Automatic underground panoramic survey panoramic imaging system for power transmission and transformation project

Abstract

The invention relates to the technical field of power transmission and transformation project surveying, and discloses a power transmission and transformation project automatic underground panoramic surveying panoramic imaging system which comprises a data acquisition module, a panoramic imaging module and a real-time monitoring module. The data acquisition module acquires multi-source detection data formed by rock and soil media, pipeline distribution and electromagnetic field data, wherein the rock and soil medium data comprises static stratum data and dynamic medium parameters. The panoramic imaging module analyzes multi-source data through a three-dimensional imaging engine, spatial consistency alignment is achieved through a data reconstruction unit, multi-modal data fusion and three-dimensional space recombination are carried out on a wave field layering layer, a feature matching layer, a correlation analysis layer and an imaging layer of a model fusion unit, and an underground space three-dimensional panoramic imaging model is constructed. The imaging layer is provided with a multi-view interaction interface and supports a user operation model. And the real-time monitoring module marks an abnormal electromagnetic field area and triggers risk early warning. According to the system, multi-source data fusion and three-dimensional panoramic imaging are realized, and the surveying precision and safety are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission and transformation engineering survey, and in particular to a panoramic imaging system for automated underground panoramic survey of power transmission and transformation engineering. Background Art

[0002] During the construction of power transmission and transformation projects, underground space environmental surveys are a crucial and fundamental step. Underground space encompasses a complex geological structure, the distribution of various pipelines, and the electromagnetic environment, among other factors. These interrelated and mutually influential factors have a direct and significant impact on the planning, design, construction, and safe operation of power transmission and transformation projects.

[0003] Traditional underground survey methods have numerous limitations. For one thing, a single detection method often only captures a single or limited set of information. For example, geological radar primarily detects the distribution of underground media and pipelines, while electromagnetic methods primarily study the distribution and variations of electromagnetic fields. This single-mindedness results in a lack of comprehensiveness and integrativeness in survey results, making it difficult to fully reflect the true state of the underground space. This can easily lead to missing or incomplete information, which in turn impacts the scientific nature and accuracy of engineering decisions.

[0004] On the other hand, traditional survey methods suffer from inefficiency and lack precision in data processing and analysis. Different types of exploration data are typically processed independently, lacking effective data fusion and collaborative analysis mechanisms. This makes it difficult to establish organic connections between various data types and fully exploit the rich information contained within them. For example, the spatial correlations and interactions between geological data, pipeline data, and electromagnetic field data are difficult to accurately identify and analyze, making it impossible to comprehensively image and model the underground space.

[0005] Furthermore, traditional surveying technologies also suffer from significant limitations in real-time and dynamic monitoring. During the construction and operation of power transmission and transformation projects, the underground environment can undergo dynamic changes, such as subtle shifts in the ground, pipeline leaks, or abnormal fluctuations in the electromagnetic field. Traditional methods often fail to capture these changes in a timely manner, making it difficult to identify potential safety hazards, and thus failing to effectively safeguard the safe operation of the project.

[0006] As power transmission and transformation projects become larger, more complex, and more intelligent, higher demands are placed on underground survey technology. An automated survey system capable of multi-source data fusion, panoramic imaging, and real-time monitoring is urgently needed to improve survey efficiency, accuracy, and comprehensiveness, meeting the demands of modern power transmission and transformation project construction and operation. Existing technologies are struggling to adapt to these new challenges. Therefore, developing an advanced, automated underground panoramic survey and imaging system for power transmission and transformation projects is of great practical significance and urgency. Summary of the Invention

[0007] The purpose of the present invention is to provide an automated underground panoramic survey and imaging system for power transmission and transformation projects to solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above-mentioned object, the present invention provides the following technical solution: a panoramic imaging system for automated underground panoramic survey of power transmission and transformation projects, the system comprising: A data acquisition module is configured to collect underground multi-source detection data from a target area. The multi-source detection data includes a first detection data set corresponding to geotechnical medium data, a second detection data set corresponding to pipeline distribution data, and a third detection data set corresponding to electromagnetic field data. The geotechnical medium data includes static stratum data retrieved from a geological database and dynamic medium parameters acquired through a distributed detection device. A panoramic imaging module, configured to input the multi-source detection data into a three-dimensional imaging engine for analysis; A three-dimensional panoramic imaging model of the underground space is constructed according to the output results of the three-dimensional imaging engine; the three-dimensional imaging engine includes a data reconstruction unit and a model fusion unit, wherein the data reconstruction unit is used to align the multi-source detection data for spatial consistency, and the model fusion unit performs multimodal data fusion based on a preset underground structure template and historical survey data; the model fusion unit includes a wave field synthesis layer, a feature matching layer, a correlation analysis layer and an imaging layer connected in sequence, the wave field synthesis layer is used to perform frequency domain conversion processing on each data set in the multi-source detection data to generate a wave field feature map; the correlation analysis layer is used to model the spatial correlation between the wave field feature maps corresponding to the multi-source detection data to generate a multi-dimensional topological relationship; the imaging layer is used to perform three-dimensional spatial reorganization based on the multi-dimensional topological relationship and the wave field feature map to generate a three-dimensional panoramic imaging model.

[0009] Preferably, the step of modeling the spatial correlation between the wavefield characteristic spectra corresponding to the multi-source detection data to generate a multi-dimensional topological relationship includes: A signal feature extraction algorithm is used to identify abnormal response areas in the wave field characteristic map, and a feature coding sequence corresponding to each type of detection data is generated based on the spectrum characteristics of each abnormal response area; Calculating a correlation coefficient between abnormal response areas having the same frequency band identifier in feature coding sequences corresponding to any two types of detection data, and determining a multi-dimensional topological relationship between the arbitrary two types of detection data based on the correlation coefficient; calculating the correlation coefficient between abnormal response areas having the same frequency band identifier in feature coding sequences corresponding to any two types of detection data, including: When the number of abnormal responses in the characteristic coding sequences corresponding to any two types of detection data is inconsistent, frequency band expansion compensation is performed based on the distribution characteristics of the highest energy frequency band in the one with the smaller number of abnormal responses, and the correlation coefficient between abnormal response areas with the same frequency band identifier is calculated based on the compensated sequence.

[0010] Preferably, the data reconstruction unit is specifically used to: performing time-frequency synchronization processing on the first detection dataset, the second detection dataset, and the third detection dataset according to a preset reference coordinate system to obtain an aligned first detection dataset, an aligned second detection dataset, and an aligned third detection dataset; An adaptive filtering method is used to suppress the interference signals in the aligned first detection data set and the aligned second detection data set, and a phase compensation method is used to correct the waveform distortion in the aligned third detection data set to obtain a first optimized detection data set, a second optimized detection data set, and a third optimized detection data set.

[0011] Preferably, the first optimized detection data set includes processed static formation data and processed dynamic medium parameters, and the data reconstruction unit is further used to: analyzing energy attenuation curves of the processed static formation data and the processed dynamic medium parameters within a historical survey period; Calculating the equivalent impedance parameters of the processed dynamic medium parameters in the specified area based on the energy attenuation curve and the wave velocity parameters of the processed static formation data in the specified area; A comprehensive medium data set is generated according to the processed dynamic medium parameters and their equivalent impedance parameters, and the detection data corresponding to the comprehensive medium data set is used as the first optimized detection data set.

