A Method for Anomaly Detection and Localization of Underwater Engineering Structures Based on Acoustic Arrays of Submersible Moorings

Abstract

This invention relates to the field of marine engineering monitoring and underwater acoustic sensing technology, specifically a method for detecting and locating anomalies in underwater engineering structures based on a moored acoustic array. The method includes the following steps: establishing an equivalent coupled acoustic propagation model; deploying a moored system in the water area above the underwater engineering structure, with an acoustic receiving array fixedly mounted on the moored system; performing time synchronization, element spatial position calibration, and channel gain calibration on the acoustic receiving array; when an abnormal acoustic event with spatially consistent propagation characteristics is detected, the acoustic receiving array enters an enhanced operating mode, outputting multi-channel enhanced acoustic signals; in the enhanced operating mode, the method outputs a determination result of whether an abnormal acoustic event has occurred, and solves for the estimated position of the abnormal event along the structural axis and the corresponding location signal interval. This method is suitable for online anomaly monitoring and precise location of underwater linear engineering structures with known spatial orientations in complex marine environments.

Description

Technical Field

[0001] This invention relates to the field of marine engineering monitoring and underwater acoustic sensing technology, and in particular to a method for detecting and locating anomalies in underwater engineering structures based on underwater mooring acoustic arrays. Background Technology

[0002] Underwater linear engineering structures are an important component of marine engineering systems, widely used in underwater energy facilities, engineering channels, protective casings, and long-distance engineering components. Their operational safety directly affects the stability of the engineering system and the safety of the marine environment. Any abnormal release or disturbance of the medium inside or near the structure can not only affect the normal operation of the engineering structure but also adversely impact the surrounding aquatic environment. Therefore, rapid, reliable detection and precise location of abnormal events in underwater linear engineering structures are of significant engineering importance.

[0003] Subsea linear engineering structures include, but are not limited to, wellheads. Wellheads are critical equipment in the oil extraction process, used to control wellhead pressure, regulate flow, and connect wells to production pipelines. Their proper operation is vital to the economic benefits of an oilfield. Subsea wellhead systems are placed on the seabed at a depth of 1500 meters along with the subsea production equipment. At this designed depth, the seabed environment is extremely harsh, highly corrosive, and with low visibility. Unlike traditional land-based wellhead systems, the seawater environment is unique, with factors such as seawater corrosion and increasing environmental pressure with depth. Over long-term operation, wellheads may suffer from corrosion, fatigue, and cracks, potentially leading to leaks, ruptures, and other accidents. Focusing on wellhead safety allows for real-time monitoring and evaluation, identifying potential safety hazards, promptly detecting faults and anomalies, ensuring the safe and reliable operation of the equipment, thereby reducing downtime for maintenance and improving the economic efficiency of the oilfield.

[0004] Existing monitoring technologies for abnormal states in underwater linear engineering structures mainly include indirect monitoring methods based on changes in operating parameters, distributed sensing methods, and acoustic sensing methods. Among these, methods based on changes in operating parameters are insensitive to weak anomalies and are easily affected by fluctuations in operating conditions. While distributed sensing methods offer high sensitivity, they are costly to deploy, complex to construct, and difficult to maintain in long-distance underwater engineering structures. In contrast, acoustic sensing methods utilize acoustic radiation generated by abnormal events for non-contact monitoring, offering advantages such as strong adaptability, wide coverage, and long-distance sensing capabilities, making them an important development direction for anomaly monitoring of underwater linear engineering structures.

[0005] However, the underwater environment is complex and variable. Abnormal acoustic events are affected by factors such as engineering structure coupling, multipath effects, marine environmental noise, and non-stationary background interference during propagation, leading to a significant decrease in the signal-to-noise ratio (SNR) of the received signal. Traditional acoustic anomaly detection methods based on single-point hydrophones or static threshold criteria suffer from high false alarm rates and poor decision stability under low SNR conditions. Three-dimensional sound source localization methods based on conventional time-of-arrival (TOA) do not fully utilize the prior structural information of underwater linear engineering structures with known spatial orientations, easily leading to multiple solutions and failing to meet the reliability and positioning accuracy requirements of engineering applications. Therefore, there is an urgent need for an acoustic monitoring method for underwater linear engineering structures that can achieve adaptive detection, stable decision-making, and highly reliable positioning in complex acoustic environments. Summary of the Invention

[0006] The purpose of this invention is to overcome the above-mentioned defects in the existing technology and propose an underwater engineering structure anomaly detection and localization method based on underwater mooring acoustic array. It utilizes the spatial location information of underwater engineering structures to perform electrical digital data processing of acoustic signals, thereby realizing online anomaly monitoring and precise localization of underwater linear engineering structures with known spatial orientation in complex marine environments.

[0007] The technical solution of this invention is: a method for detecting and locating anomalies in underwater engineering structures based on a submersible buoy acoustic array, comprising the following steps:

[0008] S1. Based on the external environmental parameters of the underwater engineering structure and the operating status information of the underwater engineering structure, an equivalent coupled acoustic propagation model is established.

[0009] S2. Deploy a mooring system in the water area above the underwater engineering structure. The mooring system is equipped with an acoustic receiving array including several scalar hydrophones. Perform time synchronization, spatial position calibration of array elements, and channel gain calibration on the acoustic receiving array.

[0010] S3. The acoustic receiving array continuously collects multi-channel acoustic signals from the water surrounding the underwater engineering structure. When an abnormal acoustic event with spatially consistent propagation characteristics is detected, the acoustic receiving array enters the enhancement working mode and outputs multi-channel enhanced acoustic signals.

[0011] S4. In enhanced working mode, output the judgment result of whether an abnormal acoustic event has occurred, and provide the probability of the abnormal event and the confidence level of the judgment.

[0012] S5. After determining that an abnormal acoustic event has occurred, solve for the estimated position of the abnormal event along the structural axis and the corresponding position information interval.

[0013] S6. Perform event-driven online correction and update of the equivalent sound velocity and equivalent coupling delay parameters in the equivalent coupled sound propagation model.

[0014] In this invention, the specific implementation process of step S1 is as follows:

[0015] S1.1 Acquisition of environmental parameters and operating status:

[0016] External environmental parameters include the temperature of the seawater surrounding the underwater engineering structure. ,salinity and water depth ;

[0017] Engineering operation status information includes the pressure of the medium inside or near the underwater linear engineering structure. Medium temperature Relative velocity and the proportion parameters of the medium components The engineering operation status information is described using interval format:

[0018] ;

[0019] S1.2 Calculation of initial value of sound velocity in external seawater;

[0020] Based on the obtained seawater temperature ,salinity and water depth information Calculate the speed of sound in external seawater The initial estimate is:

[0021] ,

[0022] in, An empirical model function representing the relationship between seawater sound speed and environmental parameters;

[0023] S1.3 Initialization of acoustic prior parameters and setting of initial equivalent sound velocity for engineering operation status information:

[0024] operating pressure Medium temperature Relative velocity and the proportion parameters of the medium components The set of prior acoustic parameters of the constituent medium:

[0025] ,

[0026] Initial value of external seawater sound speed For reference, the initial value of the equivalent speed of sound Configure settings:

[0027] ,

[0028] in, and This represents the dimensionless empirical boundary coefficient, used to define the initial value of the equivalent speed of sound. relative to the initial value of the external seawater sound speed The feasible range;

[0029] S1.4 Establishment of the equivalent coupled sound propagation model:

[0030] Assume the spatial orientation or equivalent centerline of the underwater linear engineering structure is determined by the parametric equations. It means that, among them This represents a parameter along the axial direction of an underwater linear engineering structure. Let the effective axial length of the underwater linear engineering structure be denoted by ; let the in the acoustic receiving array be denoted by . The spatial location of each scalar hydrophone is The abnormal sound source is located at the position of the structural axis parameter. When the propagation distance to the array element is expressed as:

[0031] ,

[0032] An equivalent coupled acoustic propagation model is used to represent the arrival time of the anomalous acoustic signal at the first... Arrival time of each scalar hydrophone:

[0033] ,

[0034] in, This represents the initial value of the equivalent speed of sound; This represents the initial value of the equivalent coupling delay, used to characterize the equivalent time offset introduced by factors such as structural coupling, near-field radiation effects, and system fixed delay.

