Underground interference source positioning method and system based on impact surge time difference method

Abstract

The invention belongs to the technical field of electromagnetic measurement and signal processing, and relates to an underground interference source positioning method and system based on an impact surge time difference method, and the method comprises the following steps: obtaining mixed potential time domain signals synchronously collected by a plurality of sensors distributed in a to-be-measured region; performing preliminary blind source separation processing on the mixed potential time domain signal to obtain a candidate independent signal component, and performing time difference positioning processing on the candidate independent signal component to generate a position candidate set corresponding to the candidate independent signal component; performing cross validation and screening on the position candidate set to obtain an effective source number estimation value; performing blind source separation and time difference positioning processing on the mixed potential time domain signal again by using the effective source number estimation value as a constraint condition until the positioning result converges, and determining the geographic position of the underground interference source; according to the method, the problem that the multi-source mixed signal is difficult to position due to the fact that components belonging to different sources in the mixed signal are difficult to effectively distinguish by a single time difference positioning method is solved.

Description

Technical Field

[0001] This invention belongs to the technical field of electromagnetic measurement and signal processing, and relates to a method and system for locating underground interference sources using the time difference of impulse surge. Background Technology

[0002] In fields such as power systems, pipeline networks, and geological exploration, measuring and analyzing the electrical characteristics of underground targets is an important technical means. Among these, surge current sources refer to transient, high-energy current pulses generated in underground environments due to equipment failure, lightning strikes, or geological activity. These current pulses induce a changing electromagnetic field in the surrounding medium. By deploying a sensor array on the surface and measuring the potential signal generated by this electromagnetic field, the location of the source can be determined.

[0003] Existing techniques for locating underground surge current sources, such as the single time-difference-of-arrival (STO) method, rely on measuring the minute time differences in the arrival times of a single signal at different sensors to determine the source location. This method typically deploys multiple synchronized potential sensors on the Earth's surface. When a single surge event occurs, the arrival times of the signal wavefronts recorded by each sensor are captured, the time difference is calculated, and a hyperboloid equation system is constructed using the electromagnetic wave propagation speed. Finally, the geographical location of the interference source is determined by solving the intersection of the equation system. For example, Chinese patent CN113030651B discloses a method for locating surge current sources. This method establishes at least three surface potential observation points, analyzes the time-domain waveform of the scalar potential at these observation points when the surge current enters the ground, and derives the time difference. Based on the surge current time-difference-of-arrival method, the propagation speed of electromagnetic waves in the earth is determined, and two hyperbolas are established between every three surface potential observation points. The intersection of these two hyperbolas determines the location of a surge current ingress point. This method has good location accuracy in the case of a single surge current source.

[0004] However, in real-world complex underground environments, multiple surge current sources often exist simultaneously. The electromagnetic signals generated by these current sources are linearly superimposed during propagation, resulting in surface sensors receiving mixed and complex potential field signals. In this case, the traditional single time difference positioning method cannot effectively distinguish the components belonging to different sources in the mixed signal, leading to confusion of arrival time information from different sources, which in turn causes errors or even failures in the positioning results. Summary of the Invention

[0005] In a first aspect, the present invention provides a method for locating underground interference sources using the time difference of shock surge method, employing the following technical solution:

[0006] A method for locating underground interference sources using the time-difference method of shock surges includes the following steps:

[0007] S1. Acquire the mixed potential time-domain signal synchronously collected by multiple sensors distributed in the area to be measured;

[0008] S2. Perform preliminary blind source separation processing on the mixed potential time domain signal to obtain candidate independent signal components, and perform time difference localization processing on the candidate independent signal components to generate a candidate set of positions corresponding to the candidate independent signal components.

[0009] S3. Perform cross-validation and filtering on the candidate location set to obtain an estimate of the number of valid sources;

[0010] S4. Using the estimated number of effective sources as constraints, perform blind source separation and time difference localization processing on the mixed potential time domain signal again until the localization results converge, and determine the geographical location of the underground interference source. The convergence of the localization results means that the change in the number of effective sources and the corresponding geographical locations obtained in the continuous iteration is less than the preset value.

[0011] A further aspect of the present invention includes the following steps before performing preliminary blind source separation processing on the mixed potential time-domain signal:

[0012] The mixed potential time-domain signal is subjected to bandpass filtering to remove low-frequency power frequency interference and high-frequency random noise, resulting in a filtered signal.

[0013] The filtered signal is processed by first-order differentiation to sharpen the rising edge characteristics of the signal wavefront and generate a hybrid signal with enhanced features.

[0014] The feature-enhanced hybrid signal is used as the input to the hybrid potential time-domain signal.

[0015] A further aspect of this invention involves preliminary blind source separation processing based on transient feature-driven independent component analysis, comprising the following steps:

[0016] The mixed potential time-domain signals or feature-enhanced mixed signals from multiple sensors are used to construct an observation signal matrix;

[0017] The observed signal matrix and the separation matrix obtained after iterative optimization are linearly transformed to obtain the separated signal matrix;

[0018] Each row of the signal matrix is ​​output as a statistically independent candidate independent signal component.

