A method, system and computer device for identifying a dammed aquifer
By combining data fusion from the micro-motion method, the controlled-source audio-frequency magnetotelluric method, and the equivalent reverse flux transient electromagnetic method, and dynamically adjusting the feature set weights, high-precision identification of dam aquifers was achieved. This solved the problems of limited detection depth and uneven data quality in traditional methods, and improved the reliability and rationality of the identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HOHAI UNIV
- Filing Date
- 2026-03-10
- Publication Date
- 2026-06-19
AI Technical Summary
Existing technologies are insufficient for accurately identifying the spatial distribution of aquifers beneath dams in dam engineering. Traditional methods are costly, inefficient, and pose a risk of seepage damage. Existing fusion methods have failed to achieve coordinated observation and fusion of high-resolution shallow layers, electrical structure of medium-deep layers, and water-bearing sensitivity.
By comprehensively utilizing data from the micro-motion method, the controlled-source audio-frequency magnetotelluric method, and the equivalent reverse flux transient electromagnetic method, and through a depth-constrained dynamic weighted fusion mechanism, the physical property characteristics of the strata are extracted to achieve high-precision identification of the aquifer of the dam.
It achieves high-precision and high-reliability identification of aquifers in dams, solves the problems of limited detection depth and uneven data quality in traditional methods, and improves the reliability and rationality of the fusion results.
Smart Images
Figure CN122239178A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method, system, and computer device for identifying aquifers in dams, belonging to the field of engineering geophysical exploration and data fusion technology. Background Technology
[0002] In the construction and operation of high embankment projects such as water conservancy projects, highways, and airports, accurate spatial detection of the aquifer beneath dams or roadbeds is a crucial prerequisite for assessing foundation stability, predicting seepage paths, and preventing seepage damage. High embankments bear enormous loads, and their foundations are often covered by complex Quaternary loose layers or weathered rock layers. Groundwater burial conditions are variable, easily forming hidden seepage channels. While traditional drilling methods are intuitive, they have limitations such as limited visibility, high cost, low efficiency, and the potential to disturb potential seepage paths.
[0003] Geophysical exploration methods have become an important means of investigation in this field due to their non-destructive and wide-coverage advantages. Currently, for the exploration of such shallow and medium-depth hydrogeological structures, the industry often uses a variety of single geophysical exploration methods, but all of them have significant technical bottlenecks.
[0004] In recent years, multi-source data fusion technology has provided a solution to the problem of multiple solutions in single geophysical methods. Existing technologies have attempted to fuse two geophysical methods (such as ground-penetrating radar and resistivity methods) for specific targets such as locating metal objects in tunnel rescue. However, such fusion schemes are mostly aimed at point or linear rigid targets, and the data registration and interpretation rules are simple (such as logical overlay). For the core engineering challenge of finely characterizing the three-dimensional aquifers with complex spatial distribution and intertwined physical property responses (velocity and electrical properties) beneath the dam foundation in dam engineering, existing technologies have significant shortcomings: First, there is a lack of a collaborative observation and fusion framework that can simultaneously consider shallow high resolution (micro-motion method), mid-to-deep electrical structure (controlled-source audio-frequency magnetotelluric method), and water-susceptibility (equivalent back flux transient electromagnetic method); second, existing fusion methods mostly remain at the level of qualitative or semi-qualitative image overlay, failing to establish a multi-physics quantitative coupling interpretation mechanism based on geotechnical parameters and hydrogeological models, resulting in significant uncertainty in determining the aquifer thickness, water-bearing capacity, and contact relationship with the overlying fill. Summary of the Invention
[0005] The purpose of this invention is to provide a method, system, and computer device for identifying aquifers in dams. By comprehensively utilizing the advantages of three geophysical exploration data methods—microtremor, controlled source audio-frequency magnetotelluric (CSAMT), and opposing coils transient electromagnetic (OCTEM)—and based on a depth-constrained dynamic weighted fusion mechanism, this invention achieves high-precision and high-reliability identification of the spatial distribution of aquifers in dams of dike projects.