[0035] Based on the above equivalent coupled sound propagation model, we can obtain any two scalar hydrophones. and The predicted arrival time difference is:

[0036] .

[0037] The specific implementation process of step S2 is as follows:

[0038] S2.1 Deployment of the underwater buoy system and local acoustic receiving array structure:

[0039] A mooring system is deployed in the water area above the underwater linear engineering structure. A rigid mounting frame is set on the top of the mooring system, and four scalar hydrophones are arranged on the mounting frame to form an acoustic receiving array. The acoustic receiving array is a four-element non-coplanar tetrahedral array.

[0040] S2.2 Time Synchronization and Channel Delay Calibration:

[0041] Let the first The discrete signals collected by each scalar hydrophone are The first scalar hydrophone is selected as the reference channel. Calibration signals for each channel are acquired using a calibration sound source at a known location. For any given... Normalized cross-correlation function between the calculated channel and the reference channel:

[0042] ,

[0043] Based on the peak position of the cross-correlation function, determine the first... Delay estimate of the channel relative to the reference channel:

[0044] ,

[0045] Based on this, time delay compensation is performed on the channel signal to align it with the reference channel in time.

[0046] ,

[0047] in, Indicates the number after time delay calibration Channel signal; Indicates the first The raw discrete-time acoustic signals acquired from each channel; This indicates the first value estimated through cross-correlation analysis. The time offset of each channel relative to the reference channel; This indicates that the original number will be... The signal sequence after the channel signal is shifted on the discrete time axis;

[0048] S2.3, Scalar hydrophone spatial position calibration:

[0049] Establish a local coordinate system for a four-element non-coplanar tetrahedral array. With the geometric center of the mounting bracket as the origin Let the fixed coordinates of the four scalar hydrophones in the local coordinate system be... The three scalar hydrophones are located on the same horizontal plane. The top two sides form an equilateral triangle:

[0050] ,

[0051] The fourth scalar hydrophone is located above. Location:

[0052] ,

[0053] in, Indicates the horizontal scale of the array elements; Indicates vertical height, and and They are on the same order of magnitude;

[0054] The position of the scalar hydrophone in the global coordinate system is denoted as When a mooring buoy experiences attitude angle changes due to ocean currents, the mapping relationship between the local and global coordinate systems can be expressed as a rigid body transformation:

[0055] ,

[0056] in, This indicates the position of the mounting bracket origin in the global coordinate system; This represents the rotation matrix measured by the attitude sensor (IMU);

[0057] S2.4 Channel Gain Calibration:

[0058] Under the same calibration sound source conditions, let the first... The effective signal amplitude of the channel is estimated as follows:

[0059] ,

[0060] in, This represents the total number of discrete-time sampling points used for amplitude estimation under the action of a calibrated sound source; This represents the total number of discrete-time sampling points.

[0061] If the first scalar hydrophone is selected as the reference channel, then the... The gain calibration coefficient for each channel is defined as follows:

[0062] ,

[0063] in, Indicates the effective signal amplitude of the reference channel;

[0064] Based on this, the channel acoustic signal is subjected to amplitude normalization:

[0065] .

[0066] The specific implementation process of step S3 is as follows:

[0067] S3.1 Multi-channel acoustic signal acquisition and preprocessing:

[0068] After time synchronization and channel gain calibration, the four scalar hydrophones in the underwater buoy system output calibrated multi-channel acoustic signals, denoted as:

[0069] ,

[0070] in, Indicates the first Continuous or discrete acoustic signals of a scalar hydrophone under a global time reference;

[0071] The above multi-channel signals are synchronously preprocessed, including DC component removal, bandpass filtering, and multi-channel adaptive noise reduction, to obtain multi-channel acoustic signals for subsequent event triggering and decision-making:

[0072] ;

[0073] S3.2 Construction of spatial consistency features of acoustic receiving array:

[0074] Using scalar hydrophones Calculate the signal in the sliding time window using units of measurement. Cross-correlation function within:

[0075] ,

[0076] in, This represents the time delay between channels, i.e., a scalar hydrophone. With scalar hydrophone The relative time offset between received signals; Indicates the start time of the current sliding time window;

[0077] And based on the cross-correlation function, a scalar hydrophone pair is defined. Normalized correlation coefficient:

[0078] ,

[0079] in, Indicates the first The autocorrelation function values ​​of the scalar hydrophone channel signals under zero-delay conditions; Indicates the first The autocorrelation function values ​​of the scalar hydrophone channel signals under zero-delay conditions; Indicates the first The and the first The scalar hydrophone channel signal has a time delay of Cross-correlation function values ​​under the given conditions;

[0080] For an acoustic receiving array consisting of four scalar hydrophones, six different scalar hydrophone pairs can be formed. Further analysis of the normalized correlation coefficients of each scalar hydrophone pair allows for the construction of a spatial consistency index for the acoustic receiving array.

[0081] ;

[0082] S3.3 Event Trigger Determination and Enhanced Mode Switching:

[0083] When the acoustic receiver array spatial consistency index Higher than the preset consistency threshold And the duration of this state exceeds the minimum decision window. When the current acoustic event meets the spatial propagation characteristics of a potential abnormal event, the acoustic receiving array is triggered to switch from the conventional monitoring mode to the enhanced working mode.

[0084] When the acoustic receiver array spatial consistency index Below the preset consistency threshold If this state lasts for more than the preset recovery time, the acoustic receiving array will automatically return from the enhanced working mode to the normal monitoring mode.

[0085] S3.4, Multi-channel signal output in enhanced mode:

[0086] In the enhanced working mode, the multi-channel acoustic signal preprocessed in step S3.1 is... As an acoustic receiving array, it enhances the output signal and simultaneously records the array spatial consistency index within the corresponding time window. And the relevant characteristic parameters of each scalar hydrophone.

[0087] The specific implementation process of step S4 is as follows:

[0088] S4.1 Extraction of time-domain features of the enhanced output signal from the acoustic receiving array;

[0089] In the enhanced working mode of step S3, a length of [length value] is selected. The sliding time window enhances the output signal of the acoustic receiving array. Perform temporal feature extraction, in the first... Define the array's time-domain energy characteristics within a time window. for:

[0090] ,

[0091] in, Indicates the first The start time of each sliding time window, the time interval covered by the time window is... ];

[0092] S4.2 Extraction of frequency domain features of the enhanced output signal from the acoustic receiver array:

[0093] Within the same time window as step S4.1 Within the acoustic receiver array, frequency domain analysis is performed on the enhanced output signal.