[0012] Preferably, the wave field synthesis layer includes a spectrum analysis unit and an energy mapping unit; The spectrum analysis unit is used to perform multi-scale time-frequency decomposition on each type of detection data set contained in the multi-source detection data, so as to extract corresponding frequency domain feature components from each type of detection data set; The energy mapping unit is used to perform phase alignment and superposition on the frequency domain characteristic components extracted from each type of detection data set and the original detection data set to generate a wavefield characteristic spectrum.

[0013] Preferably, the wave field synthesis layer further comprises a noise suppression unit, and the noise suppression unit is used to perform blind source separation and denoising processing on the wave field characteristic map; The modeling of the spatial correlation between the wavefield characteristic maps corresponding to the multi-source detection data includes: performing cross-validation analysis on the denoised wavefield characteristic maps corresponding to various types of detection data to generate the multi-dimensional topological relationship.

[0014] Preferably, the imaging layer includes a multi-view interactive interface, which supports the user to switch sections, filter signal strength and adjust resolution of the three-dimensional panoramic imaging model through control instructions; The multimodal data fusion based on the preset underground structure template and historical survey data includes: dynamically optimizing the physical property parameters of the underground structure template through an iterative inversion algorithm to adapt to the update of the multi-dimensional topological relationship; The three-dimensional spatial reorganization based on the multi-dimensional topological relationship and the wavefield characteristic map includes: performing spatial interpolation calculation on the multi-dimensional topological relationship according to preset imaging weights, and performing vector superposition with the wavefield characteristic map.

[0015] Preferably, the feature matching layer includes a dynamic compensation unit for performing amplitude compensation on the wave field characteristic map according to the medium density gradient of the target area; the operations performed by the dynamic compensation unit include: establishing a compensation model of the medium absorption coefficient based on the electromagnetic field intensity data of the third detection data set, and updating the energy distribution of each frequency band in the wave field characteristic map through a back propagation algorithm.

[0016] Preferably, the imaging layer also includes an error feedback unit for calculating the residual value between the three-dimensional panoramic imaging model and the historical survey data in real time during the vector superposition process; when the residual value exceeds a preset threshold, the data reconstruction unit is triggered to re-execute the time-frequency synchronization processing and update the spatial interpolation parameters of the multi-dimensional topological relationship.

[0017] Preferably, the system also includes a real-time monitoring module for marking areas in the three-dimensional panoramic imaging model where the electromagnetic field intensity exceeds a safety threshold; the real-time monitoring module is configured with an early warning trigger mechanism, which generates a risk early warning signal with coordinate positioning when abnormal field intensity is detected in the spatial overlapping area of ​​the pipeline distribution data and the electromagnetic field data.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The system's data acquisition module collects multi-source underground exploration data from the target area, including a first exploration dataset corresponding to geotechnical media data, a second exploration dataset corresponding to pipeline distribution data, and a third exploration dataset corresponding to electromagnetic field data. The geotechnical media data integrates static stratum data with dynamic media parameters. This multi-source data acquisition model transcends the limitations of traditional single-source exploration methods, enabling comprehensive capture of multi-dimensional information such as underground geological structure, pipeline distribution, and electromagnetic field environment. This ensures that survey results fully reflect the true state of the underground space, providing a rich and comprehensive data foundation for subsequent imaging and analysis, effectively addressing the information gaps inherent in traditional methods.

[0019] The three-dimensional imaging engine in the panoramic imaging module significantly improves data processing accuracy and imaging quality through the innovative design of the data reconstruction unit and model fusion unit. The data reconstruction unit performs spatial consistency alignment on multi-source detection data, and eliminates the temporal and spatial differences and noise interference between different data sets through operations such as time-frequency synchronization processing, interference signal suppression, and waveform distortion correction, ensuring data accuracy and consistency. For example, adaptive filtering is used to suppress interference signals on the first and second detection data sets, and phase compensation is performed on the third detection data set to correct waveform distortion, allowing various types of data to be fused and analyzed under a unified benchmark. At the same time, by analyzing the energy attenuation curves of static formation data and dynamic medium parameters, equivalent impedance parameters are calculated and a comprehensive medium data set is generated, further improving the reliability and integrity of geotechnical medium data.

[0020] The wavefield synthesis layer of the model fusion unit uses a spectrum analysis unit and an energy mapping unit to perform multi-scale time-frequency decomposition and phase alignment on multi-source detection data, generating a wavefield signature map that accurately reflects the data characteristics. The noise suppression unit also uses blind source separation and denoising to further improve the map quality. The dynamic compensation unit of the feature matching layer compensates the amplitude of the wavefield signature map based on the medium density gradient, establishes a medium absorption coefficient compensation model, and updates the energy distribution through a backpropagation algorithm. This effectively mitigates the impact of different media on detection signal propagation and improves feature matching accuracy. The correlation analysis layer uses a signal feature extraction algorithm and correlation coefficient calculation to fully exploit the spatial correlations between multi-source data and generate multi-dimensional topological relationships. Even when the number of anomaly responses is inconsistent, frequency band expansion compensation and other methods ensure the accuracy of correlation calculations, thereby establishing an organic connection between the various data types. The imaging layer performs three-dimensional spatial reconstruction based on the multi-dimensional topological relationships and wavefield signature maps. Using preset imaging weights, spatial interpolation and vector overlay are performed to generate a high-precision three-dimensional panoramic imaging model. A multi-view interactive interface allows users to switch sections, filter signal strength, and adjust resolution to meet analysis needs in different scenarios. The error feedback unit monitors the residual value in real time during the vector superposition process. When the residual value exceeds the preset threshold, it triggers the data reconstruction unit to reprocess and update the interpolation parameters to ensure the accuracy and reliability of the imaging model.

[0021] The system also features a real-time monitoring module that identifies areas within the 3D panoramic imaging model where electromagnetic field strength exceeds safety thresholds. It also incorporates a warning trigger mechanism. When abnormal field strength is detected in areas where pipeline distribution data and electromagnetic field data overlap, a risk warning signal with coordinate location is generated. This enables real-time dynamic monitoring of the underground environment and early warning of potential safety hazards, providing strong support for the safe operation of power transmission and transformation projects and significantly improving their safety and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a working principle diagram of the automatic underground panoramic survey and panoramic imaging system for power transmission and transformation projects according to the present invention; Figure 2 Generate flow charts for modeling spatial correlation of wavefield feature maps and multi-dimensional topological relationships; Figure 3 Flowcharts generated for the comprehensive media dataset; Figure 4 Flowchart of multimodal data fusion of imaging layer. DETAILED DESCRIPTION

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0024] See also Figure 1-Figure 4 The present invention relates to a panoramic imaging system for underground panoramic survey of power transmission and transformation engineering automation, and its specific implementation scheme is as follows: The data acquisition module is used to collect underground multi-source detection data in the target area. The multi-source detection data includes a first detection data set corresponding to geotechnical medium data, a second detection data set corresponding to pipeline distribution data, and a third detection data set corresponding to electromagnetic field data. The geotechnical medium data includes static stratum data retrieved from the geological database and dynamic medium parameters obtained through distributed detection devices. The distributed detection devices can be deployed at multiple detection points on the surface of the target area, and dynamic medium parameters such as soil resistivity and dielectric constant are collected in real time through sensor arrays. These parameters are combined with static stratum data such as historical stratum structure and lithology distribution stored in the geological database to form a first detection data set. The second detection data set is obtained through pipeline detection radar or electromagnetic induction equipment, recording information such as the spatial location, burial depth, and diameter of pipelines such as power, communication, and water supply and drainage in the target area. The third detection data set is collected using electromagnetic survey equipment and includes data such as the underground electromagnetic field intensity, distribution characteristics, and frequency characteristics in the target area.

