An acoustic emission-electromagnetic fusion underwater rock mass crack positioning and repairing method

Abstract

The application provides an underwater rock mass crack positioning and repairing method based on acoustic emission and electromagnetic fusion, acquires acoustic emission signals of underwater rock mass through an acoustic emission monitoring module, extracts acoustic emission features, and determines a crack area based on a time difference positioning algorithm; applies current excitation in the crack area through an electromagnetic imaging module, and acquires potential responses through an electrode array; establishes an electromagnetic field inverse problem model based on the potential responses, and obtains a conductivity characteristic parameter; determines a crack grade based on the acoustic emission features and the conductivity characteristic parameter, and generates a repairing instruction when the crack grade exceeds a preset threshold; and performs reinforcement work in the crack area according to the repairing instruction. Multiple modules complement each other, reduce misjudgment and missed detection, and improve the reliability and resolution of crack identification.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underwater rock crack positioning and repairing, and particularly relates to an underwater rock mass crack positioning and repairing method based on acoustic emission-electromagnetic fusion. BACKGROUND

[0002] In underwater rock mass structures such as deep-sea engineering, submarine tunnels and offshore platforms, the integrity and stability of the rock directly affect the safety of the project and the long-term operation and maintenance costs. For a long time, the monitoring and repairing of underwater rock mass cracks mainly rely on artificial diving inspection, single non-destructive testing method or semi-automatic operation means, but all have the disadvantages of response lag, inaccurate positioning and low repairing efficiency.

[0003] In the aspect of crack monitoring, although the traditional acoustic emission (AE) technology can capture the weak acoustic wave signals generated by cracks and locate them, in the complex underwater environment, it is difficult to further improve the decimeter-level positioning accuracy due to the interference of water flow, noise and multipath propagation, and the single acoustic emission data is prone to false positives, which reduces the reliability and real-time performance of the early warning. Another commonly used electromagnetic imaging and resistivity inversion technology can realize the visualization of crack orientation and opening through external electromagnetic excitation or electrode measurement, but it is extremely sensitive to the position and spacing of the excitation coil or electrode, and the spatial resolution usually stays at the meter level, which is difficult to capture subtle cracks. In addition, the existing electromagnetic monitoring means usually use fixed excitation frequency and current amplitude, which cannot be dynamically adjusted according to the real-time active state of the internal cracks of the rock mass. This "static" detection method often has the problems of mismatch between the detection frequency band and the crack size, or insufficient signal strength during the rapid development of cracks, resulting in low signal-to-noise ratio of the image, loss of effective information, and being greatly affected by changes in environmental parameters such as seawater salinity and temperature, which is prone to missed detection or false detection. The high cost of debugging and repeated calibration also limits its wide online application.

[0004] In the aspect of crack repairing, most projects still rely on professional divers or semi-autonomous ROV / AUV carrying tools to implement local repair. Artificial diving inspection has the problems of long cycle, dangerous operation environment and low efficiency; while the ROV / AUV in actual operation often needs to be remotely controlled by ground operators, and cannot completely escape human intervention, resulting in lagging repair and difficulty in forming a self-adaptive closed loop. In addition, a single repair method cannot be flexibly adjusted for different crack types and damage levels, and often needs to be operated several times to achieve the expected effect, further increasing the maintenance cost.

[0005] Although acoustic emission, electromagnetic imaging and manual or semi-automatic repair each have certain advantages, they lack data fusion and intelligent decision support, making it difficult to implement the "monitoring-locating-decision-making-repair-inspection" closed-loop process. Traditional systems have bottlenecks in large-scale continuous online monitoring, accurate crack severity discrimination and rapid response repair, and cannot meet the comprehensive requirements of modern underwater engineering for high precision, high timeliness, safety and economy. SUMMARY

[0006] The present application provides an acoustic emission-electromagnetic fusion underwater rock mass crack locating and repairing method to solve the technical problems of insufficient detection accuracy, single determination method, fixed imaging parameters and difficulty in closed-loop implementation of detection and repair in the prior art.

[0007] To solve the above technical problems, the present application provides an acoustic emission-electromagnetic fusion underwater rock mass crack locating and repairing method, comprising the following steps:

[0008] Step S1: Acoustic emission signals of the underwater rock mass are collected by an acoustic emission monitoring module, acoustic emission features are extracted, and a crack area is determined based on a time difference positioning algorithm;

[0009] Step S2: An electromagnetic imaging module is used to apply current excitation in the crack area, and potential responses are collected by an electrode array;

[0010] Step S3: An electromagnetic field inverse problem model is established based on the potential responses to obtain conductivity characteristic parameters;

[0011] Step S4: Crack levels are determined based on acoustic emission features and conductivity characteristic parameters, and a repair instruction is generated when the crack level exceeds a preset threshold;

[0012] Step S5: Reinforcement work is performed on the crack area according to the repair instruction.