[0094] Acoustic signal for each channel Perform a short-time Fourier transform to obtain its spectral representation. Based on the typical spectral distribution characteristics of the target's abnormal acoustic events, the target frequency band is pre-set. And construct the average frequency domain energy characteristics of the array. :

[0095] ;

[0096] S4.3 Construction of spatial correlation characteristics of the enhanced output signal of the acoustic receiving array;

[0097] The first [channel signal] was calculated based on the multi-channel signal of the acoustic receiving array. Acoustic receiver array spatial consistency index within a time window and will As a spatial correlation feature of the acoustic receiving array, it is directly introduced into the abnormal event judgment process to distinguish sound source events with clear spatial propagation characteristics from spatially unrelated random background noise;

[0098] For the The spatial consistency index of the acoustic receiver array within each time window is smoothed within adjacent time windows to obtain the smoothed array spatial consistency characteristics. :

[0099] ,

[0100] in, Indicates the number of time windows used for smoothing;

[0101] S4.4, Characteristic Temporal Evolution Analysis and Joint Decision:

[0102] Let the first Within a time window, a joint feature vector is constructed, consisting of array time-domain energy characteristics, array frequency-domain energy characteristics, and array spatial consistency characteristics. :

[0103] ,

[0104] By comparison The changes within adjacent time windows are analyzed to determine their temporal evolution trend, and a characteristic rate of change index is constructed. :

[0105] ,

[0106] When the joint feature vector simultaneously satisfies the following conditions within multiple consecutive windows: the improvement of both the array time-domain energy feature and the array frequency-domain energy feature relative to the background level is not less than 3 dB, preferably 3-10 dB; the spatial consistency index is consistently higher than the consistency decision threshold of 0.5-0.8; and the rate of change of the joint feature within adjacent time windows... If the amplitude does not exceed 20%-50% of the previous window's characteristic amplitude, then the current acoustic event is determined to meet the typical characteristics of the target abnormal acoustic event;

[0107] S4.5 Output of abnormal event judgment results:

[0108] When the above joint decision conditions are met, the judgment result of the abnormal acoustic event is output, and the joint feature vector within the corresponding time window is output. Characteristic rate of change index This serves as the foundational data for assessing the probability of abnormal events and the confidence level of decisions.

[0109] The specific implementation process of step S5 is as follows:

[0110] S5.1 Acquisition of the Time Difference of Arrival Observation:

[0111] No. The scalar hydrophone and the first A scalar hydrophone pair composed of scalar hydrophones in the corresponding time window Internally calculate the cross-correlation function between the two channel signals:

[0112] ,

[0113] And based on the peak position of the cross-correlation function, determine the first... The scalar hydrophone and the first The relative time difference of arrival between individual scalar hydrophones :

[0114] ,

[0115] For an acoustic receiving array consisting of four scalar hydrophones, six sets of time difference of arrival (TDOA) observations for scalar hydrophone pairs can be obtained, and these observations can be combined to form a TDOA observation vector. :

[0116] ;

[0117] S5.2 Propagation model prediction based on structural axis constraints;

[0118] Based on the equivalent coupled sound propagation model established in step S1, assuming the abnormal sound source is located on the structural axis or equivalent centerline of a known underwater linear engineering structure, at the equivalent sound velocity... Under these conditions, for any given axial position parameter Based on the equivalent propagation distance from the abnormal sound source to each array element, the corresponding predicted time difference of arrival is calculated, and a predicted time difference of arrival vector is constructed:

[0119] ;

[0120] S5.3 One-dimensional inversion localization based on axial prior constraints:

[0121] By introducing prior structural constraints that the anomalous sound source is located on a known structural axis or equivalent centerline, a cost function is constructed with the deviation between the observed and predicted arrival time differences as the objective:

[0122] ,

[0123] in, Represents the L2 norm;

[0124] Within a given axial range Internally, the cost function is optimized using one-dimensional search or numerical optimization methods. Solving for the axial position parameters that minimize the cost function yields the following:

[0125] ,

[0126] in, This represents an estimated value indicating the axial location of the abnormal sound source.

[0127] S5.4, Location Confidence Interval and Stability Assessment:

[0128] Axial position estimate Nearby, for the cost function The changing characteristics are analyzed based on preset error tolerance parameters. Define the axial confidence interval for axial position. for:

[0129] .

[0130] The specific implementation process of step S6 is as follows:

[0131] S6.1 Selection of Updatable Parameters:

[0132] Based on the equivalent coupled sound propagation model established in step S1, the model parameter vector selected to describe the propagation characteristics of abnormal sound sources is as follows:

[0133] ,

[0134] in, It represents the equivalent speed of sound and is used to comprehensively characterize the combined influence of external seawater propagation characteristics, structural coupling effects, and unexplicitly modeled factors on propagation speed. It represents the equivalent pipe wall acoustic coupling delay, used to characterize the equivalent time offset introduced by factors such as structural transmission / radiation, near-field effects, and system fixed delay;

[0135] S6.2 Calculation of parameter correction based on observation-prediction bias:

[0136] The estimated axial position of the abnormal sound source calculated in step S5 Obtain the time difference of arrival observation vector observed by the acoustic receiving array within the corresponding time window. ;

[0137] In the current model parameters Under the given conditions, based on the equivalent coupled sound propagation model, the location of the abnormal sound source is calculated. The corresponding predicted arrival time difference vector ;

[0138] The deviation vector between the observed time difference of arrival (TDOA) and the predicted time difference of arrival (TDOA) is calculated by comparing the two vectors.

[0139] ;

[0140] S6.3 Constrained Correction Update of Equivalent Sound Propagation Parameters:

[0141] Define a cost function targeting the arrival time difference residual:

[0142] ,

[0143] By minimizing the cost function The parameters of the equivalent coupled sound propagation model are updated by inversion:

[0144] ,

[0145] The update process is simplified to the equivalent speed of sound One-dimensional or low-dimensional optimization, with constraints on the update magnitude:

[0146] ,

[0147] in, This indicates the preset maximum update step size, used to ensure the smoothness and engineering controllability of the parameter update process of the equivalent coupled acoustic propagation model;

[0148] S6.4 Information Upload and Alarm Handling:

[0149] After completing the parameter correction and update of the equivalent coupled acoustic propagation model, the judgment results of abnormal events, the estimated axial position, the corresponding axial confidence interval, the probability of occurrence of abnormal events and the decision confidence information, as well as the operating status information of the underwater glider system, are transmitted to the monitoring platform.

[0150] When the information meets the preset alarm conditions, the monitoring platform triggers a leakage alarm and stores the relevant data to support subsequent operation and maintenance decisions and event retrospective analysis.

[0151] The correction and update of the equivalent coupled acoustic propagation model parameters adopts an event-driven triggering mechanism, which is executed when the following conditions are met simultaneously: the abnormal event judgment result in step S4 is true; the center localization result in step S5 is true. It exhibits stability, and the positioning information interval corresponding to the positioning result is less than a preset threshold range.