[0006] To achieve the above objectives, the present invention is implemented using the following technical solution.
[0007] In a first aspect, the present invention provides a method for identifying aquifers in dams, comprising:
[0008] Three types of detection data were acquired for the target area of the dam; the three types of detection data include micro-motion method detection data, controlled source audio-frequency magnetotelluric method detection data, and equivalent reverse magnetic flux transient electromagnetic method detection data.
[0009] Based on the three types of detection data, preprocessing is performed to remove noise and outliers, resulting in a processed data matrix.
[0010] Based on the data matrix, continuous wavelet transform processing is applied to extract three types of feature sets related to the physical properties of the formation; wherein, the three types of feature sets include micro-motion method features, controlled source audio-frequency magnetotelluric method features, and equivalent reverse flux transient electromagnetic method features.
[0011] The three types of feature sets are weighted and fused based on a depth-constrained fusion function to obtain a comprehensive feature profile reflecting the distribution of aquifers; wherein, the depth-constrained fusion function can dynamically adjust the fusion weight of the corresponding feature sets according to the data validity of various detection methods at different depths;
[0012] The aquifer of the dam is identified based on the comprehensive feature profile.
[0013] Furthermore, the preprocessing includes at least one of denoising, outlier removal, power frequency interference removal, trend term correction, and data normalization.
[0014] Furthermore, the continuous wavelet transform uses Morlet wavelets as basis functions, and extracts the corresponding features by calculating the power spectrum of the wavelet coefficient matrix.
[0015] Furthermore, the micro-motion method features include dominant frequency features and energy features extracted from the micro-motion method detection data; the controlled-source audio-frequency magnetotelluric method features include attenuation gradient features and resistivity component features extracted from the controlled-source audio-frequency magnetotelluric method detection data; and the equivalent reverse flux transient electromagnetic method features include resistivity component features extracted from the equivalent reverse flux transient electromagnetic method detection data.
[0016] Furthermore, the depth-constrained fusion function can dynamically adjust the fusion weights of the corresponding feature sets based on the data validity of various detection methods at different depths, including:
[0017] Based on the effective detection depth range of the various detection methods and the data quality indicators in different depth intervals, the initial weights of different feature sets in each depth interval are determined; wherein, the data quality indicators include at least one of data resolution, signal-to-noise ratio, and aquifer response sensitivity;
[0018] For any given depth, the initial weights are dynamically adjusted based on the effective detection capabilities of the various detection methods within the current depth range, so that the feature set corresponding to the detection method with higher data validity dominates the fusion process within the current depth range; the data validity includes at least one of detection accuracy, resolution, anti-interference capability, or sensitivity to the physical properties of the aquifer.
[0019] Furthermore, the identification of the dam aquifer based on the comprehensive feature profile includes:
[0020] The comprehensive feature profile is visualized to obtain a cloud map of two-dimensional or two-dimensional fusion results;
[0021] Based on the cloud map and in conjunction with prior borehole information, the aquifer of the dam is identified;
[0022] The cloud map uses the distance of the survey line as the horizontal axis and the depth as the vertical axis, and uses color or contour lines to represent the spatial distribution of the fused feature values; each of the fused feature values is obtained by weighted summation of the three feature sets according to the fusion weights of the corresponding depth, and the comprehensive feature profile is composed of the multiple fused feature values.
[0023] Furthermore, after obtaining a comprehensive characteristic profile reflecting the distribution of aquifers, the method further includes:
[0024] Along the survey line distance direction, the comprehensive feature profile is sliced without overlap with a fixed width to obtain a slice dataset, which is used for subsequent refined anomaly detection.