[0019] A further aspect of the present invention generates a candidate set of locations, comprising the following steps:

[0020] For each candidate independent signal component, the arrival time of its wavefront to each sensor is extracted, and a multi-channel wavefront timestamp is generated;

[0021] The arrival time difference between different sensor pairs is calculated based on the multi-channel wavefront timestamps to obtain time difference data;

[0022] Based on time difference data and the preset electromagnetic wave propagation speed, a set of hyperboloid equations is constructed;

[0023] Solve the hyperboloid equations to obtain the candidate geographic locations corresponding to the candidate independent signal components, and summarize all candidate geographic locations to form a candidate location set.

[0024] A further aspect of this invention involves cross-validation and filtering of the candidate location set, including the following steps:

[0025] Determine whether each candidate geographic location in the candidate location set falls within the preset valid geographic area and depth range;

[0026] Eliminate all candidate geographic locations that fall outside the range to obtain a spatially valid set of candidate locations.

[0027] A further aspect of the present invention, after obtaining a spatially valid set of candidate locations, includes the following steps:

[0028] Obtain the fitting residuals generated during the solution process for each candidate geographic location in the spatially valid candidate location set, forming a residual dataset;

[0029] Each fitted residual in the residual dataset is compared with a preset residual threshold;

[0030] Candidate geographical locations with fitting residuals greater than the residual threshold are removed, and a set of candidate locations after residual filtering is selected.

[0031] A further aspect of the present invention, after selecting the candidate position set after residual screening, includes the following steps:

[0032] Calculate the geographical distance between any two candidate geographic locations within the candidate location set after residual screening to obtain the distance matrix;

[0033] Based on the distance matrix and a preset distance threshold, multiple candidate geographic locations that are geographically less than the distance threshold are identified as redundant location clusters.

[0034] Multiple candidate geographic locations within a redundant location cluster are merged into a single location point, which, together with other candidate geographic locations that are not identified as redundant, constitutes a valid location result set.

[0035] A further aspect of this invention utilizes the estimated number of effective sources as a constraint for further blind source separation, comprising the following steps:

[0036] The number of location points contained in the set of valid location results is counted and determined as the estimated number of valid sources;

[0037] The estimated number of effective sources is used as the target separation quantity parameter for blind source separation processing, and a new round of iterative separation processing is performed on the mixed potential time domain signal.

[0038] A further aspect of the present invention, the blind source separation process, further includes the following steps:

[0039] From the candidate independent signal components, extract the signal components corresponding to the location points in the effective positioning result set, and use them as reference signal waveforms;

[0040] When performing a new round of iterative separation processing, the group reference signal waveform is used as prior knowledge to guide the convergence direction of the blind source separation processing.

[0041] Secondly, the present invention provides an underground interference source localization system using the time difference of shock surge method, which adopts the following technical solution:

[0042] A system for locating underground interference sources using the time-difference method for shock surges includes the following modules:

[0043] The signal acquisition module is used to acquire the mixed potential time-domain signal synchronously collected by multiple sensors distributed in the area to be measured;

[0044] The blind source separation module is used to perform preliminary blind source separation processing on the mixed potential time-domain signal to obtain candidate independent signal components;

[0045] The time difference positioning module is used to perform time difference positioning processing on candidate independent signal components and generate a candidate set of locations corresponding to the candidate independent signal components.

[0046] The cross-validation module performs cross-validation and filtering on the candidate location set to obtain an estimate of the number of valid sources.

[0047] The feedback iteration module uses the estimated number of effective sources as a constraint to perform blind source separation and time difference localization processing on the mixed potential time domain signal again;

[0048] The location module obtains the geographical location of the underground interference source based on the output of the feedback iteration module until the location result converges.

[0049] In summary, the present invention has the following beneficial technical effects:

[0050] 1. By introducing blind source separation technology and closed-loop feedback mechanism, the positioning failure caused by the mixing of multiple interference source signals is solved. It overcomes the defects of existing technologies that cause positioning confusion, accuracy reduction or even failure due to the inability to distinguish different source signals. It can decouple independent signal components from mixed signals and perform iterative optimization. It can still provide relatively stable and effective positioning in multi-source concurrent scenarios.

[0051] 2. By using cross-validation and screening steps, the problem of false localization and misjudgment of the number of sources caused by unsatisfactory blind source separation results is solved. This reduces the drawbacks of existing technologies, such as excessive reliance on algorithm preset parameters and low reliability of output results. It also reduces the impact of unreasonable space, poor solution quality and redundant candidate positions, and obtains effective source number and location information that is close to the real situation.

[0052] 3. By using the estimated number of valid sources obtained after verification as a constraint to feed back to the signal separation stage, the problem of unknown source number in blind source separation is solved. This overcomes the limitations of existing technologies, which rely on preset parameters based on human experience, resulting in unstable separation effects and inaccurate positioning results. This improves the quality of signal separation and the accuracy of subsequent positioning. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 A flowchart illustrating an embodiment of this application is disclosed.

[0055] Figure 2 Structural schematic diagrams of embodiments of this application are disclosed. Detailed Implementation

[0056] In existing technologies, such as Chinese patent CN115436734B, a method for locating underground stray current sources is proposed. This method measures the propagation speed of electromagnetic waves in the earth based on the relative permittivity of the earth, establishes an underground stray current location model with five probes to locate the underground impact source, and obtains the coordinates of the impact source through a cross array method. This method improves the location accuracy by increasing the number of observation points, but it is still mainly for the case of a single current source. As mentioned in the background technology, when faced with multi-source mixed signals, existing technologies cannot effectively distinguish signals from different sources, leading to location confusion and decreased accuracy. In order to solve the problems of difficulty in locating multi-source mixed signals, limitations of blind source separation technology, low accuracy of harmonic source location, and lack of system solutions in existing technologies, and to achieve accurate location and effective management of multiple impact surge current sources, this invention provides a multi-impact surge current source location method based on blind source separation.