[0152] The beneficial effects of this invention are:

[0153] (1) To meet the monitoring needs of abnormal acoustic events in underwater linear engineering structures, a method for detecting and locating anomalies in underwater engineering structures based on a mooring acoustic array is proposed. This method can reliably determine and stably locate abnormal events under complex ocean background noise and low signal-to-noise ratio conditions. This application uses the spatial consistency characteristics of the array to trigger and determine abnormal acoustic events, and combines the prior of the structural axis or equivalent center line to constrain the traditional three-dimensional sound source localization problem into a one-dimensional inversion problem along the structural direction. This effectively reduces the search dimension of localization, reduces the occurrence of multiple solutions and misjudgments, and significantly improves the stability and repeatability of anomaly location estimation from an engineering perspective. It is especially suitable for long-distance, linearly distributed underwater engineering target monitoring scenarios.

[0154] (2) The system adopts an event-driven enhanced working mode and a multi-feature joint decision mechanism to improve the reliability and practicality of the system under long-term online monitoring conditions: This application continuously evaluates the array spatial consistency index under the normal monitoring mode and only enters the enhanced working mode when the abnormal propagation characteristic conditions are met, thus avoiding over-response to background noise and occasional disturbances; at the same time, it combines the time evolution characteristics of array time domain energy, frequency domain energy and spatial correlation characteristics in the continuous time window for joint decision-making, effectively suppressing false alarms caused by instantaneous noise pulses, non-continuous interference and local random sound sources, thereby significantly reducing the false alarm rate and improving the credibility of abnormal event judgment results in actual engineering operation;

[0155] (3) By introducing an equivalent coupled acoustic propagation model and an event-driven online parameter update mechanism, the system can adapt to changes in environment and operating conditions and maintain positioning consistency over long-term operation: This application does not rely on detailed modeling of the acoustic parameters of the medium or structure, but adopts an engineering parameter description method of equivalent sound velocity and equivalent coupling time delay, and only performs limited online correction of model parameters under conditions of high-confidence abnormal events and stable positioning results, avoiding the damage to system performance caused by invalid or erroneous updates. This mechanism enables the system to continuously correct the propagation model under conditions of changes in marine environment, structural state, or operating conditions, improve the accuracy of time difference of arrival prediction, and thus enhance the consistency and robustness of abnormal location inversion results at different time scales;

[0156] (4) This application adopts a single underwater buoy and small-scale acoustic array deployment method, which has a simple system structure, flexible deployment, and good engineering feasibility and promotion value. Compared with underwater acoustic monitoring schemes that rely on large-scale arrays or multi-point synchronous deployment, this application only needs to deploy a single underwater buoy acoustic array above the target engineering structure to complete the detection and location of abnormal events, which significantly reduces the system deployment complexity and operation and maintenance costs. At the same time, through the structural prior constraints and event-driven processing at the algorithm level, the problem of insufficient information caused by the limited array size is made up for, which has higher adaptability and practicality in actual marine engineering applications such as monitoring of oil production trees and long-term unattended operation. Attached Figure Description

[0157] Figure 1 This is a flowchart of the method described in this application;

[0158] Figure 2 It is a simulation scenario of curved underwater engineering structures and mooring arrays;

[0159] Figure 3 This is a graph showing the RMSE of the method described in this application and existing algorithms under conditions of random noise and environmental disturbance. Detailed Implementation

[0160] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0161] Specific details are set forth in the following description to provide a full understanding of the invention. However, the invention can be practiced in many ways other than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0162] A method for detecting and locating anomalies in underwater engineering structures based on a submersible buoy acoustic array, the flowchart of which is shown below. Figure 1 As shown. Specifically, the method includes the following steps.

[0163] The first step involves acquiring environmental parameters such as temperature, salinity, and depth of the seawater outside the underwater linear engineering structure, as well as operational status information of the internal or adjacent media available from the engineering operation system. This includes, but is not limited to, measured values ​​or estimated ranges of internal pressure, media temperature, relative velocity, and media component ratios. Based on these environmental parameters and operational status information, an equivalent coupled sound propagation model is established.

[0164] The equivalent coupled acoustic propagation model is used to describe how anomalous acoustic events are generated within or near the engineering structure, coupled through the structure, radiate outwards, propagate in seawater to the acoustic receiving array, and are initialized with equivalent sound velocity and equivalent coupling time delay parameters based on the equivalent coupled acoustic propagation model for subsequent time difference of arrival prediction and location inversion calculations. The specific implementation process of this step is described below.

[0165] First, obtain environmental parameters and operating status.

[0166] Obtain external marine environmental parameters and operational status information related to underwater engineering structures to establish an equivalent coupled acoustic propagation model.

[0167] External environmental parameters include the temperature of the seawater surrounding the underwater engineering structure. ,salinity and water depth The engineering operation status information includes operating parameters related to the state of the internal or adjacent media of the structure, which are directly acquired by the engineering monitoring system or pre-given, including the pressure of the media inside or adjacent to the underwater linear engineering structure. Medium temperature Relative velocity and the proportion parameters of the medium components In practical engineering conditions, some of the above-mentioned operating status parameters may be difficult to obtain in real time in the form of deterministic values. In this case, they can be described using an interval form:

[0168] ,

[0169] The interval information is used to impose reasonable constraints on the subsequent initialization and update process of equivalent sound propagation parameters.

[0170] Second, the initial value of the sound velocity in the external seawater is calculated.

[0171] Based on the obtained seawater temperature ,salinity and water depth information Calculate the speed of sound in external seawater The initial estimate is denoted as:

[0172] ,

[0173] in, An empirical model function representing the relationship between seawater sound speed and environmental parameters; in specific implementation, the function... The initial value of the sound velocity can be obtained using existing empirical formulas for ocean sound velocity or an interpolation model constructed based on historical CTD data. This application does not limit the specific form of the empirical model.

[0174] The initial value of the external seawater sound speed obtained It is used to characterize the basic characteristics of sound wave propagation in water and serves as a reference for setting subsequent equivalent sound propagation parameters.

[0175] Third, the initialization of acoustic prior parameters and the setting of the initial value of equivalent sound velocity in the medium inside or near the underwater linear engineering structure.

[0176] Considering that the physical state of the medium inside or near the underwater engineering structure may change with the operating conditions, and that it is difficult to obtain its complete acoustic parameters in real time in practical applications, this application does not directly perform detailed modeling of the acoustic properties of the medium, but constructs a priori description of the medium acoustics based on the available operating state information.

[0177] operating pressure Medium temperature Relative velocity and the proportion parameters of the medium components The set of prior acoustic parameters of the constituent medium:

[0178] ,

[0179] When the above parameters are given in the form of intervals, the corresponding intervals are used to limit the reasonable range of values ​​for the equivalent sound propagation parameters, thereby ensuring the stability and physical rationality of the model initialization process when the medium state changes.

[0180] Based on this, the initial value of the external seawater sound speed For reference, the initial value of the equivalent speed of sound Configure settings:

[0181] ,

[0182] in, and This represents the dimensionless empirical boundary coefficient, used to define the initial value of the equivalent speed of sound. relative to the initial value of the external seawater sound speed The feasible range is to comprehensively reflect the influence of changes in the state of the medium inside or near the structure, structural coupling effects, and unexplicitly modeled factors on the sound propagation speed. and It can be preset as a constant, or it can be based on the prior range of the running status information. Automatically determined.

[0183] Fourth, an equivalent coupled sound propagation model is established.

[0184] By combining the initial value of the external seawater sound velocity, the prior acoustic parameters of the medium, and the spatial geometric information of the underwater linear engineering structure, an equivalent coupled sound propagation model is established to describe the propagation process of abnormal acoustic events.