[0025] Secondly, the present invention provides a dam aquifer identification system, comprising:
[0026] The data acquisition module is configured to acquire three types of detection data within the target area; wherein, the three types of detection data include micro-motion method detection data, controllable source audio-frequency magnetotelluric method detection data, and equivalent reverse flux transient electromagnetic method detection data;
[0027] The preprocessing module is configured to: perform preprocessing to remove noise and outliers based on the three types of detection data, and obtain the processed data matrix;
[0028] The feature extraction module is configured to: apply continuous wavelet transform processing to the data matrix to extract three types of feature sets related to the physical properties of the formation;
[0029] The depth-constrained fusion module is configured to: perform weighted fusion of the three types of feature sets based on the depth-constrained fusion function to obtain a comprehensive feature profile reflecting the aquifer distribution; wherein, the depth-constrained fusion function can dynamically adjust the fusion weights of the corresponding feature sets according to the data validity of various detection methods at different depths;
[0030] The aquifer identification module is configured to identify the aquifer of the dam based on the comprehensive feature profile.
[0031] Thirdly, the present invention provides a computer device, comprising:
[0032] Memory, used to store computer programs / instructions;
[0033] A processor for executing the computer program / instructions to implement the steps of the dam aquifer identification method according to any one of claims 1 to 8.
[0034] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:
[0035] This invention combines three geophysical methods: micro-motion, controlled-source audio-frequency magnetotellurics (CET), and equivalent reverse flux transient electromagnetic method (ECT). It comprehensively acquires data from these three methods. By leveraging the high resolution of micro-motion for stratigraphic interfaces and weak interlayers, the sensitivity of CET to mid-to-deep resistivity structures, and the high sensitivity of ECT to shallow aquifers, it achieves complementary advantages in vertical detection capabilities. Furthermore, this invention proposes a depth-constrained dynamic fusion strategy based on data validity. It dynamically allocates fusion weights to address the reliability differences of each method at different depths, solving the problems of limited detection depth for single methods and spatially uneven data quality between different methods, significantly improving the reliability and rationality of the fusion results. Moreover, this invention abandons simple image overlay and employs wavelet transform-based physical feature extraction and weighted fusion, preserving the physical essence related to aquifers while improving the signal-to-noise ratio and detail retention of the fused data. Attached Figure Description
[0036] Figure 1 The diagram shown is a flowchart of a method for identifying aquifers in dams provided by the present invention. Detailed Implementation
[0037] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.
[0038] The term "and / or" simply describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.
[0039] Example 1
[0040] See Figure 1 This embodiment proposes a method for identifying aquifers in dams. The specific implementation steps of the method are as follows:
[0041] Step S1: Obtain three types of detection data for the target area of the dam; wherein, the three types of detection data include micro-motion method detection data, controlled source audio-frequency magnetotelluric method detection data, and equivalent reverse magnetic flux transient electromagnetic method detection data;
[0042] Microtremor detection, controlled source audio-frequency magnetotelluric (CSAMT) detection, and opposing coils transient electromagnetic (OCTEM) detection are all specific technical types in geophysical exploration methods, and each of these three methods has its own technical focus.
[0043] Micromotion method: Based on natural field sources, this method extracts surface wave dispersion characteristics to invert the underground shear wave velocity structure, showing high sensitivity to stratigraphic interfaces and weak interlayers. The micromotion method acquires three-component ground vibration time series using a geophone, processes them to extract surface wave dispersion curves, and finally inverts the underground shear wave velocity structure. It mainly obtains data related to stratigraphic elastic parameters, used for classifying soil and rock layers, determining bedrock depth, and identifying concealed geological structures. It has strong resistance to electromagnetic interference and is suitable for shallow to medium-deep geological exploration in urban and complex environments. However, in dam engineering areas with strong interference from artificial vibration sources, its effective signal quality may decrease; simultaneously, this method is not directly sensitive to the resistivity characteristics of groundwater, making it difficult to directly distinguish between water-saturated zones and low-velocity weak clay layers, resulting in multiple solutions.