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] The following is in conjunction with the appendix Figure 1 , Figure 2 A preferred description of the present invention is provided below.

[0059] See attached document Figure 1 This invention proposes a method for locating underground interference sources using the time difference of impact surge method, comprising the following steps:

[0060] S1. Acquire the mixed potential time-domain signal synchronously collected by multiple sensors distributed in the area to be measured;

[0061] S2. Perform preliminary blind source separation processing on the mixed potential time domain signal to obtain candidate independent signal components, and perform time difference localization processing on the candidate independent signal components to generate a candidate set of positions corresponding to the candidate independent signal components.

[0062] S3. Perform cross-validation and filtering on the candidate location set to obtain an estimate of the number of valid sources;

[0063] S4. Using the estimated number of effective sources as constraints, perform blind source separation and time difference localization processing on the mixed potential time domain signal again until the localization results converge, and determine the geographical location of the underground interference source. The convergence of the localization results means that the change in the number of effective sources and the corresponding geographical locations obtained in the continuous iteration is less than the preset value.

[0064] In one embodiment of the present invention, before performing preliminary blind source separation processing on the mixed potential time-domain signal, the following steps are further included:

[0065] The mixed potential time-domain signal is subjected to bandpass filtering to remove low-frequency power frequency interference and high-frequency random noise, resulting in a filtered signal.

[0066] The filtered signal is processed by first-order differentiation to sharpen the rising edge characteristics of the signal wavefront and generate a hybrid signal with enhanced features.

[0067] The feature-enhanced hybrid signal is used as the input to the hybrid potential time-domain signal.

[0068] Specifically, before performing preliminary blind source separation processing on the acquired mixed potential time-domain signal, this method first implements a key preprocessing step, namely, impact transient feature enhancement processing. This processing aims to enhance and highlight the signal wavefront characteristics that characterize the initiation point of the impact surge event from the original complex mixed potential time-domain signal.

[0069] For the mixed potential time-domain signal acquired from the i-th sensor The impact transient feature enhancement processing includes two core components;

[0070] The first step is bandpass filtering, which reduces the signal... A bandpass filter, whose passband frequency range is set based on the prior physical characteristics of the signal generated by the underground shock surge source, is used to filter out low-frequency power frequency interference and high-frequency random noise unrelated to the surge front, thus obtaining the filtered signal. Generally, the energy of a typical shock surge signal is mainly concentrated in the low to mid-frequency range. Therefore, the lower passband frequency of the bandpass filter can be set in the range of 1kHz-10kHz to effectively filter out 50 / 60Hz power frequency and its harmonic interference; the upper passband frequency can be set in the range of 100kHz-1MHz to filter out high-frequency random noise. For example, a Butterworth or Chebyshev filter with a passband of 5kHz-500kHz can be selected. It should be noted that these parameter ranges can be derived from a large amount of practical experience in the spectrum analysis of underground shock surge signals, so as to suppress background noise to a large extent while preserving the core transient characteristics of the signal.

[0071] The second step is differential sharpening, which refines the filtered signal. Performing first-order differentiation yields the final feature-enhanced mixed signal. Its mathematical expression can be simplified as:

[0072]

[0073] Where t represents the time variable. The signal after filtering in the i-th channel. This is the output signal of the i-th channel obtained after impact transient characteristic enhancement processing. This differentiation operation transforms the steep rising edge of the wavefront in the original signal into a sharp pulse peak with prominent amplitude, enhancing the wavefront's discernibility on the time axis. And all channels... Together, they constitute a feature-enhanced hybrid signal. This feature-enhanced hybrid signal is then fed into the preliminary blind source separation processing module. Because the signal wavefront is sharpened into a distinct non-Gaussian peak, this provides statistical separation conditions for subsequent blind source separation algorithms, especially independent component analysis (ICA), thereby enabling more accurate decomposition of each independent candidate signal component.

[0074] In one embodiment of the present invention, the preliminary blind source separation process based on transient feature independent component analysis includes the following steps:

[0075] The mixed potential time-domain signals or feature-enhanced mixed signals from multiple sensors are used to construct an observation signal matrix;

[0076] The observed signal matrix and the separation matrix obtained after iterative optimization are linearly transformed to obtain the separated signal matrix;

[0077] Each row of the signal matrix is ​​output as a statistically independent candidate independent signal component.

[0078] Specifically, the preliminary blind source separation processing employed in this invention is preferably an independent component analysis (ICA) technique based on transient characteristics. ICA is a signal processing method that recovers the original, statistically independent source signals from a multi-channel mixed signal by finding an effective linear transformation. The basic assumption of ICA is that multiple independent source signals are mixed by an unknown linear system and then received by multiple sensors.

[0079] In one embodiment of the present invention, the input to the independent component analysis technique is the mixed potential time-domain signal obtained in the aforementioned steps, or the feature-enhanced mixed signal obtained after impact transient feature enhancement processing.