[0025] The panoramic imaging module inputs multi-source detection data into the 3D imaging engine for analysis and constructs a 3D panoramic imaging model of the underground space based on the engine's output. The 3D imaging engine includes a data reconstruction unit and a model fusion unit. The data reconstruction unit is responsible for spatially aligning the multi-source detection data. Specifically, the data reconstruction unit first performs time-frequency synchronization on the first, second, and third detection datasets according to a preset reference coordinate system (such as the National Geodetic Coordinate System). Through timestamp calibration and frequency domain matching, the consistency of the data from different sources in spatial coordinates and time dimensions is ensured, resulting in the three aligned detection datasets. The model fusion unit performs multimodal data fusion based on a preset underground structure template and historical survey data. This fusion comprises a sequentially connected wavefield synthesis layer, a feature matching layer, a correlation analysis layer, and an imaging layer. The wavefield synthesis layer performs frequency domain conversion on each dataset in the multi-source detection data to generate wavefield feature maps. The correlation analysis layer models the spatial correlations between the wavefield feature maps corresponding to the multi-source detection data, generating multidimensional topological relationships. The imaging layer performs three-dimensional spatial reconstruction based on multi-dimensional topological relationships and wave field characteristic maps to generate a three-dimensional panoramic imaging model.

[0026] The present invention will be further described below in conjunction with Examples 1 to 5: Example 1, This embodiment focuses on the optimization processing of the data reconstruction unit, specifically including time-frequency synchronization processing, interference signal suppression and waveform correction, static and dynamic data fusion, etc. Each step is closely linked to achieve spatial consistency alignment and accuracy improvement of multi-source detection data, as follows: The data reconstruction unit first performs time-frequency synchronization on the first detection dataset (geotechnical data), the second detection dataset (pipeline distribution data), and the third detection dataset (electromagnetic field data). This process uses a pre-defined reference coordinate system (such as the National Geodetic Coordinate System) as a unified reference framework, eliminating spatiotemporal misalignment between the different datasets through timestamp calibration and frequency domain matching. Dynamic medium parameters (such as soil resistivity and dielectric constant) collected by distributed detection devices are calibrated in real time using the Global Navigation Satellite System (GNSS) to ensure that the position of each detection point is consistent with the reference coordinate system. Simultaneously, the clocks of each device are calibrated using a time synchronization protocol (such as the Network Time Protocol (NTP)) to align the sampling times of the dynamic medium parameters with the time windows of the static stratigraphic data in the geological database. The second detection dataset, acquired by the pipeline detection radar, is aligned with the reference coordinate system using the radar's built-in positioning module. The data is then framed according to the detection time series, with each frame representing a spatial scan result at a specific time point. A temporal interpolation algorithm is used to unify the sampling frequency to the same time interval as the first and third datasets (e.g., 100 sampling points per second). The time-frequency synchronization of electromagnetic field data requires consideration of both the time and frequency dimensions: in terms of time, the hardware clock of the electromagnetic survey equipment is synchronized with the reference clock to ensure the accuracy of the timestamp of the electromagnetic field intensity data; in terms of frequency, the collected broadband signal is frequency resampled to unify the frequency resolution of different devices to a preset value (such as 1Hz) for subsequent frequency domain analysis.

[0027] After completing time-frequency synchronization, the data reconstruction unit performs interference signal suppression on the aligned first and second detection data sets. An adaptive filtering algorithm is used to remove background noise, such as natural electric field noise and industrial electromagnetic interference, from the static stratigraphic data (such as historical stratigraphic structure and lithologic distribution) and dynamic medium parameters in the first detection data set. This adaptive filtering algorithm uses the minimum mean square error (LMS) criterion and recursively updates filter coefficients to adapt to changes in signal statistical characteristics. For example, if a 50Hz power frequency interference signal is present in the soil resistivity signal of the dynamic medium parameter, the adaptive notch filter automatically detects this interference frequency and generates a corresponding notch, attenuating the signal energy in this frequency band while preserving the valid signal components. For the pipeline distribution data in the second detection data set, since radar echoes are susceptible to scattering interference from surface metal objects (such as manhole covers and rebar), an adaptive filtering method based on independent component analysis (ICA) is used to decompose the radar echoes into independent signal components, separating the valid components containing pipeline characteristics from the noise components, thereby suppressing clutter interference.

[0028] For the third detection dataset (electromagnetic field data), the data reconstruction unit uses phase compensation to correct waveform distortion. When electromagnetic field signals propagate underground, the uneven dielectric constant and magnetic permeability of the rock and soil medium can cause phase delays in signals of different frequency bands, resulting in waveform distortion. The specific steps for phase compensation are as follows: First, the electromagnetic field data is decomposed in the frequency domain to obtain the phase spectrum for each frequency band (e.g., low frequency 0-100Hz, medium frequency 100-1000Hz, and high frequency above 1000Hz). Then, the phase difference of the signal in each frequency band relative to a reference frequency band (e.g., medium frequency 500Hz) is calculated to generate a phase correction function. Finally, the phase of the signal in each frequency band is corrected through frequency domain multiplication to ensure that the corrected waveform remains consistent on the time axis. For example, if the high-frequency signal has a phase lag of π / 2 relative to the reference frequency band, the phase compensation algorithm will advance the phase of the high-frequency signal by π / 2, eliminating the waveform distortion caused by the phase difference.

[0029] After completing the above processing, the first optimized detection data set contains the processed static formation data and dynamic medium parameters. The data reconstruction unit further performs a fusion analysis on the two types of data: extracting the static formation data and dynamic medium parameters within the historical survey period, calculating the signal energy attenuation value at different depths, and drawing an energy attenuation curve. This curve reflects the law of signal energy attenuation with increasing formation depth, which can be fitted by an exponential function as follows: ,in is the initial value of surface energy, is the attenuation coefficient, For example, the static formation data of a certain area shows that the signal energy attenuates by 10% for every 1 meter increase in depth. The attenuation coefficient is .