[0044] Controlled-source audio-frequency magnetotellurics (CET): This method uses artificially generated radio-frequency electromagnetic signals as a field source to measure the components of the ground's electric and magnetic fields. It calculates core data such as apparent resistivity and impedance phase, and then inverts these data to form a subsurface resistivity profile. It primarily acquires information on the electrical structure of deep strata, offering deep detection depth and high resolution. It is widely used for groundwater, geothermal, mineral, and deep geological structure exploration, and responds well to low-resistivity bodies such as aquifers and fault zones. However, its detection effectiveness is affected by the field source effect, resulting in severe data distortion in the near-field region (shallow areas), leading to insufficient resolution in shallow layers (typically corresponding to key interfaces at the base of dams). Furthermore, its volume effect is significant, blurring the lateral boundaries of thin layers or small aquifer structures.
[0045] The equivalent reverse flux transient electromagnetic method utilizes an ungrounded loop to transmit a primary field and measures the attenuation process of the secondary field generated by underground eddies. It exhibits high detection sensitivity for low-resistivity bodies (such as aquifers), directly acquiring the attenuation curve of the pure secondary field induced electromotive force generated by the underground medium. This eliminates primary field interference, enabling shallow, blind-zone-free detection. Finally, high-precision resistivity profile data is obtained through inversion, focusing on acquiring detailed electrical information in shallow areas. It is suitable for fine-grained engineering geological exploration of underground pipelines, karst, and mined-out areas. However, this method has limited detection depth in high-resistivity surrounding rock areas, and its response is a volumetric comprehensive effect, resulting in weak vertical stratification capabilities. It is difficult to accurately determine the burial depth of the top and bottom plates of aquifers, and its spatial positioning accuracy needs improvement.
[0046] Therefore, this embodiment addresses the core engineering challenge of the complex spatial distribution beneath the foundation of dam projects, which makes it impossible to accurately depict the three-dimensional morphology and distribution of underground aquifers. Instead of using a single physical detection method in the data acquisition process, it configures multiple physical detection methods to simultaneously acquire detection data from micro-motion detection, controlled-source audio-frequency magnetotelluric method, and equivalent reverse flux transient electromagnetic method.
[0047] Step S2: Based on the three types of detection data, preprocessing is performed to remove noise and outliers, resulting in a processed data matrix;
[0048] This embodiment simultaneously acquires three types of detection data. These three types of detection data differ in sampling frequency, unit, coordinate system, and time scale during acquisition. Furthermore, the raw data acquired during the acquisition process is affected by instrument noise, environmental interference, and system errors, which can lead to valid anomalies in the aquifer. Additionally, gaps, missed measurements, and bad spots may occur during acquisition. Therefore, to improve the accuracy and reliability of the three types of detection data, this embodiment first preprocesses the acquired data. Preprocessing includes noise reduction, outlier removal, power frequency interference removal, trend term correction, and data normalization. Preprocessing removes noise and outliers, resulting in a preprocessed data matrix.
[0049] Step S3: Apply continuous wavelet transform processing to the data matrix to extract three types of feature sets related to the physical properties of the formation; wherein, the three types of feature sets include micro-motion method features, controlled source audio-frequency magnetotelluric method features, and equivalent reverse flux transient electromagnetic method features;
[0050] In this embodiment, Morlet wavelet is preferentially selected as the basis function in continuous wavelet transform processing, and the corresponding features are extracted by calculating the power spectrum of the wavelet coefficient matrix.
[0051] The three types of feature sets related to the physical properties of the stratigraphy include micromotion method features, controlled-source audio-frequency magnetotelluric method features, and equivalent reverse flux transient electromagnetic method features.
[0052] The micro-motion method features include dominant frequency and energy features extracted from micro-motion method detection data; the controlled-source audio-frequency magnetotelluric method features include attenuation gradient and resistivity component features extracted from controlled-source audio-frequency magnetotelluric method detection data; and the equivalent reverse flux transient electromagnetic method features include resistivity component features extracted from equivalent reverse flux transient electromagnetic method detection data.
