_i × F_i), where F_i is the hydrocarbon saturation or hydrocarbon probability output by the i-th single-parameter model, w_i is the corresponding weight, and Σ w_i = 1. Next, the initial model is applied to appraisal wells or exploration wells with existing logging data within the work area. The comprehensive hydrocarbon index at each depth point is calculated and compared with the measured hydrocarbon saturation interpreted from logging data (e.g., obtained through combined interpretation by resistivity logging, neutron density logging, and nuclear magnetic resonance logging). The root mean square error or mean absolute percentage error is calculated. It is determined whether the error meets the preset accuracy threshold, for example, a hydrocarbon saturation error of less than 5%. If the error exceeds the threshold, feedback adjustment is required: on the one hand, returning to step S3 to re-examine the rock physics relationship model, such as adjusting the exponent of the Archie formula or the form of the polarizability fitting formula; on the other hand, adjusting the weight coefficients, for example, if the resistivity model error is too large while the polarizability model error is small, appropriately increasing the polarizability weight and decreasing the resistivity weight. After adjustment, the comprehensive hydrocarbon index is recalculated and compared with well logging data. This iterative optimization continues until the verification error meets the accuracy requirements. Once the iteration converges, the final evaluation model parameters are locked, including the specific expressions of each single-parameter relationship model and the optimal weight coefficients. Finally, the validated evaluation model is applied to the temporal electromagnetic data and available seismic data of the entire exploration area to perform three-dimensional hydrocarbon prediction for the entire area. Specifically, for each inversion grid node in the entire area, the resistivity, polarizability, and other parameters obtained from the inversion are input, substituted into the locked single-parameter relationship model to calculate their respective hydrocarbon saturation, and then weighted and summed according to the optimal weights to obtain the comprehensive hydrocarbon saturation of each grid point. Simultaneously, a threshold can be set (e.g., hydrocarbon saturation greater than 50% and adjacent grids are continuous) to delineate sweet spots. The output results include a three-dimensional distribution of hydrocarbon saturation, a planar sweet spot prediction map, and hydrocarbon slices along the layers. These results can directly provide a basis for well location deployment and reserve assessment.

[0048] Based on the above method, this invention also provides a system for evaluating the hydrocarbon potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models. This system consists of several modules, and its modular design facilitates field deployment and engineering applications. The first module is an electromagnetic signal combination and noise reduction module. This module receives the raw time-domain electromagnetic signal from the acquisition device and integrates a full-cycle stacking unit, a positive and negative polarity stacking unit, and a hybrid intelligent filtering unit. The hybrid intelligent filtering unit further includes a notch filter, a median filter, an empirical mode decomposer, a long short-term memory neural network processor, and an inverse recursive filter. These components work in series in the aforementioned order, ultimately outputting smooth signal data. Users can adjust the stacking times, notch frequency points, intrinsic mode function screening thresholds for empirical mode decomposition, and the model path of the long short-term memory neural network through the module's parameter interface. The second module is an electrical parameter inversion and extraction module. This module receives the smoothed signal data from the first module and reads the geological prior information provided by the user (such as well logging curves and seismic stratigraphic interpretation). The first module implements a time-domain electromagnetic joint inversion algorithm based on the conjugate gradient method, capable of simultaneously outputting the resistivity and polarizability distributions. During the inversion process, users can set regularization parameters, the number of inversion iterations, and the convergence condition for the fitting difference. The third module is a rock physics modeling and analysis module, which has a built-in rock physics experimental database interface and model building function. Users can import core experimental data of the target reservoir into this module, which provides a variety of rock physics models (including Archie formula, Waxman-Smits model, MGEMTIP model, etc.) for selection. Through curve fitting or parameter calibration functions, the module can automatically determine the empirical coefficients in each single-parameter relationship model and display the model accuracy graphically. For work areas without experimental data, the module allows users to call preset model parameters from an empirical database of similar reservoirs, but it is recommended to perform subsequent calibration through well logging verification. The fourth module is a multi-parameter verification and joint evaluation module, which is the core decision-making part of the system. This module receives single-parameter relationship models from the third module, receives the inverted electrical parameter volume of the entire area from the second module, and reads well logging data from the work area database. The module first performs weight assignment and initial model construction. Then, it automatically extracts electrical parameter inversion values ​​along the well trajectory, calculates the hydrocarbon saturation of each single-parameter model and the integrated model, and compares it point-by-point with the well logging interpretation saturation to calculate the error. If the accuracy threshold is not met, the module will initiate a feedback iteration mechanism: the user can choose automatic or manual mode. In automatic mode, the module calls optimization algorithms (such as genetic algorithms or particle swarm optimization) to automatically search for the optimal weight combination, and fine-tunes the rock physics model parameters when necessary. In manual mode, the module provides an interactive interface for users to modify weights or model parameters according to their geological understanding. After the iteration is completed, the module locks the final parameters and generates hydrocarbon evaluation results for the entire area, outputting the results in common geophysical software formats (such as SEG-Y, VTK, etc.).The entire system can run on high-performance workstations or servers, supports multi-node parallel computing, and can process large-scale work area data with tens of thousands of measurement points.