[0030] According to the wave velocity parameters (such as P-wave velocity) in the specified area of ​​the processed static formation data , shear wave velocity ) Calculate the equivalent impedance parameters of dynamic medium parameters. Equivalent impedance Defined as the density of the medium and wave speed The product of ), which is used to characterize the medium's ability to hinder wave propagation. For static stratum data, wave velocity parameters can be obtained from geological exploration reports. For example, if the longitudinal wave velocity of a clay layer is 1500m / s and the shear wave velocity is 800m / s, combined with the density data of the layer (such as 1800kg / m³), the longitudinal wave impedance can be calculated as , the shear wave impedance is For dynamic medium parameters (such as the dielectric constant acquired in real time ), through the empirical relationship between dielectric constant and density (such as ,in The density is estimated by combining the wave velocity parameters of the static formation data with the wave velocity parameters to calculate the equivalent impedance.

[0031] The data reconstruction unit fuses the processed dynamic medium parameters with their equivalent impedance parameters to generate a comprehensive medium data set. The specific method is: to attach the corresponding equivalent impedance value to each dynamic medium parameter sample, forming a data set containing multi-dimensional features. For example, the dynamic medium parameters of a detection point are dielectric constant 20 and conductivity 0.01S / m. The equivalent impedance calculated by the above method is , then the sample is represented in the comprehensive medium data set as [dielectric constant = 20, conductivity = 0.01, equivalent impedance = 3.0×10 6 As the first optimized detection dataset, this comprehensive dataset not only retains real-time dynamic parameters but also integrates impedance characteristics derived from static data, improving the spatial consistency and physical meaning coherence of geotechnical medium data, and providing a more accurate input basis for subsequent three-dimensional imaging.

[0032] Throughout the entire processing flow, the data reconstruction unit achieves spatial alignment and noise suppression of multi-source detection data through a multi-step process involving the establishment of a reference coordinate system, time-frequency synchronization, adaptive filtering, phase compensation, and dynamic and static data fusion. This overcomes the challenges of fusing different data types due to inconsistent spatiotemporal references, noise interference, and differences in media properties, laying the data foundation for the construction of the 3D model of the panoramic imaging module. Each processing step is based on signal processing theory and geophysical principles. By combining mathematical algorithms with physical models, the scientific and accurate data processing is ensured, avoiding the introduction of subjective assumptions or fictitious descriptions.

[0033] Example 2, This embodiment focuses on the structure and processing flow of the wave field synthesis layer, covering the specific implementation of the spectrum analysis unit, energy mapping unit, and noise suppression unit. Through multi-scale time-frequency decomposition, phase alignment and superposition, and blind source separation and denoising operations, a high signal-to-noise ratio wave field feature map is generated and a multi-dimensional topological relationship is established. The details are as follows: The spectral analysis unit of the wavefield synthesis layer performs multi-scale time-frequency decomposition on each dataset in the multi-source detection data to extract frequency-domain characteristic components. For the first detection dataset (geotechnical medium data), which includes static stratigraphic data (such as lithologic distribution and stratigraphic stratification) and dynamic medium parameters (such as soil resistivity and dielectric constant), decomposition is performed using the continuous wavelet transform algorithm. The wavelet transform achieves multi-resolution analysis of the signal by convolving wavelet basis functions of different scales with the signal. For example, for seismic velocity curves in static stratigraphic data, the Mexican Hat wavelet is selected as the basis function to analyze the overall structural trends of the stratigraphic layer at large scales (low frequencies) and identify subtle features such as local interlayers or fractures at small scales (high frequencies). For the soil resistivity time series in the dynamic medium parameters, a three-layer decomposition using the db4 wavelet is performed to obtain a low-frequency approximate component (reflecting the long-term resistivity trend) and a high-frequency detail component (reflecting short-term disturbances). For the second detection data set (pipeline distribution data), time-frequency analysis was performed using the short-time Fourier transform (STFT). A fixed window function was used to divide the radar echo signal into multiple time windows. The signal within each window was Fourier transformed to generate a time-frequency matrix, where the horizontal axis is frequency and the vertical axis is time (corresponding to spatial position). The matrix element value represents the energy intensity of the frequency component at the corresponding position, thus clearly showing the frequency characteristics and spatial distribution of the pipeline reflection signal. For the third detection data set (electromagnetic field data), the empirical mode decomposition (EMD) method was used to decompose the complex electromagnetic field signal into multiple intrinsic mode functions (IMFs), each of which corresponds to a different frequency component, achieving adaptive time-frequency decomposition of non-stationary signals.

[0034] The energy mapping unit performs phase alignment and superposition on the frequency domain characteristic components and the original detection data set to generate a wave field characteristic spectrum. The core of phase alignment is to eliminate the phase difference between the frequency domain components and the original data to ensure the time consistency of the superimposed signal. The specific steps are: first, perform inverse time-frequency transformation on the frequency domain characteristic components extracted from each type of data set to restore them to time domain signals; then, calculate the cross-correlation function between the time domain signal and the original data set to determine the time delay corresponding to the maximum cross-correlation value, that is, the phase difference; then, adjust the phase of the frequency domain characteristic component to compensate for the time delay so that the two are accurately aligned on the time axis; finally, superimpose the adjusted frequency domain characteristic component with the original data set point by point to generate a wave field characteristic spectrum. Taking pipeline distribution data as an example, assuming that a high-frequency characteristic component extracted by STFT (corresponding to the strong reflection of the pipeline metal material) is 20ns in time delay with the original radar echo after inverse transformation, the phase of the characteristic component is advanced by the radian value corresponding to 20ns through phase compensation (such as the phase difference). ,in: Indicates phase difference; Indicates frequency; represents the time interval, where f is the characteristic frequency, is the delay time), and then superimposed with the original echo to accurately map the high-frequency features to the actual spatial position of the pipeline, forming a highlighted area with energy focus in the wave field characteristic map.

[0035] The noise suppression unit performs blind source separation and denoising on the wavefield signature spectrum, using the independent component analysis (ICA) algorithm to separate the signal from the noise. Blind source separation assumes that the source signals are independent and that the observed signal is a linear combination of the source signals. The wavefield signature spectrum is considered to be a mixture of valid signal sources (such as pipeline reflections and formation interface responses) and noise sources (such as ambient electromagnetic interference and instrument noise). The ICA algorithm optimizes the separation matrix to maximize the statistical independence of each separated component, thereby decomposing the spectrum into independent valid signal and noise components. In specific implementation, the spectrum is first normalized to zero mean and unity variance. Then, the separation matrix is ​​initialized, and the FastICA algorithm is used to iteratively update the matrix parameters. The negative entropy criterion is calculated as an independence measure until convergence to obtain independent components. Finally, the valid components containing the main energy are retained, while the independent components dominated by noise are removed to obtain the denoised wavefield signature spectrum. For example, in the wave field characteristic spectrum of electromagnetic field data, the ICA algorithm can separate the power frequency interference (manifested as periodic noise of a specific frequency) and the electromagnetic response signal of the underground target body, and improve the signal-to-noise ratio of the spectrum by eliminating the independent components corresponding to the power frequency interference.