[0053] The dominant frequency characteristics and energy characteristics directly correspond to the elastic physical properties and water-bearing state properties of the formation. The dominant frequency characteristics are determined by the shear wave velocity of the underground formation, which is strongly correlated with the lithology, density, porosity and water content of the formation. The energy characteristics are the core physical basis for identifying aquifers, mainly used to reflect the damping characteristics and energy propagation efficiency of the formation medium. The damping characteristics of the formation medium are affected by the pore water filling of the aquifer, which leads to varying degrees of attenuation or enhancement of micro-motion energy.
[0054] Attenuation gradient characteristics and resistivity component characteristics are direct characterizations of the electrical physical properties of strata. Attenuation gradient characteristics correspond to the decay rate of electromagnetic response with time or frequency, reflecting the spatial variation rate of the electrical structure of underground strata. They are directly related to the lateral continuity of strata physical properties and the burial depth of interfaces, accurately identifying key physical property interfaces such as dam bases and the top and bottom plates of aquifers. Resistivity component characteristics are the most core electrical physical properties of strata, determined by strata lithology, porosity, water content, and mineralization. By identifying resistivity component characteristics, aquifer boundaries can be accurately characterized.
[0055] This embodiment obtains three types of feature sets related to the physical properties of the formation by applying continuous wavelet transform to the data matrix, including four features with clear physical meaning: dominant frequency features, energy features, gradient features, and resistivity component features.
[0056] Step S4: The three types of feature sets are weighted and fused based on the depth-constrained fusion function to obtain a comprehensive feature profile reflecting the distribution of the aquifer; wherein, the depth-constrained fusion function can dynamically adjust the fusion weight of the corresponding feature sets according to the data validity of various detection methods at different depths;
[0057] This embodiment constructs a depth-constrained fusion function, which includes a fusion weight function. The fusion weight function can dynamically adjust the weights of each feature source according to the data validity of different detection methods at different exploration depths. Specifically, it can adjust the weights of micro-motion method features, controllable source audio-frequency magnetotelluric method features, and equivalent reverse flux transient electromagnetic method features according to different exploration depths.
[0058] The following principles apply when dynamically adjusting the weights using the fusion weight function:
[0059] Based on the effective detection depth range of various detection methods and the data quality indicators in different depth intervals, the initial weights of different feature sets in each depth interval are determined; among them, the data quality indicators include at least one of data resolution, signal-to-noise ratio, and aquifer response sensitivity.
[0060] For any given depth, the initial weights are dynamically adjusted based on the effective detection capabilities of various detection methods within the current depth range, so that the feature set corresponding to the detection method with higher data validity dominates the fusion process within the current depth range; data validity includes at least one of detection accuracy, resolution, anti-interference capability, or sensitivity to the physical properties of the aquifer.
[0061] As a specific implementation case:
[0062] Shallow region: that is, the region with a depth less than the effective detection depth of the equivalent anti-magnetic flux transient electromagnetic method. In the shallow region, the weighting function is integrated to improve the weights of the micro-motion method characteristics and the equivalent anti-magnetic flux transient electromagnetic method characteristics.
[0063] The middle layer region is the region between the effective detection depth of the equivalent reverse flux transient electromagnetic method and the effective starting depth of the controllable source audio-frequency magnetotelluric method. In the middle layer region, the fusion weight function mainly relies on the micro-motion method features, increases the weight of the micro-motion method features, and ensures that the weight of the micro-motion method features is the largest among the three types of features.