[0013] Preferably, the acoustic emission monitoring module includes not less than four piezoelectric acoustic emission sensors that are waterproof packaged and coupled with the underwater rock mass surface through soft coupling pads, and the sensors are arranged in a symmetrical array with a spacing of 1-2 m and coupled with the underwater rock mass surface through mechanical clamping and soft coupling pads.

[0014] Preferably, the electromagnetic imaging module includes:

[0015] A matrix arrangement of a transmitting coil array, the transmitting coil being spaced apart from the crack area by a distance of 0.5-1 m and supporting multi-frequency alternating excitation;

[0016] A two-dimensional electrode array attached to the surface of the crack area, the electrode spacing being 0.5-1 m, and the arrangement direction being orthogonal to the transmitting coil array.

[0017] Preferably, the Tikhonov regularization method is used in step S3, and the expression of the inversion objective function is as follows:

[0018] ;

[0019] wherein, represents a system matrix; represents a response vector measured by the electromagnetic imaging module; represents a regularization parameter; represents a to-be-solved vector, i.e., a conductivity image;

[0020] Based on the inversion result, a conductivity distribution map is extracted, and a conductivity characteristic parameter is analyzed, the conductivity characteristic parameter including a conductivity mutation amplitude, an abnormal area proportion, and a resistivity gradient characteristic.

[0021] Preferably, the first determination unit is used to determine the crack grade in step S4.

[0022] The first determination unit constructs a threshold decision tree based on the acoustic emission event count in the acoustic emission characteristic and the conductivity mutation amplitude and the abnormal area proportion in the conductivity characteristic parameter, and outputs the crack grade.

[0023] The threshold decision tree includes:

[0024] When the acoustic emission event count per unit time is less than a first threshold value, it is determined that there is no crack or false positive.

[0025] When the acoustic emission event count per unit time is greater than or equal to the first threshold value and the conductivity mutation amplitude is less than a second threshold value, it is determined that there is a mild crack.

[0026] When the acoustic emission event count per unit time is greater than or equal to the first threshold value, the conductivity mutation amplitude is greater than or equal to the second threshold value, and the abnormal area proportion is less than a third threshold value, it is determined that there is a moderate crack.

[0027] When the acoustic emission event count per unit time is greater than or equal to the first threshold value, the conductivity mutation amplitude is greater than or equal to the second threshold value, and the abnormal area proportion is greater than or equal to the third threshold value, it is determined that there is a severe crack.

[0028] Preferably, the second determination unit is used to determine the crack grade in step S4.

[0029] The second determination unit is configured to perform weighted fusion on the acoustic emission characteristic and the conductivity characteristic parameter, and compare the weighted fusion result with a set threshold value to determine the crack grade.

[0030] The weights and the threshold value are optimized through self-learning of historical sample data during training.

[0031] Preferably, the expression for the weighted fusion of the acoustic emission feature and the electrical conductivity feature parameter is:

[0032] ;

[0033] ;

[0034] wherein, represents the comprehensive score; represents the weight coefficient, = 1, 2, 3, 4, 5; represents the acoustic emission event count per unit time; represents the maximum acoustic emission event number allowed for safe operation; represents the average energy of acoustic emission events per unit time; represents the maximum observed energy; represents the maximum variation amplitude of electrical conductivity; represents the maximum variation amplitude of resistivity history; represents the area of the abnormal region detected in the electromagnetic image; represents the total area of the entire monitoring region; represents the maximum abnormal area proportion threshold in the resistivity image; represents the average gradient of the resistivity image; the historical maximum value of the average gradient of the resistivity image.

[0035] Preferably, the second determination unit uses the weighted least square error as the loss function when training.

[0036] Preferably, the step S4 further comprises an adaptive excitation step: dynamically adjusting the excitation frequency range of the electromagnetic imaging module according to the acoustic emission energy, and adjusting the excitation current amplitude according to the acoustic emission event count.