[0185] The equivalent coupled acoustic propagation model is used to characterize the equivalent propagation process of an anomalous acoustic event generated inside or in the vicinity of a structure, coupled by the structure and radiating outward, propagating in the external seawater to the acoustic receiving array.

[0186] Assume the spatial orientation or equivalent centerline of the underwater linear engineering structure is determined by the parametric equations. It means that, among them This represents a parameter along the axial direction of an underwater linear engineering structure. Let represent the effective axial length of the underwater linear engineering structure. Let be the length of the in the acoustic receiving array. The spatial location of each scalar hydrophone is The abnormal sound source is located at the position of the structural axis parameter. When the propagation distance to the array element is expressed as:

[0187] ,

[0188] To facilitate subsequent calculations of the time difference of arrival and the inversion of anomaly locations based on prior constraints of the structural axial direction, this application employs an equivalent coupled acoustic propagation model to represent the arrival time of the anomalous acoustic signal at the [missing information]. Arrival time of each scalar hydrophone:

[0189] ,

[0190] in, This represents the initial value of the equivalent speed of sound; This represents the initial value of the equivalent coupling delay, used to characterize the equivalent time offset introduced by factors such as structural coupling, near-field radiation effects, and system fixed delay. During the model initialization phase, the equivalent coupling delay... It can be set to zero, or a fixed compensation value obtained from system calibration.

[0191] Based on the above equivalent coupled sound propagation model, any two scalar hydrophones can be obtained. and The predicted arrival time difference is:

[0192] ,

[0193] The above-mentioned predicted time difference of arrival expression serves as the basis for calculation in subsequent steps to introduce prior constraints on the structural axis and perform one-dimensional inversion localization of abnormal sound sources.

[0194] The second step involves deploying a mooring system in the water area above the underwater linear engineering structure. Four scalar hydrophones are fixedly mounted on the mooring system to form a localized small-scale acoustic receiving array. The acoustic receiving array is then time-synchronized, its elements are spatially calibrated, and its channel gain is adjusted to obtain the array geometric priors and channel consistency conditions used for time difference of arrival estimation and positioning inversion. The specific implementation process of this step is described below.

[0195] First, deploy the underwater buoy system and local acoustic receiving array structure.

[0196] This application employs a single-moor deployment method, deploying a mooring system in the water area above an underwater linear engineering structure. A rigid mounting frame is installed on top of the mooring system, on which several scalar hydrophones are arranged to form a local small-scale acoustic receiving array. Given that anomalous acoustic events around underwater linear engineering structures typically exhibit broadband propagation, low signal-to-noise ratio, and multiple paths coexisting with background interference in seawater, to improve the stability of the array's time-of-arrival estimation and avoid geometric degradation caused by collinearity or coplanarity, this embodiment arranges four scalar hydrophones on the rigid mounting frame, fixing the local array as a four-element non-coplanar tetrahedral array.

[0197] Second, time synchronization and channel delay calibration.

[0198] In a submersible buoy system, multiple scalar hydrophone channels typically share the same acquisition unit clock. However, in practical engineering applications, each channel may still experience fixed group delays and sampling phase deviations introduced by analog front-end differences, circuit variations, or digital links. To ensure subsequent arrival time differences... To improve the estimation accuracy, this application performs relative time delay calibration on each channel.

[0199] Let the first The discrete signals collected by each scalar hydrophone are The first scalar hydrophone is selected as the reference channel. Calibration signals for each channel are acquired using calibration sound sources at known locations. For any given channel... Normalized cross-correlation function between the calculated channel and the reference channel:

[0200] ,

[0201] Based on the peak position of the cross-correlation function, determine the first... Delay estimate of the channel relative to the reference channel:

[0202] ,

[0203] Based on this, time delay compensation is performed on the channel signal to align it with the reference channel in time.

[0204] ,

[0205] in, Indicates the number after time delay calibration Channel signal; Indicates the first The raw discrete-time acoustic signals acquired from each channel; This indicates the first value estimated through cross-correlation analysis. The time offset of each channel relative to the reference channel; This indicates that the original number will be... The signal sequence after the channel signal is shifted on the discrete time axis is that equivalent time delay compensation is applied to the channel signal to align it with the effective signal components of the reference channel in time.

[0206] Third, scalar hydrophone spatial position calibration.

[0207] Since the array geometry parameters are directly involved in the prediction calculation of the propagation distance and time difference of arrival of anomalous acoustic events, it is necessary to obtain the scalar hydrophone spatial position. Reliable prior information. For the single underwater rigid mounting frame used in this application, the relative position of the scalar hydrophone in the local coordinate system. The parameters can be determined through precise measurements under laboratory conditions and written into the system as fixed array geometry parameters. After the mooring is deployed, this is combined with the mooring's global position. With attitude rotation matrix Get global array element positions .

[0208] Establish a local coordinate system for a four-element non-coplanar tetrahedral array. With the geometric center of the mounting bracket as the origin Let the fixed coordinates of the four scalar hydrophones in the local coordinate system be... The three scalar hydrophones are located on the same horizontal plane. The top two sides form an equilateral triangle:

[0209] ,

[0210] The fourth scalar hydrophone is located above. Location:

[0211] ,

[0212] in, Indicates the horizontal scale of the array elements; Indicates vertical height, and and They are on the same order of magnitude to ensure the non-coplanarity of the overall array structure.

[0213] To maintain consistency with the subsequent anomaly location inversion model, the position of the scalar hydrophone in the global coordinate system is denoted as... The global coordinate system can be taken as a fixed local coordinate system of the Earth over the monitored sea area, with its origin taken as the buoy deployment point and the coordinate axes pointing east, north, and up, respectively. When the buoy's attitude angle changes under the influence of ocean currents, the mapping relationship between the local and global coordinate systems can be expressed as a rigid body transformation:

[0214] ,

[0215] in, This indicates the position of the mounting bracket origin in the global coordinate system; This represents the rotation matrix measured by the attitude sensor (IMU). In cases where the buoy system is not equipped with an attitude measurement module, the rotation matrix can be used during the deployment phase. It is treated as a constant, and the attitude change error introduced by it is incorporated into the subsequent positioning error term for unified processing.

[0216] Fourth, channel gain calibration.

[0217] Different hydrophones may have varying sensitivities and gain characteristics in their front-end amplifier circuits. Without calibration, this will affect the accuracy of scalar hydrophone feature extraction and subsequent location reliability assessment. Therefore, this application performs amplitude consistency calibration on each channel of the scalar hydrophone.

[0218] Under the same calibration sound source conditions, let the first... The effective signal amplitude of the channel is estimated as follows:

[0219] ,

[0220] in, This represents the total number of discrete-time sampling points used for amplitude estimation under the action of a calibrated sound source; This represents the total number of discrete-time sampling points.

[0221] If the first scalar hydrophone is selected as the reference channel, then the... The gain calibration coefficient for each channel is defined as follows:

[0222] ,

[0223] Among them, the first scalar hydrophone is a uniformly selected reference channel within the array. Its number is only used to identify the hydrophone order and has no special physical meaning. Any other hydrophone in the array can also be selected as the reference channel without affecting the gain calibration results. This indicates the effective signal amplitude of the reference channel.

[0224] Based on this, the channel acoustic signal is subjected to amplitude normalization:

[0225] .