[0064] Deep region: This refers to the region with a depth greater than the effective starting depth of the controlled-source audio-frequency magnetotelluric method. In the deep region, the weighting function is used to balance the micro-motion method features and the controlled-source audio-frequency magnetotelluric method features. By balancing and increasing the weights of the micro-motion method features and the controlled-source audio-frequency magnetotelluric method features, the sum of the weight ratios of the micro-motion method features and the controlled-source audio-frequency magnetotelluric method features is not less than a preset threshold, and the weight difference between the micro-motion method features and the controlled-source audio-frequency magnetotelluric method features is within a preset reasonable range. At the same time, the weight ratio of the equivalent reverse flux transient electromagnetic method features is reduced.
[0065] The fusion weights of the three feature sets extracted based on the fusion weight function are weighted and summed to obtain a unified comprehensive feature profile.
[0066] Step S5: Identify the aquifer of the dam based on the comprehensive feature profile.
[0067] This embodiment first visualizes the comprehensive feature profile to obtain a cloud map of the two-dimensional or two-dimensional fusion result. The cloud map uses the distance of the survey line as the horizontal axis and the depth as the vertical axis, and uses color or contour lines to represent the spatial distribution of the fused feature values. The fused feature values are obtained by weighting and summing the three types of feature sets obtained in step S4 according to the fusion weights of the corresponding depths, and the comprehensive feature profile is composed of multiple fused feature values.
[0068] In practice, staff identify the aquifer of the dam based on cloud maps and prior information from borehole drilling.
[0069] Furthermore, the comprehensive feature profile can be sliced non-overlappingly along the survey line at a fixed width, generating a series of slice datasets for subsequent refined anomaly detection, which can then be used by deep learning or pattern recognition models. This function transforms the detection results of long survey lines into standardized analysis units, greatly facilitating subsequent automated anomaly detection and refined aquifer interpretation based on artificial intelligence algorithms, and improving the efficiency and standardization of engineering interpretation.
[0070] Example 2
[0071] Based on Example 1, this example demonstrates the specific implementation of a method for identifying aquifers in dams for a high-fill canal embankment project with a survey line length of 6400 meters.
[0072] Step A1S1: Lay out a survey line approximately 6400 meters long along the axis of the canal embankment project. Use the micro-motion method with a point spacing of 100 meters to obtain data from 65 measuring points; use the controlled-source audio-frequency magnetotelluric method with a point spacing of 40 meters to obtain data from 161 measuring points; align the equivalent reverse flux transient electromagnetic method measurement with the controlled-source measuring points.
[0073] Step A1S2: Bandpass filtering (0.5-50 Hz) and wavelet denoising are applied to the micro-motion method data to preserve effective seismic wave signals; notch filtering is applied to the controlled-source audio-frequency magnetotelluric method data to remove 50Hz power frequency interference, and static migration correction is performed; mean filtering is applied to the equivalent back flux transient electromagnetic method data to remove impulse noise and trend term correction is performed. Finally, all data are interpolated and aligned to a unified grid of 161 measurement points (0-6400 meters) and 100 depth layers (0-500 meters), forming three data matrices of shape (161, 100, 256), where 256 represents the number of time / frequency sampling points.
[0074] Step A1S3: For each preprocessed data matrix, apply the Continuous Wavelet Transform (CWT) with Morlet wavelets as the basis function. For the micro-motion method data matrix, calculate the wavelet power spectrum at each measurement point-depth location, normalize the frequency corresponding to the scale with the highest power as the dominant frequency feature, which is related to the formation shear wave velocity; simultaneously calculate the sum of the power spectra as the energy feature, reflecting the formation response intensity. For the controlled-source audio-frequency magnetotelluric method data matrix, extract the attenuation curve of the wavelet coefficient energy with time (frequency) at each location, fit its attenuation rate using an exponential function to obtain the attenuation gradient feature; and invert a resistivity component feature based on this attenuation gradient. For the equivalent reverse flux transient electromagnetic method data matrix, mainly extract the resistivity component feature from its late attenuation characteristics of the wavelet transform. All extracted features are normalized to the [0, 1] interval and moderately smoothed to reduce noise.