[0037] Preferably, the expression for the excitation frequency range is:

[0038] ;

[0039] wherein, represents the minimum excitation frequency; represents the maximum excitation frequency; represents the center frequency; represents the maximum allowed expansion bandwidth; represents the maximum observed energy; represents the average energy of acoustic emission events per unit time;

[0040] the expression for the excitation current amplitude is:

[0041] ;

[0042] ;

[0043] wherein, represents the default current; represents the acoustic emission event count per unit time; represents the acoustic emission count threshold; represents the maximum number of acoustic emission events allowed for safe operation; represents the adjustment factor of the maximum allowed current amplitude.

[0044] The beneficial effects of the present application include at least:

[0045] 1) High detection accuracy: The multi-channel time difference positioning of the acoustic emission module quickly captures micro-crack signals, preliminarily determining the suspicious crack area; then the multi-frequency excitation and electrode array measurement of the electromagnetic imaging module are combined with the regularization inversion algorithm to extract the conductivity characteristic parameters, realizing fine crack identification. AE and EM complement each other, reducing misjudgment and missed detection, and improving the reliability and resolution of crack identification.

[0046] The present application realizes the linkage control of acoustic emission and electromagnetic imaging by introducing an adaptive excitation mechanism. It can dynamically adjust the frequency range of electromagnetic excitation according to the energy intensity captured by acoustic emission, and adjust the excitation current amplitude in real time according to the density of acoustic emission events. This mechanism enables the system to automatically match the best frequency band to capture subtle features in the early stage of cracks, and to improve the penetration depth and imaging contrast in the active expansion period of cracks. Compared with the traditional fixed parameter detection method, it improves the recognition accuracy and environmental adaptability of cracks in different evolution stages.

[0047] 2) Intelligent decision-making: The control and decision-making module integrates multi-dimensional features such as acoustic emission count, crack energy, conductivity mutation amplitude, and abnormal area proportion, and realizes automatic grading judgment using threshold decision tree and multi-modal evaluation function. When it is determined as moderate or severe crack, it can automatically generate repair instructions, avoiding human intervention and improving response speed and judgment objectivity.

[0048] 3) Repair automation: The system relies on the underwater robot platform to carry mechanical arms and various repair tools (grouting device, spray head, flexible sealing device, etc.), which can implement remote or autonomous repair for target crack areas, avoiding manual diving operation and significantly improving operation efficiency and safety.

[0049] 4) Closed-loop control: After repair is completed, AE or EM modules can be called again for re-inspection, comparing the acoustic emission activity and conductivity characteristic parameters before and after repair to verify the repair effect, thereby forming a closed-loop process of "monitoring - positioning - decision-making - repair - re-inspection", realizing the continuous closed-loop management of the health status of underwater structures.

[0050] 5) Applicability and Engineering Value: It has high integration, strong detection sensitivity, and good adaptability. It is especially suitable for early crack diagnosis and maintenance of important underwater rock structures such as deep-sea tunnels, seabed rock foundations, and offshore platforms, providing an efficient and reliable technical solution for improving the safe operation and intelligent maintenance capabilities of marine engineering. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the system structure according to an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram of the underwater acoustic coupling piezoelectric acoustic emission sensor array structure according to an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of the electromagnetic excitation coil arrangement of the electromagnetic imaging module according to an embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of the electrode array arrangement of the electromagnetic imaging module according to an embodiment of the present invention;

[0055] Figure 5 This is a schematic diagram of the logical decision-making of the first determination module in an embodiment of the present invention;

[0056] Figure 6 This is a schematic diagram illustrating the usage process of the system according to an embodiment of the present invention. Detailed Implementation

[0057] 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, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0058] like Figure 1 As shown, this embodiment of the invention provides an underwater rock mass crack location and repair system that integrates acoustic emission and electromagnetic fusion, including an acoustic emission monitoring module, an electromagnetic imaging module, an image inversion module, a control and decision module, and an underwater repair module installed on an underwater work platform, and correspondingly proposes an underwater rock mass crack location and repair method that integrates acoustic emission and electromagnetic fusion.

[0059] For example, the present invention provides an underwater rock fracture location and repair system that integrates acoustic emission and electromagnetic fusion, which is installed on a remotely operated underwater vehicle (ROV) or autonomous underwater vehicle (AUV) platform, but this is not intended to limit the embodiments of the present invention.

[0060] An acoustic emission monitoring module (AE module) is configured to collect acoustic emission signals of the underwater rock mass, extract acoustic emission features, and determine the crack region based on a time difference positioning algorithm.