[0049] The following specific embodiment further illustrates the implementation process of the present invention. In the exploration of a tight sandstone gas reservoir in a basin, the target reservoir has a burial depth of approximately 3000 m, an average porosity of 8%, and a permeability of less than 0.1 millidarcy, classifying it as a typical low-porosity, low-permeability tight reservoir. Conventional seismic methods are insufficient to effectively identify oil and gas-bearing areas. Using the method of the present invention, a time-domain electromagnetic survey network was first deployed, with a survey line spacing of 200 m and a survey point spacing of 50 m, covering an area of ​​100 square kilometers. The transmitted waveform was a bipolar trapezoidal wave with a fundamental frequency of 5 Hz and a transmitted current of 60 amperes. Magnetic induction coils were used for reception, and each survey point recorded 128 cycles of signal. After full-cycle superposition and positive-negative superposition processing, the signal-to-noise ratio of the collected raw data was significantly improved, but interference from 50 Hz power frequency and random pulses still existed. Next, a hybrid intelligent filtering method was applied: notch filtering removed 50Hz and its second and third harmonics; the median filter window length was set to 7, successfully eliminating several spikes caused by nearby vehicles; empirical mode decomposition decomposed the remaining signal into 9 intrinsic mode functions (IMFs). Analysis showed that the first and second IMFs were high-frequency noise, and the eighth and ninth IMFs were baseline drift components. After eliminating these IMFs, the signal was reconstructed; then, the reconstructed signal was input into a long short-term memory neural network pre-trained with a small amount of high signal-to-noise ratio data from the work area for smoothing; finally, inverse recursive filtering was used to compensate for the phase. The resulting secondary field attenuation curve was smooth and continuous, and the late-stage signal-to-noise ratio improved from 10dB to 35dB. Based on these data, a three-dimensional resistivity and polarizability joint inversion was performed, using the resistivity logging curves of three existing exploration wells in the work area as constraints. After 32 iterations of the inversion, the fitting error (RMS) decreased to within 2.5. The inversion results showed a high resistivity (80-120 Ω·m) anomaly zone in the central part of the work area, while the polarizability exhibited moderately low values ​​(5-10 mV / V). To quantitatively evaluate hydrocarbon potential, core samples from the target layer were tested in a rock physics laboratory, obtaining complex resistivity spectra of a series of samples with porosity ranging from 5% to 12% and water saturation from 20% to 100%. The experimental data were fitted using the MGEMTIP model, and the cementation index m = 1.85, saturation index n = 1.92, and lithology coefficient a = 1.0 were determined in the Archie formula; the relationship between polarizability and water saturation was fitted as m = 12.5 * S_w^1.3. Based on these parameters, resistivity and polarizability relationship models were constructed. The initial weights were set as follows: resistivity model weight 0.6, polarizability model weight 0.4. The above model was applied to well A, and the calculated overall hydrocarbon saturation was 62% using the inverted resistivity and polarizability. The actual well logging interpretation showed a hydrocarbon saturation of 58%, an error of 4%, which met the preset 5% threshold. To verify robustness, further verification was conducted at well B, where the error was 6%, slightly exceeding the threshold. Therefore, through feedback adjustments, the exponent in the polarizability model was adjusted from 1.3 to 1.45, and the weights were adjusted to resistivity 0.55 and polarizability 0.45.Recalculation showed that the oil and gas saturation error at well B decreased to 4%, while that at well A remained around 4%, meeting the requirements. After locking in the final model, the inversion parameters for the entire work area were evaluated, generating a three-dimensional distribution map of oil and gas saturation. The results showed that the oil and gas saturation corresponding to the high resistivity anomaly zone in the middle was generally between 55% and 75%, predicting a favorable sweet spot area of ​​approximately 12 square kilometers. Subsequently, a horizontal well was deployed within this sweet spot area, encountering a good gas layer, and industrial gas flow was obtained, verifying the effectiveness of the invention.