[0075] Step A1S4: The system presets the maximum reliable depth of the fused detection methods. During fusion, the weight of each depth layer is dynamically calculated based on the maximum reliable depth of each detection method. Specifically, the weights are dynamically adjusted according to step S4 to obtain the comprehensive feature profile.
[0076] Step A1S5: Input the integrated feature profile (a 161×100 matrix) obtained in Step A1S4 into the visualization module. This module uses the survey line distance (0-6400 meters) as the abscissa and depth (0-500 meters) as the ordinate. Using "jet" color mapping, it renders the fused feature value at each survey point and depth as a color and draws contour lines, generating a high-precision two-dimensional fusion result profile cloud map. This cloud map clearly shows the continuous spatial variation of the integrated physical parameters of the subsurface medium, where low-value anomaly areas (usually represented in blue) may correspond to aquifers or weak interlayers. This cloud map is the core output of this method, used by geological engineers to perform final aquifer identification and delineation in conjunction with site data.
[0077] Step A1S6: To further automate the analysis, the slice generation module is invoked. This module uses a fixed width of 40 meters to slide and capture comprehensive feature data volumes without overlap, starting from the survey line's starting point. For example, the first slice covers 0-40 meters, the second covers 40-80 meters, and so on. Each slice is saved as an independent image file and data file (e.g., NumPy .npy format), forming a slice dataset containing approximately 160 samples. This dataset can be used to train deep learning models in the field of object detection, such as You Only Look Once (YOLO) and Real-Time Detection Transformer (RT-DETR), to achieve automatic identification and classification of aquifer anomalies.
[0078] Example 3
[0079] Based on the same inventive concept as Embodiment 1, this embodiment introduces a dam aquifer identification system, including:
[0080] The data acquisition module is configured to acquire three types of detection data within the target area; wherein, the three types of detection data include micro-motion method detection data, controllable source audio-frequency magnetotelluric method detection data, and equivalent reverse flux transient electromagnetic method detection data;
[0081] The preprocessing module is configured to: perform preprocessing to remove noise and outliers based on the three types of detection data, and obtain the processed data matrix;
[0082] The feature extraction module is configured to: apply continuous wavelet transform processing to the data matrix to extract three types of feature sets related to the physical properties of the formation;
[0083] The depth-constrained fusion module is configured to: perform weighted fusion of the three types of feature sets based on the depth-constrained fusion function to obtain a comprehensive feature profile reflecting the aquifer distribution; wherein, the depth-constrained fusion function can dynamically adjust the fusion weights of the corresponding feature sets according to the data validity of various detection methods at different depths;
[0084] The aquifer identification module is configured to identify the aquifer of the dam based on the comprehensive feature profile.
[0085] The specific functions of each module described above are explained in the relevant content of the method in Embodiment 1, and will not be repeated here.
[0086] Example 4
[0087] This embodiment describes a computer device, including:
[0088] Memory, used to store computer programs / instructions;
[0089] A processor is used to execute the computer program / instructions to implement the steps of the dam aquifer identification method described in Embodiment 1.
[0090] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0091] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0092] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0093] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0094] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.
Claims
1. A method for identifying a dammed aquifer, characterized in that, include: Three types of detection data were acquired for the target area of the dam; the three types of detection data include micro-motion method detection data, controlled source audio-frequency magnetotelluric method detection data, and equivalent reverse magnetic flux transient electromagnetic method detection data. Based on the three types of detection data, preprocessing is performed to remove noise and outliers, resulting in a processed data matrix. Based on the data matrix, continuous wavelet transform processing is applied to extract three types of feature sets related to the physical properties of the formation; wherein, the three types of feature sets include micro-motion method features, controlled source audio-frequency magnetotelluric method features, and equivalent reverse flux transient electromagnetic method features. The three types of feature sets are weighted and fused based on a depth-constrained fusion function to obtain a comprehensive feature profile reflecting the distribution of aquifers; wherein, the depth-constrained fusion function can dynamically adjust the fusion weight of the corresponding feature sets according to the data validity of various detection methods at different depths; The aquifer of the dam is identified based on the comprehensive feature profile.