[0061] Specifically, the module includes a plurality of waterproof packaged piezoelectric acoustic emission sensors arranged in an array and coupled to the underwater rock mass through a soft coupling pad. The array is installed on the end of a mechanical arm or a deployment support and arranged close to the surface of the rock mass to be measured. The main function of the module is to collect acoustic emission signals generated during the micro-crack propagation and stress release of the underwater rock mass, record waveforms through a multi-channel signal acquisition device, and calculate the propagation delay to achieve preliminary positioning and identification of suspicious damage areas.

[0062] The arrangement principle of the AE sensor array in the acoustic emission monitoring subsystem is shown in FIG. 1, and the arrangement points include: Figure 2

[0063] Arrangement points:

[0064] 1) More than three points in an array, usually ≥4: a minimum triangular structure is used for two-dimensional positioning, and four or more points form redundancy to improve positioning accuracy and stability.

[0065] 2) Array spacing: 1-2 m: This spacing can balance resolution and coverage, and is suitable for underwater mesoscale rock surfaces.

[0066] 3) Symmetrically distributed around the potential risk area: The sensors are arranged to surround the suspicious area or tectonic belt as much as possible to avoid one-sided positioning errors.

[0067] 4) Good coupling with the rock surface: The installation method ensures that the acoustic signals can be effectively transmitted from the rock mass to the sensor, which can be completed by mechanical clamping and a soft coupling pad.

[0068] 5) Avoid noise source areas as much as possible: such as sea currents, propellers, and tool areas to reduce false triggers.

[0069] The purpose is to determine the crack source position through time difference positioning (TDOA) and capture high-frequency pulse events.

[0070] An electromagnetic imaging module (EM module) is configured to apply current excitation in the crack region and collect potential responses through an electrode array.

[0071] Specifically, the electromagnetic imaging module includes a set of coupled alternating current transmission coil arrays and a set of rock surface electrode arrays, wherein the transmission coils are vertically arranged and installed on the mechanical arm / deployment support, maintaining a fixed distance of 0.5-1 m from the rock surface. The electrode array is made of flexible conductive electrodes and is attached to the rock surface, and is distributed in a two-dimensional plane, with a 3×3 or more array for synchronous acquisition of the rock surface potential response under alternating electric field.

[0072] ​The arrangement principles of the electromagnetic excitation coil and the electrode array in the electromagnetic imaging module are shown in Figs. Figure 3 and Figure 4

[0073] The arrangement points of the transmitting coil array include:

[0074] 1) The height is 0.5-1 m away from the rock surface, and a mechanical arm or a support is used to keep the constant distance;

[0075] 2) The grid / matrix arrangement is used, and a 3*3 array is shown in Fig. Figure 3 , which covers the suspicious area of the AE mark;

[0076] 3) Multi-frequency excitation can be used, each coil can work in turn, or the excitation can be alternated to improve the resolution.

[0077] The arrangement points of the electrode measurement array include:

[0078] 1) The electrodes are arranged close to the rock surface or arranged along the edge of the rock surface;

[0079] 2) The measurement points are dense enough, and the typical interval is 0.5-1 m, so that the spatial resolution of the inversion image is ensured;

[0080] 3) The arrangement direction is orthogonal to the coil array, which is more conducive to multi-directional imaging;

[0081] 4) The electrode contact must be good, and the water body or the surface should be free of insulating attachments.

[0082] Overall, the imaging area should cover the periphery of 0.5-1 m of the AE marked area, so as to avoid missing the crack edge. The purpose is to obtain the crack map of the conductive / insulating anomaly through multi-frequency electromagnetic excitation and potential measurement.

[0083] The image inversion module is used to establish an electromagnetic field inverse problem model based on the potential response, and to obtain the conductivity characteristic parameter.

[0084] Specifically, the image inversion module obtains the local voltage response based on the electrode array, and inversely calculates the conductivity characteristic map of the target rock mass area by constructing an electromagnetic field inverse problem model. In the embodiment of the application, the Tikhonov regularization algorithm is used for conductivity image inversion. The algorithm suppresses the ill-posedness and noise interference of the inverse problem by introducing a regularization term, and ensures the stability and physical reasonableness of the inversion image.

[0085] Specifically, the system matrix A is an operator linearized from the mapping relationship between the conductivity distribution x and the corresponding electrode potential b, the data include the excitation parameters of the transmitting coil, the spatial arrangement of the coil and the electrode array, and the finite element modeling result based on the Maxwell equation; the response vector measured by the electrode is b, and the conductivity image is the to-be-solved vector x, and the inversion objective function is: ​

[0086] ;

[0087] where λ is a regularization parameter used to control the smoothness of the solution. The analytical solution of the optimization problem is:

[0088] ;

[0089] The inversion process in the embodiment of the application can be run based on an embedded processor, a cloud platform or an image processing module, supports real-time data input and atlas output, and automatically completes crack identification and imaging processing.