[0050] In summary, this invention, through high-precision time-domain electromagnetic data acquisition and processing, joint inversion of electrical parameters, rock physics-constrained modeling, and multi-parameter iterative verification, forms a complete technical solution that effectively solves the problems of multiple solutions and quantification in the evaluation of hydrocarbon-bearing properties in complex reservoirs, and has significant industrial practical value. The specific embodiments of this invention are not limited to the above examples. Those skilled in the art can make various modifications and variations based on the technical concept of this invention, and all such modifications and variations fall within the protection scope of this invention.

[0051] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the scope of the invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0052] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for evaluating the hydrocarbon-bearing potential of complex reservoirs based on time-domain electromagnetic methods and rock physics models, characterized in that, Includes the following steps: Step S1: Acquire the original time-domain electromagnetic signal, and sequentially perform full-cycle superposition, positive and negative polarity superposition, and hybrid intelligent filtering on the original time-domain electromagnetic signal to obtain smooth signal data with high signal-to-noise ratio; Step S2: Perform electromagnetic inversion based on the smoothed signal data to extract electrically sensitive parameters such as resistivity and polarizability; Step S3: Based on the results of rock physics testing and analysis, establish an evaluation model for electrical sensitive parameters and reservoir hydrocarbon content; Step S4: Assign weights to each single-parameter response relationship model, construct a multi-parameter fusion hydrocarbon evaluation model, use well logging data to verify and iterate the initial model, and output the hydrocarbon evaluation results of the target complex reservoir based on the evaluation model parameters locked after verification.

2. The method according to claim 1, characterized in that, The hybrid intelligent filtering in step S1 further includes: sequentially performing notch filtering, median filtering, empirical mode decomposition, long short-term memory neural network prediction filtering, and recursive filtering on the signal data after positive and negative superposition.

3. The method according to claim 1, characterized in that, The construction of the single-parameter relationship model in step S3 further includes: using a modified generalized equivalent medium-induced polarization model to establish a quantitative physical mapping relationship between the rock micropore structure, multiphase medium, complex resistivity, and polarizability.

4. The method according to claim 1, characterized in that, In step S4, the verification and iteration of the initial model using well logging data specifically involves comparing the evaluation calculation results of the initial model with the actual well logging data of the work area. If the verification error does not meet the preset accuracy threshold, it is fed back to step S3 to readjust the single-parameter oil and gas relationship model or update the weight coefficients until the error meets the preset accuracy threshold.

5. The method according to claim 1, characterized in that, The oil and gas content evaluation results in step S4 include an oil and gas saturation distribution map and / or a sweet spot prediction map.

6. The method according to claim 1, characterized in that, The acquisition of time-domain electromagnetic signals in step S1 adopts a multi-waveform and / or high-power transmission method.

7. The method according to claim 1, characterized in that, The method also includes applying the validated evaluation model to the combined seismic and electromagnetic interpretation of the work area to achieve combined seismic and electromagnetic detection of hydrocarbon potential.

8. A system for evaluating the hydrocarbon-bearing properties of complex reservoirs based on time-domain electromagnetic methods and rock physics models, used to implement the method as described in any one of claims 1 to 7, characterized in that, The system includes: Electromagnetic signal combination noise reduction module: Used to receive the original time-domain electromagnetic signal and obtain smooth signal data by sequentially performing full-cycle superposition, positive and negative superposition and hybrid intelligent filtering; Electrical parameter inversion and extraction module: used to process the smoothed signal data and invert to obtain the electrical sensitive parameters of the formation, wherein the electrical sensitive parameters include at least resistivity and polarizability; Rock physics modeling and analysis module: Based on rock physics test and analysis results, establish an evaluation model for electrical sensitive parameters and reservoir hydrocarbon content; Multi-parameter verification and joint evaluation module: Weights are assigned to each single-parameter response relationship model to construct a multi-parameter fusion hydrocarbon evaluation model. The initial model is verified and iterated using well logging data. Based on the evaluation model parameters locked after verification, the hydrocarbon evaluation results of the target complex reservoir are output.

9. The system according to claim 8, characterized in that, The hybrid intelligent filtering in the electromagnetic signal combination noise reduction module specifically includes a notch filter unit, a median filter unit, an empirical mode decomposition unit, a long short-term memory neural network filter unit, and a recursive filter unit.

10. The system according to claim 8, characterized in that, The rock physics modeling and analysis module specifically uses a modified generalized equivalent medium-induced polarization model to construct the single-parameter relationship model.

Images