[0080] The observed signals from N sensors are represented as matrix X, where each row represents the time-domain signal from the sensor. The goal of independent component analysis is to find a separation matrix W such that the signals in each row of the output matrix S obtained through the following linear transformation... The statistical independence between them has increased.

[0081] The linear transformation formula for blind source separation is as follows:

[0082]

[0083] Where X is the observed mixed signal matrix, W is the separation matrix to be solved, and S is the separated signal matrix, with each row... These are the candidate independent signal components. According to the central limit theorem, the mixture of multiple independent random variables tends to a Gaussian distribution. Conversely, to separate the original signal from the mixed signal, it is necessary to find a direction in which the probability distribution of the signal component projected onto the mixed signal is furthest from the Gaussian distribution, i.e., the strongest non-Gaussianity. The signal generated by the surge current source has a typical transient pulse shape, and its probability distribution exhibits peaks and heavy tails, which is a manifestation of strong non-Gaussianity, especially after feature enhancement processing. Through iterative optimization, the coefficients of the separation matrix W are continuously adjusted to measure non-Gaussianity indicators, such as kurtosis or negative entropy, ultimately improving the statistical independence of the output candidate independent signal components.

[0084] Taking the iterative optimization algorithm based on negative entropy, FastICA, as an example:

[0085]

[0086] in, This is the separation vector, which is a row in the separation matrix W; The observed signal matrix; It is a nonlinear function; It is a standard Gaussian random variable; For expectation operators; Let be the negative entropy approximation objective function, used to maximize non-Gaussianity;

[0087] The iterative update rule is as follows:

[0088]

[0089] in, and They are The derivative and second derivative.

[0090] In one embodiment of the present invention, generating a candidate location set using the time difference positioning method includes the following steps:

[0091] For each candidate independent signal component, the arrival time of its wavefront to each sensor is extracted, and a multi-channel wavefront timestamp is generated;

[0092] The arrival time difference between different sensor pairs is calculated based on the multi-channel wavefront timestamps to obtain time difference data;

[0093] Based on time difference data and the preset electromagnetic wave propagation speed, a set of hyperboloid equations is constructed;

[0094] Solve the hyperboloid equations to obtain the candidate geographic locations corresponding to the candidate independent signal components, and summarize all candidate geographic locations to form a candidate location set.

[0095] Specifically, the step of performing time difference positioning to generate a candidate location set in this invention is a systematic process of independently processing each separated candidate independent signal component.

[0096] First, for a given candidate independent signal component, its waveform at each sensor is analyzed to extract its wavefront arrival time. This step is typically achieved by identifying the instant when the signal amplitude first exceeds a preset arrival threshold, or by finding the point where the first derivative of the signal reaches its maximum value, thereby generating the wavefront arrival time for that signal component at the i-th sensor. By combining the arrival times of the wavefronts corresponding to this signal component on all N sensors, a set of multi-channel wavefront timestamps is formed.

[0097] Wave head arrival time It can be achieved using either the threshold method or the derivative method, where the threshold method is expressed as:

[0098]

[0099] in, The time when the wavefront arrives at the i-th sensor; Let be the candidate independent signal components of the i-th sensor; The preset threshold is set based on the signal-to-noise level. The setting depends on the signal noise level. For example: first, calculate the noise standard deviation of the candidate independent signal component within a preset time window before the arrival of the wavefront, such as 100 μs before the signal starts, and then set the threshold. Set as ,in For coefficients, generally speaking, Based on experience, the value can be selected within the range of 3 to 8. This method can adapt to signals with different signal-to-noise ratios, ensuring the robustness of wavefront detection;

[0100] The derivative method is expressed as:

[0101]

[0102] in, This refers to the time point corresponding to the maximum absolute value of the derivative; It is the first derivative of the signal.

[0103] Secondly, based on the multi-channel wavefront timestamps, the arrival time difference between different sensor pairs is calculated. For example, for sensor i and sensor j, their arrival time difference is... It is obtained through simple subtraction, that is equal - By selecting multiple different sensor pairs, a time difference dataset containing multiple time differences can be obtained.

[0104] Next, based on this time difference dataset and the preset propagation speed of electromagnetic waves in the underground medium... For each time difference, a hyperboloid equation is constructed. In three-dimensional space, the location of a source... To two sensors and The distance difference is constant, and this constant value is equal to Multiply This definition geometrically forms a hyperboloid with the two sensors as foci. Its equation can be expressed as:

[0105]

[0106] in, These are the coordinates of the source point to be solved. and These are the sensor locations with known geographic coordinates for the i-th and j-th sensors, respectively. This is the preset electromagnetic wave propagation speed, a key parameter for time-of-flight positioning. Its value depends on the electrical parameters of the underground medium, including its dielectric constant and conductivity. In practical applications, this speed can be determined through one of the following methods:

[0107] Based on geological survey data: By reviewing geological exploration reports of the target area or conducting on-site measurements, the electrical parameters of typical media are obtained and estimated. For example, for typical dry soil, the relative permittivity is about 4-9, corresponding to a propagation velocity of about 0.1c-0.15c, or 0.03m / ns-0.045m / ns.

[0108] Based on field calibration: A reference source is placed at a known location in the area to be tested. By measuring the actual time difference of its signal reaching different sensors, the equivalent propagation speed is calculated. This method is preferred for the actual propagation speed in complex media.