[0090] The control and decision module determines the crack grade based on the acoustic emission feature and the conductivity feature parameter, and generates a repair instruction when the crack grade exceeds a preset threshold.

[0091] Specifically, in the embodiment of the application, the control and decision module is installed on an ROV host or a background control platform, and includes a signal acquisition card, a current source controller, an image inversion module and a decision algorithm module. The control unit receives acoustic emission data of the AE module and locates the damage area, controls the electromagnetic excitation process and performs crack imaging reconstruction at the same time, and automatically triggers the repair mechanism after identifying the existence of structural cracks.

[0092] Exemplarily, the control and decision module in the embodiment of the application is a first decision unit or a second decision unit, wherein the first decision unit constructs a threshold decision tree based on the acoustic emission event count in the acoustic emission feature and the conductivity mutation amplitude and the abnormal area proportion in the conductivity feature parameter, and outputs the crack grade; the second decision unit is used for weighting fusion of the acoustic emission feature and the conductivity feature parameter, and compares with a set threshold to determine the crack grade.

[0093] Specifically, the first decision unit integrates multi-parameter feature extraction and automatic classification judgment mechanism based on a rule tree in the signal processing and decision subsystem.

[0094] The acoustic emission subsystem constructed by the above modules real-time monitors the micro crack activity or local damage event inside the rock mass, and extracts the acoustic emission features, including:

[0095] 1) Acoustic emission event count : The number of acoustic emission events recorded in a unit time, reflecting the crack activity level. The higher the crack activity or growth.

[0096] 2) Event energy : The average energy of acoustic emission events in a unit time, which can be calculated by the amplitude and duration of the AE signal. The greater the energy, the greater the crack energy release and the more intense the damage.

[0097] 3) Cluster location coordinates: The spatial location of AE sources is deduced by TDOA algorithm. If multiple events are clustered in a point or area, it indicates that the crack is growing intensively.

[0098] Through the magnetic imaging module and the image inversion module, the conductivity distribution map of the crack area is inverted, and the extracted conductivity characteristic parameters include:

[0099] 1) Maximum change amplitude of conductivity in the image , expressed as the difference between the local resistivity peak value and the surrounding background, The larger the value, the stronger the insulating crack.

[0100] 2) Abnormal area ratio : The proportion of the abnormal area to the entire observation area, referred to as "crack ratio", if it exceeds a certain threshold, it means that the problem is serious, represents the area of the abnormal area detected in the electromagnetic imaging map, represents the total area of the entire monitoring area (electromagnetic imaging coverage).

[0101] 3) Resistivity gradient feature: The crack edge will produce a resistivity mutation, and the gradient of these changes or "boundary definition" is extracted to assist in determining whether the crack profile is clear.

[0102] The decision tree is a highly interpretable machine learning model, and each judgment node corresponds to a threshold judgment, and finally gives a classification result. In the embodiments of the present application:

[0103] The judgment conditions include:

[0104] 1) > Indicates that acoustic emission is abnormally frequent;

[0105] 2) > Indicates that the conductivity changes dramatically;

[0106] 3) > Indicates that the crack range is large.

[0107] Wherein: is the acoustic emission count threshold, such as 10 times / min; is the resistivity change threshold, such as > 30%; is the crack image abnormal area ratio, such as > 0.2.

[0108] The output classification is shown in Table 1, and the determination process is shown in Figure 5 :

[0109] Table 1

[0110]

[0111] When a crack is determined to be of moderate or severe severity, the system will automatically trigger a repair warning signal.

[0112] The system incorporates a decision tree model, comparing each feature with preset thresholds and sequentially determining the current rock mass damage level using logical nodes. When the overall judgment result indicates moderate or severe cracking, the system automatically generates a repair early warning signal and invokes an underwater repair unit to perform targeted reinforcement work, forming a closed-loop structural diagnosis and response mechanism.

[0113] This invention also provides a second determination module and a multimodal damage evaluation function. The steps involved in fusing the multidimensional features of AE and EM into a unified index for grade determination and self-learning optimization include:

[0114] 1) Normalization of various features

[0115] ;

[0116] in The average gradient of the resistivity image is given by the denominator, which is the historical maximum value. This indicates the largest historical change in resistivity; This represents the maximum percentage threshold of abnormal regions in a resistivity image.