[0226] The multi-channel acoustic signal after the above time synchronization, delay calibration and gain calibration processing As the array input signal for subsequent steps, it ensures that each channel of the array is consistent in time and amplitude, thereby improving the stability and reliability of time difference of arrival estimation and anomalous position inversion based on structural axis constraints.

[0227] The third step involves continuously acquiring multi-channel acoustic signals from the water surrounding the underwater linear engineering structure using the acoustic receiving array from the second step. The acquired signals undergo synchronous preprocessing and noise reduction, and event triggering is performed on the acoustic signals. When an abnormal acoustic event with spatially consistent propagation characteristics is detected, the acoustic receiving array is driven into an enhanced operating mode to output multi-channel enhanced acoustic signals for abnormal event detection and location calculation. The specific implementation process of this step is described below.

[0228] First, multi-channel acoustic signal acquisition and preprocessing.

[0229] After time synchronization and channel gain calibration, the four scalar hydrophones in the underwater buoy system output calibrated multi-channel acoustic signals, denoted as:

[0230] ,

[0231] in, Indicates the first A scalar hydrophone provides continuous or discrete acoustic signals under a global time reference.

[0232] The aforementioned multi-channel signals undergo synchronous preprocessing, including DC component removal, bandpass filtering, and multi-channel adaptive noise reduction. The bandpass filter frequency band is set based on the typical spectral distribution range of the target anomalous acoustic event in seawater, used to suppress the influence of low-frequency flow noise, low-frequency interference caused by mooring / attitude disturbances, and high-frequency environmental noise on anomalous signal detection. After the above preprocessing, the multi-channel acoustic signals used for subsequent event triggering and decision-making are obtained:

[0233] .

[0234] Second, the construction of spatial consistency characteristics of the acoustic receiving array.

[0235] Considering that anomalous acoustic events related to underwater linear engineering structures usually manifest as coherent sound sources with spatially consistent propagation characteristics within the array scale, while marine environmental background noise and local random interference usually exhibit spatially non-consistent characteristics between array channels, this application utilizes the spatial consistency characteristics of the local acoustic receiving array as an important criterion for triggering anomalous events.

[0236] Using scalar hydrophones Calculate the signal in the sliding time window using units of measurement. Cross-correlation function within:

[0237] ,

[0238] in, This represents the time delay between channels, i.e., a scalar hydrophone. With scalar hydrophone The relative time offset between received signals; This indicates the start time of the current sliding time window.

[0239] And based on the cross-correlation function, a scalar hydrophone pair is defined. Normalized correlation coefficient:

[0240] .

[0241] in, Indicates the first The autocorrelation function value of a scalar hydrophone channel signal under zero time delay condition is used to characterize the average energy level of the channel signal within the sliding time window; Indicates the first The autocorrelation function values ​​of each scalar hydrophone channel signal under zero-delay conditions are used to characterize the signal energy level of the corresponding channel; Indicates the first The and the first The scalar hydrophone channel signal has a time delay of The cross-correlation function value under the given conditions is used to characterize the similarity between two channel signals under different time alignment conditions, and its peak position corresponds to the relative propagation delay characteristics between the two channel signals.

[0242] For an acoustic receiving array consisting of four scalar hydrophones, six different scalar hydrophone pairs can be formed. Further analysis of the normalized correlation coefficients of each scalar hydrophone pair allows for the construction of a spatial consistency index for the acoustic receiving array.

[0243] .

[0244] The above-mentioned acoustic receiver array spatial consistency index Used to characterize whether the acoustic signal received by the acoustic receiving array exhibits spatially consistent propagation characteristics within the current time window.

[0245] Third, event trigger determination and enhanced mode switching.

[0246] When the acoustic receiver array spatial consistency index Higher than the preset consistency threshold And the duration of this state exceeds the minimum decision window. When the current acoustic event meets the spatial propagation characteristics of a potential abnormal event, the acoustic receiving array is triggered to switch from the normal monitoring mode to the enhanced operating mode.

[0247] In enhanced operating mode, the system adaptively adjusts the signal acquisition and processing parameters of the acoustic receiving array in response to triggered potential anomalous acoustic events. Compared to conventional monitoring mode, the system improves the temporal resolution of multi-channel acoustic signals, shortening the sampling time interval to within twice that of the conventional mode, thereby enhancing the estimation accuracy of the relative time difference of arrival between array elements. Simultaneously, it expands the effective processing bandwidth of the acoustic signals to cover the frequency bands of the main energy distribution of anomalous acoustic events, thus reducing the attenuation of anomalous acoustic characteristics by band-limited processing. In this enhanced operating mode, the system continuously records complete raw data from the acoustic receiving array's multi-channels and their corresponding characteristic parameters for subsequent anomalous event determination, time difference of arrival estimation, and location inversion calculations based on structural prior constraints.

[0248] When the acoustic receiver array spatial consistency index Below the preset consistency threshold If this state lasts longer than the preset recovery time, the acoustic receiving array will automatically return from the enhanced working mode to the normal monitoring mode in order to reduce the long-term operating power consumption of the acoustic receiving array and improve the overall operating stability.

[0249] Fourth, multi-channel signal output in enhanced mode.

[0250] In the enhanced working mode, the multi-channel acoustic signal preprocessed in the above steps is... As an acoustic receiving array, it enhances the output signal and simultaneously records the array spatial consistency index within the corresponding time window. The acoustic receiving array enhances the output signal and related characteristic parameters of each hydrophone. These parameters serve as input data for the subsequent fourth step (abnormal event determination) and the fifth step (time difference of arrival estimation and anomalous location inversion). The related characteristic parameters include: the normalized cross-correlation coefficient between hydrophone pairs and their corresponding peak delay positions, the overall spatial consistency index of the array, the estimated time difference of arrival between array elements, and the energy and spectral characteristic parameters of each channel and the array itself.

[0251] The fourth step, in enhanced operating mode, extracts time-domain features, frequency-domain features, and array spatial correlation features from the enhanced output signal of the acoustic receiving array from the previous step. These features are then combined with their temporal evolution characteristics within a continuous time window to make a joint decision, outputting a determination of whether an abnormal acoustic event has occurred. The probability of the abnormal event and the confidence level of the decision are also provided. The specific implementation process is described below.

[0252] First, extraction of the time-domain characteristics of the enhanced output signal from the acoustic receiving array.

[0253] In the enhanced working mode of the third step, a length of [length value] is selected. The sliding time window enhances the output signal of the acoustic receiving array. Temporal feature extraction is performed. To reduce the impact of occasional interference from a single channel on the decision result, this application uses an array averaging method to construct temporal statistical features. Taking the array temporal energy features as an example, in the first... Define the array's time-domain energy characteristics within a time window. for:

[0254] ,

[0255] in, Indicates the first The start time of each sliding time window, the time interval covered by the time window is... ].

[0256] The array time-domain energy characteristics are used to characterize the overall energy level of the acoustic signals received by the array within the current time window.

[0257] Second, extraction of frequency domain features of the enhanced output signal from the acoustic receiving array.