[0036] After denoising, the wavefield synthesis layer performs cross-validation analysis on the wavefield characteristic maps corresponding to each type of detection data to generate multidimensional topological relationships. This cross-validation analysis compares the signal characteristics of different map types at the same spatial location to determine data consistency and correlation. The specific operation is as follows: First, the three types of maps (geotechnical media, pipeline distribution, and electromagnetic field) are spatially registered based on a reference coordinate system to ensure that points with the same geographic coordinates are located consistently across all maps. Then, a region of interest (ROI) is selected from the registered maps, such as a suspected pipeline anomaly area, and characteristic parameters of this region in the three types of maps are extracted (e.g., equivalent impedance for geotechnical media maps, radar reflectivity for pipeline maps, and field strength for electromagnetic field maps). Next, correlation metrics (e.g., covariance and mutual information) are calculated between the characteristic parameters to assess the degree of correlation between different data types in the region. Finally, the correlation across the entire region is represented as a graph structure, with nodes representing spatial locations and edges representing correlations between characteristic parameters, forming a multidimensional topological relationship. For example, at a certain coordinate point (X, Y, Z), if the geotechnical medium map shows a sudden change in equivalent impedance, the pipeline map has a strong reflection signal, and the electromagnetic field map detects a field strength anomaly, then these three types of data are significantly correlated here, and the edge weight (correlation coefficient) between this node and the adjacent nodes in the topological relationship diagram is high, indicating the possibility of an underground structural anomaly here.

[0037] The entire processing flow of the wavefield synthesis layer strictly adheres to the mathematical principles of signal processing. It extracts frequency features from the data through multi-scale time-frequency decomposition, spatially localizes these features using phase alignment and energy superposition, suppresses noise with the aid of blind source separation algorithms, and establishes spatial correlations among multi-source data through cross-validation. Each step is based on deterministic algorithms and physical models, avoiding the introduction of fictitious experimental descriptions. This ensures the accuracy of the wavefield feature maps and the reliability of the multi-dimensional topological relationships, providing high-quality input data for the subsequent feature matching and correlation analysis layers.

[0038] Example 3, This embodiment details the modeling process of the spatial correlation between wavefield feature maps in the correlation analysis layer, covering steps such as abnormal response area identification, feature coding sequence generation, correlation coefficient calculation, and multi-dimensional topological relationship construction. It implements spatial correlation analysis of multi-source data through signal processing algorithms and mathematical modeling, as follows: The correlation analysis layer first uses the signal feature extraction algorithm to identify the abnormal response area in the wave field characteristic spectrum. The signal feature extraction algorithm is based on the combination logic of threshold detection and edge detection: for the energy distribution matrix in the wave field characteristic spectrum, a global energy threshold is set. (such as 2 times the average energy value), the energy value is greater than The pixel points are marked as potential abnormal points; then, the Canny edge detection algorithm is used to extract the contour edges of the potential abnormal points to form a continuous abnormal response area. For example, in the radar wave field characteristic map corresponding to the pipeline distribution data, the reflection signal of the metal pipeline is manifested as a local energy peak. The threshold detection is used to identify the energy higher than The pipeline's spatial contours are then outlined using edge detection to determine the geometric range of the abnormal response area. For the wavefield characteristic spectrum of electromagnetic field data, the abnormal response area may be caused by underground cable leakage, manifested as an abnormally high field intensity at a specific frequency (such as 50Hz). Frequency domain threshold detection and spatial contour extraction are used to locate the source of the abnormal field.

[0039] After identifying the abnormal response area, a feature coding sequence corresponding to each type of detection data is generated based on the spectral characteristics of each area. The feature coding sequence is in the form of a multidimensional vector, containing parameters such as frequency band identification, energy level, and spatial coordinate offset. The specific coding rules are as follows: Frequency band identification: divide the frequency range of the wave field characteristic spectrum into frequency bands (such as low frequency, medium frequency, and high frequency bands), each frequency band corresponds to a unique identifier ( ).

[0040] Energy level: Normalize the energy value of the abnormal response area to the interval [1, M] (such as M=5), corresponding to the energy level ( ), where 1 is the lowest level and 5 is the highest level.

[0041] Spatial coordinate offset: Calculate the coordinates of the centroid of the abnormal response area with the origin of the reference coordinate system as the reference , and converted to an offset relative to the center of the current detection area .

[0042] Taking a certain abnormal response area of ​​the first detection data set (geotechnical medium data) as an example, if its main energy is concentrated in the intermediate frequency ( ), the normalized energy value is 4 ( ), the center of mass coordinate offset is ( ), then its feature encoding is The feature codes of all abnormal response areas are arranged in order of spatial coordinates to form a feature code sequence for this type of data. ,in is the number of abnormal response areas.

[0043] When calculating any two types of detection data (such as Class and When the feature coding sequence of the two classes is correlated, first determine whether the number of abnormal responses of the two classes is consistent. , then based on the number of abnormal responses with less (assuming it is the Class, quantity ) to perform frequency band expansion compensation based on the distribution characteristics of the highest energy frequency band in the image. The specific steps of frequency band expansion compensation are: Identify The highest energy level in the class feature coding sequence (e.g. ) abnormal response area, count its frequency band identification distribution, and determine the dominant frequency band (such as the frequency band with the highest frequency).

[0044] In the Insertion in class sequence virtual abnormal response points, and the frequency band identifier of the virtual point is set to , the energy level is set to (Second highest level), the spatial coordinate offset is generated based on the coordinate interpolation of adjacent real points to ensure that the distribution of virtual points conforms to the spatial trend of the original sequence.

[0045] After compensation, the length of the feature coding sequence of the two types of data is , recorded as and Next, the correlation coefficient between the abnormal response regions with the same frequency band identification is calculated. ,extract and All frequency bands are marked as feature encoding to form a subsequence and , each subsequence contains the energy level and spatial coordinate offset in the frequency band. The Pearson correlation coefficient is used to calculate the correlation between subsequences, and the formula is: , in, For the Class and Class data in frequency band The correlation coefficient under For subsequence and The length of (i.e., the number of abnormal responses in this frequency band); 、 They are 、 Middle Energy level of abnormal response; 、 They are 、 The average value of the medium energy level.

[0046] This formula measures the linear correlation of energy levels by standardized covariance, which ranges from [-1,1]. The larger the absolute value, the stronger the correlation. For example, if two types of data are in the medium frequency The energy level sequence under the ), indicating that the abnormal responses of the two in this frequency band have similar energy distribution characteristics and may correspond to the same underground target (such as the electromagnetic coupling effect between metal pipelines and the surrounding rock and soil media).

[0047] After calculating the correlation coefficients of all frequency bands, based on the weight of each frequency band (preset value, such as that set according to frequency band resolution or data reliability), the comprehensive correlation coefficient of any two types of data is obtained by weighted averaging: , in Comprehensive correlation coefficient It reflects the spatial correlation between the two types of data in the entire frequency band.

[0048] Finally, the multi-dimensional topological relationship between any two types of detection data is determined based on the comprehensive correlation coefficient. express: Among them, the node Corresponding to the abnormal response area of ​​each detection data, each node contains the spatial coordinates and frequency band information in the feature code; Connecting nodes with correlation, the weight of the edge is the corresponding comprehensive correlation coefficient , the larger the weight value is, the stronger the correlation between the two types of data in this area is.