[0117] 2) Assign weights and calculate the overall score

[0118]

[0119] The weights are initially set as follows: =0.30, =0.25, =0.20, =0.20, =0.05, where the coefficient for counting AE in this embodiment of the invention is... Set to maximum because crack activity is the most direct indicator of damage; the next highest value is the coefficient of AE energy. A coefficient reflecting the fracture strength; the magnitude of the change in conductivity. Affecting image contrast; coefficient of abnormal proportion Coefficients reflecting crack size and gradient characteristics. It helps to enhance edge sharpness, but has little impact on the overall rating, so it is set to the lowest level.

[0120] In this embodiment of the invention, in order to achieve the evaluation function of multimodal modes The risk grading and adaptive tuning, the signal processing and decision subsystem further sets a three-stage threshold and learning optimization mechanism, the specific steps are as follows:

[0121] 1) Threshold definition

[0122] The system sets three increasing threshold parameters in advance:

[0123] ;

[0124] Respectively used to define mild cracks, moderate cracks and severe cracks.

[0125] 2) Classification rules

[0126] When the fusion score function is calculated, according to the comparison result of and the threshold, the following classification is executed:

[0127] If , it is determined as a mild crack, and the system only continues acoustic emission and local electromagnetic monitoring, without immediate intervention;

[0128] If , it is determined as a moderate crack, triggering primary reinforcement measures such as low-pressure epoxy grouting;

[0129] If , it is determined as a severe crack, triggering high-priority anchor implantation, micro-cement filling and other reinforcement;

[0130] If , it can be further classified as an extremely severe crack and an emergency manual intervention alarm is issued.

[0131] For most engineering applications, severe means that the highest priority conventional reinforcement measures such as anchor reinforcement need to be started; the extremely severe subdivision is often triggered only under very rare or special safety events. Therefore, only and are analyzed in the following description.

[0132] The crack characteristics, environmental noise and measurement errors of each engineering site are different, and a scheme that simply relies on artificial setting of empirical weights and determination thresholds often cannot take into account all scenarios. Therefore, on the basis of the above embodiment, the K-Fold cross-validation and least squares optimization are introduced to adjust the weights and determination thresholds, and the method comprises:

[0133] 1) Data preparation: collect a multi-modal data sample set for typical crack cases in the system's historical operation, including the corresponding AE / EM feature vectors and expert hand-labeled true crack grades, and label true cracks and false cracks.

[0134] 2) K-fold cross validation: evenly split the sample set into K parts, and cycle through selecting K-1 parts for training and 1 part for validation, repeat K-fold training / validation to evaluate the classification performance of different combinations of A and T, to ensure the model does not overfit and can perform well on different subsets. 1) K-fold cross validation: evenly split the sample set into K parts, and cycle through selecting K-1 parts for training and 1 part for validation, repeat K-fold training / validation to evaluate the classification performance of different combinations of A and T, to ensure the model does not overfit and can perform well on different subsets. 1) K-fold cross validation: evenly split the sample set into K parts, and cycle through selecting K-1 parts for training and 1 part for validation, repeat K-fold training / validation to evaluate the classification performance of different combinations of A and T, to ensure the model does not overfit and can perform well on different subsets.

[0135] 3) Loss function selection: for the classification result and the true label, weighted least squares error is used as the loss function:

[0136] ;

[0137] wherein is the weight of the multi-modal evaluation function; represents the number of samples; is the comprehensive score of the i-th sample; is the true score of the sample. The probability or score output by the model is regarded as a continuous value, and the difference between the label and the square sum is minimized to adjust and T.

[0138] In this way, the system can automatically and based on data find the group of , and that can best distinguish light / medium / heavy cracks.

[0139] An underwater repair module is used to reinforce the crack area according to the repair instruction.

[0140] Specifically, the system of the embodiment of the present application integrates underwater repair devices such as crack grouting devices, flexible adhesive nozzles, and low-pressure sealers, and the repair action is performed by a mechanical arm, and after the repair is completed, AE or EM detection is performed again to verify the repair effect.

[0141] The system of the embodiment of the present application can also include an EM module excitation parameter adaptive generation function, further improving the measurement accuracy of the system of the present application.

[0142] 1) Excitation frequency range selection

[0143] AE event energy The greater, usually means that the crack releases stronger fracture energy, which may be accompanied by a wider frequency band of high or low frequency components. Therefore, the EM excitation frequency band of the embodiment of the present application is dynamically expanded around a center frequency , the greater the energy, the wider the bandwidth, and the better the nonlinear electromagnetic response can be captured.