[0258] Considering that anomalous acoustic events associated with underwater linear engineering structures typically manifest as broadband energy enhancement or significant energy concentration within specific frequency bands in seawater, within the same time window as the steps described above... Within this, frequency domain analysis is performed on the enhanced output signal of the acoustic receiving array. This analysis is conducted on the acoustic signal of each channel. Perform a short-time Fourier transform to obtain its spectral representation. Based on the typical spectral distribution characteristics of the target anomalous acoustic events, the target frequency band is pre-defined. And construct the average frequency domain energy characteristics of the array. :

[0259] ,

[0260] The array frequency domain energy characteristics are used to reflect the energy changes related to the characteristic frequency bands of the target abnormal event within the current time window.

[0261] Third, the construction of spatial correlation features of the acoustic receiving array to enhance the output signal.

[0262] The first [channel signal] was calculated based on the multi-channel signal of the acoustic receiving array. Acoustic receiver array spatial consistency index within a time window This index is used to characterize whether the sound field received by the acoustic receiving array has spatially consistent propagation characteristics.

[0263] The first Acoustic receiver array spatial consistency index within a time window The spatial correlation characteristics of the acoustic receiving array are directly introduced into the abnormal event decision-making process to distinguish between sound source events with clear spatial propagation characteristics and spatially independent random background noise. To improve the stability of the decision-making process, the first... The spatial consistency index of the acoustic receiver array within each time window is smoothed within adjacent time windows to obtain the smoothed array spatial consistency characteristics. :

[0264] ,

[0265] in, This indicates the number of time windows used for smoothing.

[0266] Fourth, characteristic temporal evolution analysis and joint decision-making.

[0267] Real-world anomalous acoustic events typically manifest as continuous or gradually evolving changes in acoustic characteristics over time, rather than as instantaneous anomalies within a single time window. Based on this, this application incorporates the temporal evolution characteristics of acoustic features, including time-domain features, frequency-domain features, and spatial correlation features, into the joint decision-making process.

[0268] Let the first Within a time window, a joint feature vector is constructed, consisting of array time-domain energy characteristics, array frequency-domain energy characteristics, and array spatial consistency characteristics. :

[0269] ,

[0270] By comparison The changes within adjacent time windows are analyzed to determine their temporal evolution trend, and a characteristic rate of change index is constructed. :

[0271] ,

[0272] When the joint feature vector simultaneously satisfies the following conditions within multiple consecutive windows: the improvement of both the array time-domain energy feature and the array frequency-domain energy feature relative to the background level is not less than 3 dB, preferably 3-10 dB; the spatial consistency index is consistently higher than the consistency decision threshold of 0.5-0.8; and the rate of change of the joint feature within adjacent time windows... If the amplitude of the feature value does not exceed 20%-50% of the previous window, the current acoustic event is determined to meet the typical characteristics of the target anomalous acoustic event. The background level is calculated as the statistical mean of the corresponding features within a historical time window under conditions of no anomalous events.

[0273] Fifth, output the results of abnormal event determination.

[0274] When the above joint decision conditions are met, the determination result of the abnormal acoustic event is output. The joint feature vector within the corresponding time window is then output. Characteristic rate of change index As the basic data for assessing the probability of abnormal events and the confidence level of judgments, it is used for subsequent steps such as abnormal location inversion and alarm decision-making by the monitoring platform.

[0275] The fifth step involves calculating the arrival time difference between scalar hydrophones based on the abnormal acoustic signals received by different scalar hydrophones in the acoustic receiving array after determining that an abnormal acoustic event has occurred. Combining this with the established equivalent coupled sound propagation model and introducing prior structural constraints that the abnormal sound source is located near the axis of an underwater linear engineering structure with a known spatial orientation or the centerline of an equivalent structure, the problem of locating the abnormal sound source is transformed into a one-dimensional parametric inversion problem along the axial direction of the engineering structure. Solving this problem yields the estimated position of the abnormal event along the structural axis and the corresponding location information interval. The specific implementation process is described below.

[0276] First, the observation of the time difference of arrival is obtained;

[0277] In the enhanced working mode of the third step, the time window determined as an abnormal acoustic event in the fourth step is selected, and the multi-channel acoustic signals output by the acoustic receiving array are processed. Estimate the time difference of arrival observations. (Based on the first...) The scalar hydrophone and the first Taking a scalar hydrophone pair consisting of scalar hydrophones as an example, in the corresponding time window Internally calculate the cross-correlation function between the two channel signals:

[0278] ,

[0279] And based on the peak position of the cross-correlation function, determine the first... The scalar hydrophone and the first The relative time difference of arrival between individual scalar hydrophones :

[0280] ,

[0281] For an acoustic receiving array consisting of four scalar hydrophones, six sets of time difference of arrival (TDOA) observations for scalar hydrophone pairs can be obtained, and these observations can be combined to form a TDOA observation vector. :

[0282] .

[0283] Second, propagation model prediction based on structural axis constraints.

[0284] Based on the equivalent coupled sound propagation model established in the first step, it is assumed that the abnormal sound source is located on the structural axis or equivalent center line of a known underwater linear engineering structure.

[0285] equivalent speed of sound Under these conditions, for any given axial position parameter Based on the equivalent propagation distance from the abnormal sound source to each array element, the corresponding predicted time difference of arrival can be calculated, and a predicted time difference of arrival vector can be constructed:

[0286] ,

[0287] The predicted time difference of arrival vector is used to characterize the theoretical time difference of arrival distribution characteristics generated on the acoustic receiving array when the anomalous sound source is located at different positions on the structural axis.

[0288] Third, one-dimensional inversion positioning based on axial prior constraints.

[0289] By introducing prior structural constraints that the anomalous sound source is located on a known structural axis or equivalent centerline, the original three-dimensional sound source localization problem is transformed into a problem concerning only a single axial parameter. The one-dimensional inversion localization problem. A cost function is constructed that targets the deviation between the observed and predicted time difference of arrival:

[0290] ,

[0291] in, This represents the L2 norm.

[0292] Within a given axial range Within this process, the cost function is solved using a one-dimensional search or numerical optimization method to obtain the axial position parameters that minimize the cost function.

[0293] ,

[0294] in, This represents an estimated value indicating the axial position of the abnormal sound source.

[0295] Compared with traditional three-dimensional sound source localization methods that do not introduce structural constraints, this application utilizes prior information about structural axes to significantly reduce the search dimension, reduce multiple solutions, and improve the localization stability and convergence reliability in low signal-to-noise ratio environments.

[0296] Fourth, positional confidence interval and stability assessment.

[0297] To evaluate the reliability of the axial position estimation results, the axial position estimate is... Nearby, for the cost function The changing characteristics are analyzed based on preset error tolerance parameters. (The parameters are related to the environmental noise level and the uncertainty in the time difference of arrival estimation), defining the axial confidence interval for the axial position. for:

[0298] ,

[0299] The length of the axial confidence interval is used to characterize the uncertainty of the axial position estimation and can serve as an important basis for subsequent alarm classification, result screening and decision support.

[0300] The sixth step involves using the multi-channel acoustic signals corresponding to the abnormal acoustic event, the observed time difference of arrival, and the inversion results of the abnormal location to perform event-driven online correction and update of the equivalent sound velocity and equivalent coupling time delay parameters in the equivalent coupled sound propagation model. The abnormal event judgment results, abnormal location estimates, location information intervals, probability of abnormal event occurrence, decision confidence level, and underwater buoy system operating status information are then uploaded to the monitoring platform via underwater communication links and / or wireless communication links to achieve alarm triggering and data storage. The specific implementation process is described below.

[0301] First, the selection and physical meaning of updatable parameters.