[0049] For example, if the comprehensive correlation coefficient between geotechnical media data and pipeline distribution data in a certain area is 0.7, then in the topological relationship diagram, the corresponding nodes of the two are connected by an edge with a weight of 0.7, indicating that the geotechnical media anomaly in this area is strongly correlated with the pipeline distribution, which may be due to changes in soil compaction or corrosion around the pipeline. Abnormal media parameters.

[0050] Through the aforementioned process, the correlation analysis layer extracts spatial correlation features from multi-source data from the wavefield feature map, constructing multidimensional topological relationships encompassing frequency bands, energy, and spatial location. This provides a quantitative correlation basis for the subsequent multimodal data fusion in the model fusion unit. This entire process, based on signal processing algorithms and statistical analysis theory, achieves correlation modeling through rigorous mathematical definitions and logical deduction, avoiding the introduction of fictitious experimental effect descriptions and ensuring the scientific and interpretable nature of the topological relationships.

[0051] Example 4, This embodiment focuses on the functions of the imaging layer, covering aspects such as the multi-view interactive interface, iterative inversion algorithm optimization, spatial interpolation calculation and vector superposition, and error feedback mechanism. The construction and optimization process of the 3D panoramic imaging model is detailed in conjunction with specific application scenarios. The details are as follows: The multi-view interactive interface of the imaging layer provides users with interactive analysis tools for 3D panoramic imaging models. For example, in a substation underground survey scenario within a power transmission and transformation project, users input control commands through the graphical user interface (GUI). Sliding the mouse wheel adjusts the model's display resolution, switching from macroscopic stratigraphic structure (1m resolution) to pipeline details (0.1m resolution). Clicking buttons on the interface switches the section direction, such as displaying a horizontal section along the X-axis to visualize stratigraphic layers or a vertical section along the Z-axis to examine the vertical distribution of pipelines. The signal strength filtering function allows users to set an energy threshold using a slider, for example, to display only areas with electromagnetic field intensities greater than 50μT, allowing for quick location of high-field anomalies. The interactive interface renders model changes in real time, allowing users to intuitively compare underground structural features at different viewpoints, resolutions, and signal strengths. For example, if a stratigraphic impedance anomaly is detected in a horizontal section, switching to a vertical section reveals that the anomaly overlaps with a metal pipeline buried 2 meters deep, thus identifying a possible connection between the two.

[0052] During the multimodal data fusion phase, the imaging layer dynamically optimizes the pre-set underground structure template using an iterative inversion algorithm. The underground structure template is constructed based on the geological survey report for the target area and initially includes three strata (silty clay, sand, and bedrock) and known pipeline locations (e.g., an east-west power cable buried at a depth of 1.5 meters). After the wavefield synthesis layer generates a new wavefield signature map, the iterative inversion algorithm compares the anomalous response areas in the map with the template model. For example, the wavefield signature map shows a high-frequency, strong energy response at coordinates (X=10m, Y=5m, Z=3m), whereas the template model indicates this location as a sand layer with no pipeline. The algorithm uses the least squares method to calculate the difference between the model prediction and the measured data, adjusts the dielectric parameters of that area in the template (e.g., adjusting the dielectric constant of the sand layer from 8 to 12 to match the measured high-frequency response), and adds a virtual node for the unknown pipeline. After five iterations, the sand layer distribution range of the template model was narrowed, and a suspected pipeline running north-south was added. Its spatial position was completely consistent with the abnormal area in the wave field characteristic map, realizing the dynamic adaptation of the model to the measured data.

[0053] During the 3D spatial reconstruction process, the imaging layer performs spatial interpolation of multidimensional topological relationships based on preset imaging weights. Imaging weights are set based on the reliability of the data type. For example, electromagnetic field data is assigned a weight of 0.4 due to its high positioning accuracy (±0.2m error), pipeline radar data is assigned a weight of 0.3 due to its high resolution (0.1m), and geotechnical data is assigned a weight of 0.3 due to its wide coverage. For example, a node in the multidimensional topological relationship, corresponding to coordinates (X=15m, Y=8m, Z=2.5m), is associated with an equivalent impedance anomaly in the geotechnical data, a weak reflection signal from the pipeline radar, and a moderate electromagnetic field response (weights of 0.3, 0.3, and 0.4, respectively). Spatial interpolation is performed using the kriging interpolation method. Based on the correlation coefficients of the node's 10 neighboring nodes (e.g., a correlation of 0.6 with the node to the north and 0.7 with the node to the south), the 3D spatial attribute values ​​of the point are estimated, generating a continuous correlation field distribution. Then, the correlation field distribution and the wave field characteristic map are vector superimposed: for example, in the wave field characteristic map of the pipeline radar, the reflection intensity of the point is 15dB. After superimposing the weight contribution of the electromagnetic field data in the correlation field, the comprehensive intensity value of the point in the final imaging model is (Assuming the corresponding electromagnetic field intensity value is 10), it is displayed in orange with a gradient color block, indicating a medium abnormality level.

[0054] The error feedback unit monitors model accuracy in real time during the vector stacking process. For example, a set of electromagnetic field intensity profiles from historical survey data records field intensity values ​​at 1-meter intervals along the X-axis from 0 to 20 meters (Z = 2 meters). The error feedback unit calculates the absolute error between the predicted value of the 3D panoramic imaging model and the historical data at the corresponding location. For example, at X = 5 meters, the predicted value is 45 μT, while the historical data is 48 μT, resulting in a residual error of 3 μT. If the residual value in a certain area exceeds a preset threshold (e.g., 5 μT) for five consecutive calculations, the data reconstruction unit is triggered to re-execute time-frequency synchronization. For example, if a systematic 100-ms delay in the timestamps of the third survey dataset (electromagnetic field data) is detected, resulting in a time-frequency synchronization error, the delay can be eliminated by recalibrating the device clock and adjusting the phase compensation parameters. Simultaneously, the imaging layer updates the spatial interpolation parameters of the multi-dimensional topological relationships, such as reducing the kriging interpolation search radius from 3 meters to 2 meters to improve interpolation accuracy in the local area. After feedback optimization, the residual error between the model prediction and the historical data is reduced to 2 μT, meeting the accuracy requirements.

[0055] In practical applications in power transmission and transformation projects, the imaging layer processing process can fully reproduce the complex structure of underground space. For example, in a transmission line route survey, the 3D panoramic imaging model revealed: the surface layer from 0 to 1.5 meters is a silty clay layer containing an east-west communication pipeline (buried at a depth of 1.2 meters, with a diameter of 50 mm); the 1.5-3 meters layer is a sand layer with localized water-rich areas (identified by equivalent impedance jumps in geotechnical data); below 3 meters is bedrock, above which lies a north-south power cable (buried at a depth of 2.8 meters, located by high-field intensity areas in electromagnetic field data). The multi-view interactive interface allows engineers to observe the spatial relationship between the pipeline and the strata from different angles. For example, rotating the model to view the intersection of the power cable and the communication pipeline, and analyzing the electromagnetic field coupling effects at the intersection through cross-sections. An iterative inversion algorithm automatically optimizes the undulating shape of the bedrock interface to match the seismic wave velocity data; an error feedback mechanism continuously monitors model accuracy, ensuring that the location of the water-rich areas is within 0.5 meters.