[0144] The following excitation frequency range selection formula is proposed:

[0145] ;

[0146] is the maximum observed energy of the system history or pre-calibration, which is used to normalize , to ensure is always in the interval [0, 1], so that no matter how the energy changes, the expansion coefficient will not exceed [0, ], ensuring that the wideband adjustment is controllable and gradual. is the maximum value used to control the bandwidth expansion. is the maximum allowed expansion bandwidth set according to engineering requirements at the design time, if the system is expected to expand ±500Hz bandwidth at high energy events, then .

[0147] 2) Excitation current amplitude adjustment

[0148] AE event count The higher the AE event count means the higher the crack activity, which may require a stronger electromagnetic field excitation, i.e. a larger current, to improve the imaging contrast and penetration depth. Therefore, the following excitation current amplitude adjustment formula is proposed:

[0149] ;

[0150] is the warning threshold, when , , i.e. the default current. is the maximum AE event number allowed for safe operation, which is used to normalize to [0, 1]. The coefficient is used to control the sensitivity of the current amplitude increase, and the configuration selection is the same as . Increasing can make the current more sensitive to AE activity, but also prevent excessive energy consumption or safety risks.

[0151] To avoid unintended behavior of current reduction when the warning is not triggered, the excitation current amplitude adjustment module further performs "lower limit truncation" processing on the normalized ratio:

[0152] ;

[0153] The embodiment adds excitation current amplitude adjustment and excitation frequency range selection, ensures signal-to-noise ratio and imaging intensity through excitation current amplitude adjustment, and balances resolution and penetration depth through excitation frequency range selection, so as to realize adaptive optimization of electromagnetic imaging.

[0154] The system of the embodiment of the application is used, and the flowchart is as shown in Figure 6 .​

[0155] Step one: preliminary arrangement

[0156] ROV or AUV carrying the system dives to the area of the rock surface to be tested, starts the acoustic emission monitoring module, and the mechanical arm expands and deploys the AE array, which is close to or attached to the rock surface, for continuous monitoring.

[0157] Step two: acoustic emission warning and suspicious area identification

[0158] The AE sensor array continuously collects environmental acoustic emission event data, and calculates the acoustic source positioning results through multi-channel analysis algorithm. When obvious acoustic emission aggregation or crack propagation events are detected, the system automatically judges the suspicious damage area and records the coordinate position.

[0159] Step three: electromagnetic excitation imaging positioning and confirmation

[0160] The mechanical arm deploys the EM module in front of the suspicious area, and the transmit coil array is arranged vertically facing the rock surface at a distance of 0.5-1m, while the electrode array is attached to the rock surface. The system controls the coil to excite alternating current of multiple frequencies in turn, inducing electromagnetic field to penetrate the rock mass and form potential response on the electrode surface.

[0161] Step four: crack identification and inversion of atlas

[0162] The control unit collects the voltage response of the electrode array, and reconstructs the local conductive structure atlas through the preset conductivity / insulation inversion algorithm, such as the finite element forward and inverse model. Since the rock crack disturbs the current path, the system can accurately draw the crack distribution map and judge its geometric characteristics and development trend.

[0163] Step five: intelligent decision and repair triggering

[0164] If the system determines that the crack size, shape or conductivity parameter exceeds the safety threshold, it automatically triggers the repair warning. The mechanical arm starts the crack repair tool to perform glue injection, grouting, sealing or reinforcement treatment on the target area.

[0165] Step six: post-repair inspection

[0166] After the repair is completed, the AE or EM module is activated again to re-inspect the same area, compare the damage parameter changes, and ensure that the repair effect meets the standard.

[0167] The technical features of the above embodiments can be combined in any way. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, only the preferred embodiments of the present application are expressed, which are more specific and detailed, but it cannot be understood as limiting the scope of the present application. As long as the combination of these technical features does not exist contradictory, it should be considered as the scope of the present application.