[0109] Each time difference corresponds to one of the equations described above, thus forming a set of hyperboloid equations.

[0110] Finally, through numerical optimization, such as the nonlinear least squares method, the set of hyperboloid equations is solved to find the points in space that simultaneously satisfy or effectively fit all the equations. This solution represents the candidate geographic location corresponding to the candidate independent signal component. The candidate geographic locations calculated from all candidate independent signal components through the above complete process are then aggregated to form the final location candidate set.

[0111] Nonlinear least squares objective function:

[0112]

[0113] in That is, the coordinates of the source point to be optimized.

[0114] In one embodiment of the present invention, cross-validation and filtering of the candidate location set based on preset physical and geometric consistency rules includes the following steps:

[0115] Determine whether each candidate geographic location in the candidate location set falls within the preset valid geographic area and depth range;

[0116] Eliminate all candidate geographic locations that fall outside the range to obtain a spatially valid set of candidate locations.

[0117] Specifically, the step of cross-validating and filtering the candidate location set based on preset physical and geometric consistency rules in this invention is the first key line of defense against false location results. This step introduces prior geospatial constraints to verify the physical rationality of the candidate location set generated in the previous step.

[0118] Specifically, the first step is to obtain a pre-defined effective geographical area and depth range. This range is determined based on the actual needs of the exploration mission and common-sense knowledge of the geological structure of the target area. For example, it is known that the burial depth of underground pipelines or facilities will not exceed a certain specific depth, or the horizontal boundary of the exploration area is known. This range defines a "credible region" in three-dimensional space, which can be represented as a set of coordinates. It satisfies:

[0119]

[0120] in, The three-dimensional coordinates of the candidate geographical location, and The minimum and maximum boundaries of the effective region on the x-axis, such as longitude or eastward coordinates; and The minimum and maximum boundaries of the effective region on the y-axis, such as latitude or northward coordinates; and The minimum and maximum boundaries of the effective region along the z-axis, such as depth or elevation; typically, Indicates the maximum depth (negative or absolute value). Indicates the minimum depth or surface elevation (e.g., 0 or a positive value).

[0121] Next, each candidate geographic location in the candidate location set is traversed. For each candidate geographic location, a simple logical judgment is performed, namely, checking whether its coordinates fall within the effective geographic area and depth range, that is, determining whether the candidate geographic location belongs to the set. If the coordinates of a candidate geographic location exceed this preset range—for example, if its z-coordinate is greater than the ground elevation, its depth is far beyond engineering norms, or its horizontal position falls outside the detection area—it will be deemed non-compliant, a false location caused by noise or algorithm error. For example, if the detection area is a 100m × 100m rectangular area, the ground elevation is set to 0, and the target source depth is expected to be between 1m and 20m underground, then the following settings can be configured: =0, =100; =0, =100; and =0; These ranges exclude unreasonable locations above the ground surface and too deep, and then each fitted residual in the residual dataset is compared with a preset residual threshold.

[0122] Finally, candidate geographic locations that are deemed not to conform to physical reality will be directly removed from the candidate location set. After this round of screening, the remaining candidate geographic locations, whose coordinates are all within the effective geographic area and depth range, together constitute a new, smaller but more physically meaningful set, namely the spatially effective candidate location set.

[0123] In one embodiment of the present invention, after obtaining a spatially valid set of candidate locations, the method further includes the following steps:

[0124] Obtain the fitting residuals generated during the solution process for each candidate geographic location in the spatially valid candidate location set, forming a residual dataset;

[0125] Each fitted residual in the residual dataset is compared with a preset residual threshold;

[0126] Candidate geographical locations with fitting residuals greater than the residual threshold are removed, and a set of candidate locations after residual filtering is selected.

[0127] Specifically, after generating a set of spatially valid candidate locations through spatial validity identification, the cross-validation and screening process of this invention enters a deeper stage of solution quality evaluation. This step aims to further filter out candidate geographical locations that, although reasonably located, have low solution quality from the perspective of inherent data consistency.

[0128] First, the fitting residuals generated for each candidate geographic location during the solution of the hyperboloid equations are obtained. The fitting residual is a quantitative indicator that measures the geometric deviation between the final calculated location and the hyperboloid formed by all time-difference data. For candidate geographic locations within the spatially valid candidate location set... The corresponding fitting residual It is usually obtained through the least squares method and can be expressed as the square root of the sum of the squares of the errors of all hyperboloid equations at that point, i.e.:

[0129]

[0130] The summation symbol Σ iterates through all sensor pairs used for positioning calculations; Indicates candidate geographical location To the sensor The Euclidean distance, i.e. , It is the speed of electromagnetic wave propagation. It is a sensor pair The time difference of arrival measured between them. Candidate geographic locations The fitting residuals are used to measure the quality of the localization solution. Collecting the fitting residuals of all spatially valid candidate geographic locations forms the residual dataset.

[0131] Each fitted residual in this residual dataset Compared with the preset residual threshold The comparison is then performed. This residual threshold is an acceptable upper limit of error, set comprehensively based on factors such as accuracy requirements, sensor deployment accuracy, and environmental noise levels. It represents the inherent data consistency standard that the selected positioning solutions should meet.