[0056] The imaging layer achieves high-precision 3D modeling of underground spaces through multi-view interaction, dynamic model optimization, spatial interpolation and fusion, and closed-loop error feedback. Each step is based on actual data processing logic and engineering application scenarios, avoiding fictitious descriptions and ensuring the reliability and engineering practicality of the imaging results. This provides an intuitive and accurate basis for decision-making in underground surveys, route planning, and risk assessment for power transmission and transformation projects.

[0057] Example 5, This embodiment involves the specific implementation of the dynamic compensation unit and real-time monitoring module of the feature matching layer. Combined with the underground survey scenario of power transmission and transformation projects, it details the amplitude compensation based on medium density gradient and the electromagnetic field abnormal area marking and risk warning process.

[0058] The dynamic compensation unit in the feature matching layer performs amplitude compensation on the wavefield signature spectrum based on the dielectric density gradient in the target area. For example, in an underground survey of a substation, the target area consists of a backfill soil layer (density 1.8×10³kg / m³) on the surface, followed by a clay layer (density 2.0×10³kg / m³) and a sandstone layer (density 2.3×10³kg / m³). The dielectric density increases stepwise with depth. The dynamic compensation unit first constructs a dielectric absorption coefficient compensation model based on the electromagnetic field intensity data from the third detection dataset. By analyzing the electromagnetic field intensity attenuation at different depths, it was found that the signal attenuation rate in the backfill soil layer is 0.5dB / m, the clay layer is 1.2dB / m, and the sandstone layer is 1.8dB / m. Based on this, a piecewise linear absorption coefficient model was established: the absorption coefficient α = 0.5 for depths 0-2 meters (backfill soil), 1.2 for 2-5 meters (clay), and 1.8 for depths below 5 meters (sandstone).

[0059] The dynamic compensation unit uses a backpropagation algorithm to update the energy distribution of each frequency band in the wavefield signature map. For example, a power cable buried at a depth of 3 meters shows an energy response in the 50Hz frequency band in the electromagnetic field data wavefield signature map. Because this location is within a clay layer (α = 1.2), the signal energy attenuates by approximately 3.6dB (1.2 × 3) after 3 meters of propagation, resulting in an underestimated energy value in the map. Based on the absorption coefficient model, the backpropagation algorithm compensates the energy value in this frequency band by the attenuation amount, adding 3.6dB. This restores the cable's energy response in the map to a value close to the true intensity at the surface transmitting end, thus more clearly displaying the cable's spatial trajectory. Similarly, for pipelines traversing different dielectric layers, the dynamic compensation unit calculates the attenuation of each dielectric layer segment by segment and applies compensation, ensuring consistent amplitudes for targets at different depths in the wavefield signature map, facilitating feature matching and identification.

[0060] The real-time monitoring module marks areas in the 3D panoramic imaging model where electromagnetic field strength exceeds safety thresholds. Safety thresholds are set according to the "Limits of Electromagnetic Environment Control" (GB8702-2014). For example, the public exposure limits for power-frequency electromagnetic fields are 4000 V / m for electric field strength and 100 μT for magnetic induction. During an underground cable survey for a transmission line, the 3D model showed a magnetic induction strength of 120 μT near the cable terminal (coordinates X=20m, Y=15m, Z=1.8m), exceeding the safety threshold of 100 μT. The real-time monitoring module automatically marked this area in red and displayed a detailed information box on the right side of the interface, displaying the field strength value, the magnitude of the excess (+20%), and the dielectric layer (backfill soil) in which it occurred. This alerted engineering personnel to the electromagnetic radiation risk in this area and the need to implement shielding measures or adjust the construction plan.

[0061] The real-time monitoring module's early warning trigger mechanism activates upon detecting abnormal field strength in the spatial overlap area between pipeline distribution data and electromagnetic field data. For example, a natural gas pipeline buried 1.5 meters deep (the second detection dataset) and an operating 10kV power cable (the third detection dataset) are only 0.3 meters apart at coordinates (X=10m, Y=8m, Z=1.6m), which is lower than the safe distance (0.5 meters) required by the "Electric Power Facility Protection Regulations." The real-time monitoring module detects a sudden increase in electromagnetic field strength to 150μT in this overlapping area (80μT in normal operation), indicating a possible leakage caused by cable insulation damage. The system immediately generates a risk warning signal with coordinate positioning, including the following information: Warning type: pipeline electromagnetic coupling abnormality Coordinate position: X=10m, Y=8m, Z=1.5-1.6m Abnormal description: The field strength in the area adjacent to the power cable and natural gas pipeline suddenly increases Response suggestion: Immediately stop excavation work in the area and conduct cable insulation testing Early warning signals are delivered through multiple channels, including system pop-up windows, text messages (sent to the project manager's phone), and on-site alarms (with audible and visual alarms), ensuring timely dissemination of risk information. Upon receiving the alert, engineers can access historical wavefield characteristic maps of the area through a multi-view interactive interface, comparing and analyzing time series changes in electromagnetic field intensity to confirm that the anomaly is a sudden increase rather than equipment noise, thereby initiating emergency investigation procedures.

[0062] In another application scenario, a power transmission and transformation project needed to assess the electromagnetic environmental impact of aging underground pipelines. The real-time monitoring module mapped the electromagnetic field intensity distribution of multiple abandoned cables within the 3D model. The field intensity in the area surrounding one abandoned cable (de-energized) at a depth of 4 meters was close to the background value. Another abandoned cable (possibly not fully de-energized) at a depth of 3 meters had a field intensity of 60 μT. While not exceeding the safety threshold, it was significantly higher than the surrounding area (background value of 20 μT). The monitoring module marked the latter as a yellow warning area, alerting engineers that the cable might still be energized and requiring further testing and confirmation. The wavefield characteristic spectrum after dynamic compensation clearly illustrates the relationship between the cable's direction and changes in the medium density. When passing through sandstone layers (high-density media), the energy response of the cable in the spectrum is more continuous due to the amplitude enhancement of the dynamic compensation unit. However, when passing through backfill layers, the energy response is slightly weaker due to the smaller compensation, but still above the background value, assisting in assessing the cable's integrity.

[0063] The feature matching layer and real-time monitoring module, through dielectric density compensation and real-time electromagnetic field monitoring, address the issue of feature ambiguity caused by signal attenuation in complex underground dielectric environments, as well as the challenge of identifying electromagnetic safety risks in power transmission and transformation projects. The dynamic compensation unit achieves physical consistency correction of wavefield characteristics based on actual dielectric parameters and electromagnetic field propagation laws. The real-time monitoring module provides quantitative risk assessment and early warning according to national standards and engineering specifications. The entire process is based on measured data and engineering logic, avoiding fictitious descriptions of effects and ensuring the system's safety and practicality in underground surveys of power transmission and transformation projects.