[0168] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A method for locating and repairing underwater rock fractures using a combination of acoustic emission and electromagnetic methods, characterized in that: The method comprises the following steps: Step S1: Collecting acoustic emission signals of the underwater rock mass by an acoustic emission monitoring module, extracting acoustic emission features, and determining a crack area based on a time difference positioning algorithm; Step S2: Applying current excitation in the crack area by an electromagnetic imaging module, and collecting potential response by an electrode array; Step S3: Establishing an electromagnetic field inverse problem model based on the potential response to obtain a conductivity characteristic parameter; Step S4: Determining a crack grade based on the acoustic emission features and the conductivity characteristic parameter, and generating a repair instruction when the crack grade exceeds a preset threshold; Step S5: Performing reinforcement work on the crack area according to the repair instruction; Step S4 further comprises an adaptive excitation step: dynamically adjusting an excitation frequency range of the electromagnetic imaging module according to acoustic emission energy, and adjusting an excitation current amplitude according to acoustic emission event count; The acoustic emission monitoring module comprises no less than four waterproof packaged piezoelectric acoustic emission sensors coupled with underwater rock mass surface through soft coupling pads, which are symmetrically arranged in an array with a spacing of 1-2 m and coupled with the underwater rock mass surface through mechanical clamping and soft coupling pads; The electromagnetic imaging module comprises: A matrix arranged transmitting coil array, the transmitting coil is kept a distance of 0.5-1 m from the crack area, and supports multi-frequency alternating excitation; A two-dimensional electrode array attached to the surface of the crack area, the electrode spacing is 0.5-1 m, and the arrangement direction is orthogonal to the transmitting coil array; In step S3, the Tikhonov regularization method is adopted, and the expression of the inversion target function is: ； wherein, represents a system matrix; represents a response vector measured by the electromagnetic imaging module; represents a regularization parameter; represents a to-be-solved vector, i.e., a conductivity image; Based on the inversion result, a conductivity distribution map is extracted, and a conductivity characteristic parameter is analyzed, the conductivity characteristic parameter includes conductivity mutation amplitude, abnormal area proportion and resistivity gradient characteristics.

2. The method according to claim 1, wherein the method is characterized in that: In step S4, a first determination unit is used to determine the crack grade; The first determination unit constructs a threshold decision tree based on acoustic emission event count in the acoustic emission features and conductivity mutation amplitude and abnormal area proportion in the conductivity characteristic parameter, and outputs the crack grade; The threshold decision tree comprises: When the acoustic emission event count per unit time is less than a first threshold, it is determined that there is no crack or false positive; When the acoustic emission event count per unit time is greater than or equal to the first threshold and the conductivity mutation amplitude is less than a second threshold, it is determined that there is a slight crack; When the acoustic emission event count per unit time is greater than or equal to the first threshold, the conductivity mutation amplitude is greater than or equal to the second threshold, and the abnormal area proportion is less than a third threshold, it is determined that there is a moderate crack; When the acoustic emission event count per unit time is greater than or equal to the first threshold, the conductivity mutation amplitude is greater than or equal to the second threshold, and the abnormal area proportion is greater than or equal to the third threshold, it is determined that there is a severe crack.

3. The method according to claim 1, wherein the method is characterized in that: In step S4, a second determination unit is used to determine the crack grade; The second determination unit is used for weighted fusion of the acoustic emission features and the conductivity characteristic parameter, and compares with the set threshold to determine the crack grade; In training, the weight and threshold are self-optimized through historical sample data.

4. The method according to claim 3, wherein the method is characterized in that: The expression for weighted fusion of the acoustic emission features and the conductivity characteristic parameter is: ； ； In the formula, represents the comprehensive score; represents the weight coefficient, =1, 2, 3, 4, 5; represents the acoustic emission event count per unit time; represents the maximum acoustic emission event number allowed for safe operation; represents the average energy of acoustic emission events per unit time; represents the maximum observed energy; represents the maximum change range of electrical conductivity; represents the maximum change range of resistivity history; represents the area of the abnormal region detected in the electromagnetic image; represents the total area of the entire monitoring region; represents the maximum proportion threshold of the abnormal region in the resistivity image; represents the average gradient of the resistivity image; the historical maximum value of the average gradient of the resistivity image.

5. The method of claim 4, wherein the method further comprises: The second determination unit uses a weighted least square error as a loss function when training. 6.The method according to claim 1, characterized in that: The expression of the excitation frequency range is: ； wherein represents the minimum excitation frequency; represents the maximum excitation frequency; represents the center frequency; represents the maximum allowed spread bandwidth; represents the maximum observed energy; represents the average energy of acoustic emission events per unit time; The excitation current amplitude The expression for the excitation current amplitude is ； ； wherein represents the default current; represents the count of acoustic emission events per unit time; represents the acoustic emission count threshold; represents the maximum number of acoustic emission events allowed for safe operation; represents the adjustment factor for the maximum allowed current amplitude.

Images