[0132] Finally, iterate through all spatially valid candidate geographic locations. If the fitting residual of a candidate geographic location is greater than a preset residual threshold, i.e. This indicates that the location was calculated from a set of time difference data with significant internal contradictions that could not intersect at a single point, and the residual threshold... The accuracy of the positioning calculation is used to measure the internal symbol accuracy. Its setting is related to the sensor position error, time synchronization error, and wave velocity error. A reasonable range is 5 meters to 15 meters. =10 meters. Fitting residuals This indicates that the inherent consistency of the time difference data used in the location calculation is poor, and the positioning result is unreliable. This usually means that the wavefront characteristics of the corresponding candidate independent signal components are severely distorted or contaminated by noise. Therefore, this candidate geographical location will be judged as a low-quality solution and eliminated. Through this screening, a subset with higher solution quality is finally selected from the original set of spatially valid candidate locations, namely the candidate location set after residual screening.

[0133] In one embodiment of the present invention, after selecting the candidate position set after residual screening, the method further includes the following steps:

[0134] Calculate the geographical distance between any two candidate geographic locations within the candidate location set after residual screening to obtain the distance matrix;

[0135] Based on the distance matrix and a preset distance threshold, multiple candidate geographic locations that are geographically less than the distance threshold are identified as redundant location clusters.

[0136] Multiple candidate geographic locations within a redundant location cluster are merged into a single location point, which, together with other candidate geographic locations that are not identified as redundant, constitutes a valid location result set.

[0137] Specifically, after the candidate location set is selected through residual screening, the cross-validation and screening process proceeds to the final step: source point redundancy clustering. This step aims to address the source splitting problem that may occur in blind source separation algorithms, where the real physical source is incorrectly decomposed into multiple highly correlated signal components, resulting in multiple redundant location points with very close geographical locations in the candidate location set after residual screening.

[0138] First, the geographical distance between each pair of candidate geographic locations in the candidate location set after residual filtering is calculated. For any two candidate geographic locations in this set... and The geographical distance between them is calculated using the standard three-dimensional Euclidean distance formula:

[0139]

[0140] in, Candidate geographic locations and The three-dimensional Euclidean distance between them.

[0141] By summing up the calculation results for all location pairs, a symmetric distance matrix can be constructed, which contains the spatial distance information between any two points within the set:

[0142] Next, based on this distance matrix and the preset distance threshold... Cluster analysis is performed to identify redundant location clusters. This distance threshold is set based on the theoretical accuracy of the positioning system and the acceptable source point resolution in practical application scenarios. It defines the quantitative standard for "location proximity," i.e., the distance threshold. The spatial resolution that can be considered as the same source point is defined, and its value is related to the theoretical accuracy of the positioning system and engineering requirements. For example, considering the accuracy of TDOA positioning under typical deployment, the distance threshold can be selected in the range of 10 meters to 30 meters, such as 20 meters. This means that multiple candidate locations that are spatially clustered within a 20-meter range will be considered as repeated estimates of the same physical source.

[0143] By traversing the distance matrix, if a set of candidate geographic locations is found where the geographical distances between them are all less than the preset distance threshold, then this set of locations is identified as a redundant location cluster. This indicates that they are very likely multiple location results generated by the same physical source due to source splitting effects.

[0144] For each identified redundant location cluster, a geometric center calculation is performed, merging all candidate geographic locations within the cluster into a single location point. The coordinates of this single location point are typically the arithmetic mean of the coordinates of all location points within the cluster. For example, if a redundant location cluster contains... Location points Then the merged single location point The coordinates are:

[0145]

[0146] The summation symbol Σ traverses all points within the cluster. The coordinates of the merged single location point; The coordinates of the i-th candidate geographic location in the redundant location cluster; This represents the number of location points in the redundant location cluster. This merged single location point is considered a valid estimate of the physical source location. All merged single location points, along with independent candidate geographic locations not identified as any redundant location clusters, constitute the final set of valid location results.

[0147] In one embodiment of the present invention, the estimated number of effective sources is used as a constraint for further blind source separation, including the following steps:

[0148] The number of location points contained in the set of valid location results is counted and determined as the estimated number of valid sources;

[0149] The estimated number of effective sources is used as the target separation quantity parameter for blind source separation processing, and a new round of iterative separation processing is performed on the mixed potential time domain signal.

[0150] Specifically, the step of using the estimated number of effective sources as a constraint for further processing in this invention is a key execution link of the closed-loop mechanism. This step transforms the physical insights obtained in the preceding cross-validation and screening stages into direct and quantitative guidance for the signal separation algorithm.

[0151] Specifically, key information is first extracted from the set of valid location results generated in the previous iteration. The estimated number of valid sources is determined by simply counting the total number of independent location points contained in the set. This estimated number represents the number of real underground interference sources most likely to exist in the current scenario, inferred after multiple physical and geometric rule checks. For example, if the set of valid location results contains three discrete location points, the estimated number of valid sources is 3.

[0152] This estimate of the effective source number will be used as a mandatory input parameter for blind source separation algorithms (such as independent component analysis). In many implementations of blind source separation algorithms, the number of source signals to be separated needs to be specified in advance. In conventional applications, this number is unknown and can only be set by guess or experience, which is often the main reason for poor separation results. In this invention, the estimate of the effective source number is directly set as the target separation number parameter for the blind source separation algorithm. This means that in the next execution, the algorithm's goal is no longer to blindly search for an arbitrary number of independent components, but is constrained to be able to separate only the estimated number of signal components.