2. The dammed aquifer identification method of claim 1, wherein, The preprocessing includes at least one of the following: noise reduction, outlier removal, power frequency interference removal, trend term correction, and data normalization.
3. The dammed aquifer identification method of claim 1, wherein, The continuous wavelet transform uses Morlet wavelets as basis functions, and extracts corresponding features by calculating the power spectrum of the wavelet coefficient matrix.
4. The dammed aquifer identification method of claim 1, wherein, The micro-motion method features include dominant frequency features and energy features extracted from the micro-motion method detection data; the controlled-source audio-frequency magnetotelluric method features include attenuation gradient features and resistivity component features extracted from the controlled-source audio-frequency magnetotelluric method detection data; the equivalent reverse flux transient electromagnetic method features include resistivity component features extracted from the equivalent reverse flux transient electromagnetic method detection data.
5. The dammed aquifer identification method of claim 1, wherein, The depth-constrained fusion function can dynamically adjust the fusion weights of the corresponding feature sets based on the data validity of various detection methods at different depths, including: Based on the effective detection depth range of the various detection methods and the data quality indicators in different depth intervals, the initial weights of different feature sets in each depth interval are determined; wherein, the data quality indicators include at least one of data resolution, signal-to-noise ratio, and aquifer response sensitivity; For any given depth, the initial weights are dynamically adjusted based on the effective detection capabilities of the various detection methods within the current depth range, so that the feature set corresponding to the detection method with higher data validity dominates the fusion process within the current depth range; the data validity includes at least one of detection accuracy, resolution, anti-interference capability, or sensitivity to the physical properties of the aquifer.
6. The dam aquifer identification method according to claim 5, wherein, The identification of the dam aquifer based on the comprehensive feature profile includes: The comprehensive feature profile is visualized to obtain a cloud map of two-dimensional or two-dimensional fusion results; Based on the cloud map and in conjunction with prior borehole information, the aquifer of the dam is identified; The cloud map uses the distance of the survey line as the horizontal axis and the depth as the vertical axis, and uses color or contour lines to represent the spatial distribution of the fused feature values; each of the fused feature values is obtained by weighted summation of the three feature sets according to the fusion weights of the corresponding depth, and the comprehensive feature profile is composed of the multiple fused feature values.
7. The dammed aquifer identification method of claim 1, wherein, After obtaining a comprehensive characteristic profile reflecting the distribution of aquifers, the method further includes: Along the survey line distance direction, the comprehensive feature profile is sliced without overlap with a fixed width to obtain a slice dataset, which is used for subsequent refined anomaly detection.
8. A dam aquifer identification system characterized by, include: The data acquisition module is configured to acquire three types of detection data for the target area of the dam; wherein, the three types of detection data include micro-motion method detection data, controlled source audio-frequency magnetotelluric method detection data, and equivalent reverse magnetic flux transient electromagnetic method detection data; The preprocessing module is configured to: perform preprocessing to remove noise and outliers based on the three types of detection data, and obtain the processed data matrix; The feature extraction module is configured to: apply continuous wavelet transform processing to the data matrix to extract three types of feature sets related to the physical properties of the formation; wherein, the three types of feature sets include micro-motion method features, controlled source audio-frequency magnetotelluric method features, and equivalent reverse flux transient electromagnetic method features; The depth-constrained fusion module is configured to: perform weighted fusion of the three types of feature sets based on the depth-constrained fusion function to obtain a comprehensive feature profile reflecting the aquifer distribution; wherein, the depth-constrained fusion function can dynamically adjust the fusion weights of the corresponding feature sets according to the data validity of various detection methods at different depths; The aquifer identification module is configured to identify the aquifer of the dam based on the comprehensive feature profile.
9. A computer apparatus, comprising: include: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of the dam aquifer identification method according to any one of claims 1 to 8.