[0064] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0065] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A panoramic imaging system for automated underground panoramic survey of power transmission and transformation projects, characterized in that: include: A data acquisition module is configured to collect underground multi-source detection data from a target area. The multi-source detection data includes a first detection data set corresponding to geotechnical medium data, a second detection data set corresponding to pipeline distribution data, and a third detection data set corresponding to electromagnetic field data. The geotechnical medium data includes static stratum data retrieved from a geological database and dynamic medium parameters acquired through a distributed detection device. A panoramic imaging module, configured to input the multi-source detection data into a three-dimensional imaging engine for analysis; A three-dimensional panoramic imaging model of the underground space is constructed according to the output results of the three-dimensional imaging engine; the three-dimensional imaging engine includes a data reconstruction unit and a model fusion unit, wherein the data reconstruction unit is used to align the multi-source detection data for spatial consistency, and the model fusion unit performs multimodal data fusion based on a preset underground structure template and historical survey data; the model fusion unit includes a wave field synthesis layer, a feature matching layer, a correlation analysis layer and an imaging layer connected in sequence, the wave field synthesis layer is used to perform frequency domain conversion processing on each data set in the multi-source detection data to generate a wave field feature map; the correlation analysis layer is used to model the spatial correlation between the wave field feature maps corresponding to the multi-source detection data to generate a multi-dimensional topological relationship; the imaging layer is used to perform three-dimensional spatial reorganization based on the multi-dimensional topological relationship and the wave field feature map to generate a three-dimensional panoramic imaging model.

2. The system for automatic underground panoramic survey and imaging of power transmission and transformation engineering according to claim 1, characterized in that: The spatial correlation between the wave field characteristic spectra corresponding to the multi-source detection data is modeled to generate a multi-dimensional topological relationship, including: A signal feature extraction algorithm is used to identify abnormal response areas in the wave field characteristic map, and a feature coding sequence corresponding to each type of detection data is generated based on the spectrum characteristics of each abnormal response area; Calculating a correlation coefficient between abnormal response areas having the same frequency band identifier in feature coding sequences corresponding to any two types of detection data, and determining a multi-dimensional topological relationship between the arbitrary two types of detection data based on the correlation coefficient; calculating the correlation coefficient between abnormal response areas having the same frequency band identifier in feature coding sequences corresponding to any two types of detection data, including: When the number of abnormal responses in the characteristic coding sequences corresponding to any two types of detection data is inconsistent, frequency band expansion compensation is performed based on the distribution characteristics of the highest energy frequency band in the one with the smaller number of abnormal responses, and the correlation coefficient between abnormal response areas with the same frequency band identifier is calculated based on the compensated sequence.

3. The system for automatic underground panoramic survey and imaging of power transmission and transformation engineering according to claim 1, characterized in that: The data reconstruction unit is specifically used for: performing time-frequency synchronization processing on the first detection dataset, the second detection dataset, and the third detection dataset according to a preset reference coordinate system to obtain an aligned first detection dataset, an aligned second detection dataset, and an aligned third detection dataset; An adaptive filtering method is used to suppress the interference signals in the aligned first detection data set and the aligned second detection data set, and a phase compensation method is used to correct the waveform distortion in the aligned third detection data set to obtain a first optimized detection data set, a second optimized detection data set, and a third optimized detection data set.

4. The system for automatic underground panoramic survey and imaging of power transmission and transformation engineering according to claim 3, characterized in that: The first optimized detection data set includes processed static formation data and processed dynamic medium parameters, and the data reconstruction unit is further used to: analyzing energy attenuation curves of the processed static formation data and the processed dynamic medium parameters within a historical survey period; Calculating the equivalent impedance parameters of the processed dynamic medium parameters in the specified area based on the energy attenuation curve and the wave velocity parameters of the processed static formation data in the specified area; A comprehensive medium data set is generated according to the processed dynamic medium parameters and their equivalent impedance parameters, and the detection data corresponding to the comprehensive medium data set is used as the first optimized detection data set.

5. The system for automatic underground panoramic survey and imaging of power transmission and transformation engineering according to claim 1, characterized in that: The wave field synthesis layer includes a spectrum analysis unit and an energy mapping unit; The spectrum analysis unit is used to perform multi-scale time-frequency decomposition on each type of detection data set contained in the multi-source detection data, so as to extract corresponding frequency domain feature components from each type of detection data set; The energy mapping unit is used to perform phase alignment and superposition on the frequency domain characteristic components extracted from each type of detection data set and the original detection data set to generate a wavefield characteristic spectrum.

6. The system for automatic underground panoramic survey and imaging of power transmission and transformation engineering according to claim 5, characterized in that: The wave field synthesis layer further includes a noise suppression unit, which is used to perform blind source separation and denoising processing on the wave field characteristic map; The modeling of the spatial correlation between the wavefield characteristic maps corresponding to the multi-source detection data includes: performing cross-validation analysis on the denoised wavefield characteristic maps corresponding to various types of detection data to generate the multi-dimensional topological relationship.

7. The system for underground panoramic survey and panoramic imaging of power transmission and transformation project automation according to claim 1, characterized in that: The imaging layer includes a multi-view interactive interface, which supports the user to switch sections, filter signal strength and adjust resolution of the three-dimensional panoramic imaging model through control instructions; The multimodal data fusion based on the preset underground structure template and historical survey data includes: dynamically optimizing the physical property parameters of the underground structure template through an iterative inversion algorithm to adapt to the update of the multi-dimensional topological relationship; The three-dimensional spatial reorganization based on the multi-dimensional topological relationship and the wavefield characteristic map includes: performing spatial interpolation calculation on the multi-dimensional topological relationship according to preset imaging weights, and performing vector superposition with the wavefield characteristic map.

8. The system for automatic underground panoramic survey and imaging of power transmission and transformation engineering according to claim 1, characterized in that: The characteristic matching layer includes a dynamic compensation unit for performing amplitude compensation on the wave field characteristic spectrum according to the medium density gradient of the target area; The operations performed by the dynamic compensation unit include: establishing a compensation model of the dielectric absorption coefficient based on the electromagnetic field intensity data of the third detection data set, and updating the energy distribution of each frequency band in the wave field characteristic map through a back propagation algorithm.

9. The system for automatic underground panoramic survey and imaging of power transmission and transformation engineering according to claim 7, characterized in that: The imaging layer also includes an error feedback unit for calculating the residual value between the three-dimensional panoramic imaging model and the historical survey data in real time during the vector superposition process; when the residual value exceeds a preset threshold, the data reconstruction unit is triggered to re-execute the time-frequency synchronization processing and update the spatial interpolation parameters of the multi-dimensional topological relationship.

10. The automatic underground panoramic survey and panoramic imaging system for power transmission and transformation engineering according to claim 1, characterized in that: The system also includes a real-time monitoring module for marking areas in the three-dimensional panoramic imaging model where the electromagnetic field intensity exceeds a safety threshold; the real-time monitoring module is configured with an early warning trigger mechanism, which generates a risk early warning signal with coordinate positioning when abnormal field intensity is detected in the spatial overlap area between pipeline distribution data and electromagnetic field data.

Images