[0153] Using this blind source separation algorithm with a clearly defined target separation number parameter, a new round of iterative separation processing is performed on the initial, unprocessed mixed potential time-domain signal. This separation is no longer blind but guided. The algorithm will re-optimize the separation matrix under new constraints, searching for a solution that can decompose the mixed signal into an estimated number of effective sources and statistically independent components.

[0154] In one embodiment of the present invention, the process of performing blind source separation again using the estimated number of effective sources as a constraint condition further includes the following steps:

[0155] From the candidate independent signal components, extract the signal components corresponding to the location points in the effective positioning result set, and use them as reference signal waveforms;

[0156] When performing a new round of iterative separation processing, the group reference signal waveform is used as prior knowledge to guide the convergence direction of the blind source separation processing.

[0157] Specifically, based on constraining the blind source separation algorithm using the effective source quantity estimate, this invention further introduces a deeper feedback mechanism, namely, using the verified effective signal waveform as prior knowledge to guide the subsequent iterative separation process.

[0158] First, it is necessary to establish the correspondence between the valid positioning result set and the candidate independent signal components. In the previous processing flow, each candidate geographic location was calculated from a specific candidate independent signal component. Therefore, we can backtrack and accurately extract the signal components that ultimately generated the location points in the valid positioning result set from the original set of separated candidate independent signal components. These extracted signal components are considered valid and reliable because their positioning results have passed multiple rigorous checks of physical and geometric consistency, and thus constitute a set of reference signal waveforms.

[0159] During the next round of iterative separation processing, this set of reference signal waveforms will be used as prior knowledge to guide the convergence direction of the blind source separation algorithm. The specific implementation depends on the chosen algorithm. For example, when using certain variants of independent component analysis, this set of reference signal waveforms can be used as known sources or fixed-pattern inputs. When optimizing the separation matrix, the algorithm's objective function will be modified to not only improve the independence of all output components but also ensure that a portion of the output components have a high similarity to the provided reference signal waveforms.

[0160] For example, the ICA objective function based on the reference signal:

[0161]

[0162] in, The objective function of standard ICA is, for example, negative entropy; For the first Each separated signal component; For the first A reference signal waveform, which is extracted from the corresponding signal component of the effective positioning result set; This is a regularization parameter that controls the influence of the reference signal.

[0163] It should be noted that the reference signal waveform The regularization parameter originates from the candidate independent signal components that generated the set of valid localization results in the previous iteration. These signal components are considered reliable because their corresponding localization results passed the physical and geometric consistency test; Used to balance standard ICA targets And similarity with the reference signal; The value of can be determined through cross-validation. The typical initial trial range is 0.1 to 1.0, for example, 0.5. This parameter controls the weight of prior knowledge in the separation process. The larger the value, the more the algorithm tends to output components similar to the reference signal.

[0164] Therefore, the new round of iterative separation processing is carried out under dual constraints: the estimated number of effective sources serves as a macro-constraint, determining the target number of separations; while the set of reference signal waveforms serves as a micro-constraint, providing specific waveform morphology guidance for the separation process.

[0165] In one embodiment of the present invention, location result convergence means that the number of valid sources and the corresponding changes in geographical locations obtained in continuous iterations are less than a preset value. Preferably, the criteria for determining location result convergence are as follows: In the feedback iteration process, let the set of valid location results obtained in the nth iteration be denoted as... The effective localization result set obtained in the (n-1)th iteration is First, determine whether the number of valid sources is stable, that is, require that the estimated number of valid sources remains unchanged for m consecutive iterations. Where m is the preset number of stable iterations, typically 2 or 3. Next, based on the stability of the number of iterations, it is further determined whether the source location is stable. The current iteration result is then calculated. Compared with the previous iteration result The average Euclidean distance offset between all paired source locations. If this average offset is less than a preset distance threshold (i.e., a preset value), for example, set to 5m, the positioning result is considered converged, and the iteration stops.

[0166] See appendix Figure 2 The present invention also proposes an underground interference source localization system based on the time difference of impact surge method, comprising the following modules:

[0167] The signal acquisition module is used to acquire the mixed potential time-domain signal synchronously collected by multiple sensors distributed in the area to be measured;

[0168] The blind source separation module is used to perform preliminary blind source separation processing on the mixed potential time-domain signal to obtain candidate independent signal components;

[0169] The time difference positioning module is used to perform time difference positioning processing on candidate independent signal components and generate a candidate set of locations corresponding to the candidate independent signal components.

[0170] The cross-validation module performs cross-validation and filtering on the candidate location set to obtain an estimate of the number of valid sources.

[0171] The feedback iteration module uses the estimated number of effective sources as a constraint to perform blind source separation and time difference localization processing on the mixed potential time domain signal again;

[0172] The location module obtains the geographical location of the underground interference source based on the output of the feedback iteration module until the location result converges.

[0173] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0174] It should be noted that the human information (including but not limited to human device information and personal information) and data (including but not limited to data used for analysis, data stored and data displayed) involved in this invention are all information and data authorized by the human body or fully authorized by all parties. The collection, use and processing of related data require relevant legal standards.

[0175] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for locating underground interference sources using the time-difference method of shock surges, characterized in that, Includes the following steps: S1. Acquire the mixed potential time-domain signal synchronously collected by multiple sensors distributed in the area to be measured; S2. Perform preliminary blind source separation processing on the mixed potential time domain signal to obtain candidate independent signal components, and perform time difference localization processing on the candidate independent signal components to generate a candidate set of positions corresponding to the candidate independent signal components. S3. Perform cross-validation and filtering on the candidate location set to obtain an estimate of the number of valid sources; S4. Using the estimated number of effective sources as constraints, perform blind source separation and time difference localization processing on the mixed potential time domain signal again until the localization results converge, and determine the geographical location of the underground interference source. The convergence of the localization results means that the change in the number of effective sources and the corresponding geographical locations obtained in the continuous iteration is less than the preset value.

2. The method for locating underground interference sources using the time-difference method of shock surges according to claim 1, characterized in that, Before performing preliminary blind source separation processing on the mixed potential time-domain signal, the following steps are also included: The mixed potential time-domain signal is subjected to bandpass filtering to remove low-frequency power frequency interference and high-frequency random noise, resulting in a filtered signal. The filtered signal is processed by first-order differentiation to sharpen the rising edge characteristics of the signal wavefront and generate a hybrid signal with enhanced features. The feature-enhanced hybrid signal is used as the input to the hybrid potential time-domain signal.

3. A method for locating underground interference sources using the time-difference method of shock surges according to claim 1 or 2, characterized in that, Preliminary blind source separation processing, based on transient feature-driven independent component analysis, includes the following steps: The mixed potential time-domain signals or feature-enhanced mixed signals from multiple sensors are used to construct an observation signal matrix; The observed signal matrix and the separation matrix obtained after iterative optimization are linearly transformed to obtain the separated signal matrix; Each row of the signal matrix is ​​output as a statistically independent candidate independent signal component.

4. The method for locating underground interference sources using the time-difference method of shock surges according to claim 1, characterized in that, Generating a candidate set of locations includes the following steps: For each candidate independent signal component, the arrival time of its wavefront to each sensor is extracted, and a multi-channel wavefront timestamp is generated; The arrival time difference between different sensor pairs is calculated based on the multi-channel wavefront timestamps to obtain time difference data; Based on time difference data and the preset electromagnetic wave propagation speed, a set of hyperboloid equations is constructed; Solve the hyperboloid equations to obtain the candidate geographic locations corresponding to the candidate independent signal components, and summarize all candidate geographic locations to form a candidate location set.

5. The method for locating underground interference sources using the time-difference method of shock surges according to claim 1, characterized in that, Cross-validation and filtering of the candidate locations include the following steps: Determine whether each candidate geographic location in the candidate location set falls within the preset valid geographic area and depth range; Eliminate all candidate geographic locations that fall outside the range to obtain a spatially valid set of candidate locations.

6. The method for locating underground interference sources using the time-difference method of shock surges according to claim 5, characterized in that, After obtaining the spatially valid candidate location set, the following steps are also included: Obtain the fitting residuals generated during the solution process for each candidate geographic location in the spatially valid candidate location set, forming a residual dataset; Each fitted residual in the residual dataset is compared with a preset residual threshold; Candidate geographical locations with fitting residuals greater than the residual threshold are removed, and a set of candidate locations after residual filtering is selected.

7. The method for locating underground interference sources using the time-difference method of shock surges according to claim 6, characterized in that, After selecting the candidate location set based on residual filtering, the following steps are also included: Calculate the geographical distance between any two candidate geographic locations within the candidate location set after residual screening to obtain the distance matrix; Based on the distance matrix and a preset distance threshold, multiple candidate geographic locations that are geographically less than the distance threshold are identified as redundant location clusters. Multiple candidate geographic locations within a redundant location cluster are merged into a single location point, which, together with other candidate geographic locations that are not identified as redundant, constitutes a valid location result set.

8. The method for locating underground interference sources using the time-difference method of shock surges according to claim 7, characterized in that, Using the estimated number of effective sources as a constraint, a second blind source separation process is performed, including the following steps: The number of location points contained in the set of valid location results is counted and determined as the estimated number of valid sources; The estimated number of effective sources is used as the target separation quantity parameter for blind source separation processing, and a new round of iterative separation processing is performed on the mixed potential time domain signal.

9. The method for locating underground interference sources using the time-difference method of shock surges according to claim 8, characterized in that, The further blind source separation process also includes the following steps: From the candidate independent signal components, extract the signal components corresponding to the location points in the effective positioning result set, and use them as reference signal waveforms; When performing a new round of iterative separation processing, the group reference signal waveform is used as prior knowledge to guide the convergence direction of the blind source separation processing.

10. A system for locating underground interference sources using the time-difference method for shock surges, characterized in that, Includes the following modules: The signal acquisition module is used to acquire the mixed potential time-domain signal synchronously collected by multiple sensors distributed in the area to be measured; The blind source separation module is used to perform preliminary blind source separation processing on the mixed potential time-domain signal to obtain candidate independent signal components; The time difference positioning module is used to perform time difference positioning processing on candidate independent signal components and generate a candidate set of locations corresponding to the candidate independent signal components. The cross-validation module performs cross-validation and filtering on the candidate location set to obtain an estimate of the number of valid sources. The feedback iteration module uses the estimated number of effective sources as a constraint to perform blind source separation and time difference localization processing on the mixed potential time domain signal again; The location module obtains the geographical location of the underground interference source based on the output of the feedback iteration module until the location result converges.

Images