[0302] Based on the equivalent coupled sound propagation model established in the first step, the model parameter vector selected to describe the propagation characteristics of abnormal sound sources is as follows:

[0303] ,

[0304] in, It represents the equivalent speed of sound and is used to comprehensively characterize the combined influence of external seawater propagation characteristics, structural coupling effects, and unexplicitly modeled factors on propagation speed. It represents the equivalent pipe wall acoustic coupling delay, used to characterize the equivalent time offset introduced by factors such as structural transmission / radiation, near-field effects, and system fixed delay.

[0305] The parameters mentioned above do not require the actual physical acoustic constants of the corresponding medium or structure, but rather serve as engineering equivalent model parameters for time-of-arrival prediction and axial positioning inversion calculation. By correcting these parameters within a limited range, the adaptability of the equivalent coupled acoustic propagation model to changes in environment and operating conditions can be improved, thereby enhancing the stability and reliability of the axial positioning results.

[0306] Second, the calculation of parameter correction based on the observation-prediction bias.

[0307] Based on the estimated axial position of the abnormal sound source calculated in step five... Obtain the time difference of arrival observation vector observed by the acoustic receiving array within the corresponding time window. .

[0308] In the current model parameters Under these conditions, based on the equivalent coupled sound propagation model, the location of the abnormal sound source can be calculated. The corresponding predicted arrival time difference vector .

[0309] The deviation vector between the observed time difference of arrival (TDOA) and the predicted time difference of arrival (TDOA) is calculated by comparing the two vectors.

[0310] ,

[0311] The deviation vector is used to characterize the degree of matching between the prediction results and actual observations under the current equivalent coupled acoustic propagation model parameters.

[0312] Third, the limited correction update of the equivalent sound propagation parameters.

[0313] Based on the observation-prediction bias obtained from the above steps, the parameters of the equivalent coupled acoustic propagation model are updated with constrained correction to ensure consistency between the predicted time difference of arrival (TDOA) and the observation results from the acoustic receiving array. A cost function is defined with the TDOA residual as the objective factor:

[0314] ,

[0315] The parameters of the equivalent coupled acoustic propagation model are updated by minimizing the above cost function:

[0316] ,

[0317] In practical implementation, considering the low dimensionality of the equivalent coupled sound propagation model structure and the need for online computation, the update process can be simplified to updating the equivalent sound velocity. One-dimensional or low-dimensional optimization, with constraints on the update magnitude:

[0318] ,

[0319] in, This indicates the preset maximum update step size, used to ensure the smoothness and engineering controllability of the parameter update process of the equivalent coupled acoustic propagation model.

[0320] The correction and update of the equivalent coupled acoustic propagation model parameters are not performed continuously, but rather through an event-driven triggering mechanism, executing only when the following conditions are met simultaneously: 1) the abnormal event determination result in step four is valid; 2) the localization result in step five is valid. It is stable, and the positioning information interval corresponding to the positioning result is less than the preset threshold range.

[0321] By performing parameter correction only under high-confidence anomalous event conditions, frequent adjustments to the equivalent coupled acoustic propagation model parameters can be avoided in the absence of events or when localization is unreliable, thus ensuring the long-term stability of the system. Updated parameters As a new equivalent sound propagation parameter, it is used for subsequent abnormal event arrival time difference prediction and axial positioning inversion calculation, enabling the system to gradually adapt to changes in environment and working conditions.

[0322] Fourth, information uploading and alarm processing.

[0323] After completing the parameter correction and update of the equivalent coupled acoustic propagation model, the judgment results of abnormal events, the estimated axial position, the corresponding axial confidence interval, the probability of occurrence of abnormal events and the decision confidence level, as well as the operating status information of the underwater glider system, are uploaded to the monitoring platform via wireless communication links and / or underwater communication links. When the information meets the preset alarm conditions, the monitoring platform triggers a leakage alarm and stores the relevant data to support subsequent operation and maintenance decisions and event retrospective analysis.

[0324] To verify the technical effectiveness of the event-driven detection and localization method based on underwater buoy acoustic arrays and structural prior constraints described in this application, simulation tests were conducted on the MATLAB platform. It is assumed that the monitored object is an underwater engineering structure located near the seabed, extending horizontally but with planar curvature, its axis exhibiting a gently curving shape in the horizontal plane, and a total equivalent length of approximately 1000 m. An abnormal sound source occurs near this structure, and its radiated acoustic signal propagates into the surrounding water after coupling through the structure.

[0325] A single underwater buoy system is deployed at a fixed location in the water above the structure. Four scalar hydrophones are mounted on the buoy, forming a small-scale, non-coplanar acoustic receiving array. The array size is much smaller than the propagation distance from the sound source to the array, satisfying the far-field approximation condition. The equivalent propagation speed of sound waves in the water is assumed to be constant, and background noise and random disturbances are superimposed in the environment. By changing the signal-to-noise ratio (SNR), the arrival direction of the abnormal sound source is estimated using both a traditional orientation estimation algorithm based on the array manifold and an orientation estimation algorithm based on prior structural constraints introduced in this invention. The stability and error characteristics of the estimation results are compared and analyzed. A schematic diagram of the simulation scenario is shown below. Figure 2 As shown.

[0326] In the above simulation scenario of curved underwater engineering structures, the positioning error performance of various typical array orientation estimation algorithms under different signal-to-noise ratio conditions was compared and analyzed. Figure 3 The positioning error shown is the root mean square (RMSE) value of the estimation error of the anomalous sound source's direction of arrival, which is used to characterize the estimation stability of each algorithm under random noise and environmental disturbance conditions.

[0327] Simulation results show that the traditional Delayed Summation (DSB) algorithm is extremely sensitive to noise under low signal-to-noise ratio (SNR) conditions, and the positioning error increases significantly as the SNR decreases. The MVDR algorithm has a certain ability to suppress interference under medium SNR conditions, but there are still large error fluctuations in the low SNR region. The MUSIC algorithm shows high angle resolution under medium to high SNR conditions, but its performance is significantly affected by the accuracy of covariance matrix estimation under low SNR conditions, and its stability decreases.

[0328] In contrast, the method proposed in this application does not rely on a specific array orientation estimation algorithm. Instead, it addresses the anomalous acoustic localization requirements of underwater linear engineering structures by introducing the stable prior constraint of the spatial orientation of the engineering structure into the array signal processing. This constraint provides engineering consistency limitations and soft guidance to the feasible solution space of the orientation estimation, effectively suppressing random fluctuations in the estimation results under low signal-to-noise ratio (SNR) conditions. Simulation results show that in the low SNR region, the orientation estimation error of the method described in this application is significantly lower than that of traditional DSB and MVDR algorithms, and it exhibits better stability. Under medium-to-high SNR conditions, its performance is on par with traditional high-resolution algorithms, without introducing significant systematic bias.

[0329] The above results show that the method described in this application does not improve performance by pursuing the theoretical limit resolution, but rather improves the stability and robustness of array orientation estimation in complex underwater engineering environments by introducing prior constraints on engineering structures, thereby providing a more reliable orientation information basis for subsequent one-dimensional anomaly location inversion based on structural axis constraints.

[0330] The above provides a detailed description of the underwater engineering structure anomaly detection and localization method based on a submersible buoy acoustic array provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make various improvements and modifications to this invention without departing from the principles of this invention, and these improvements and modifications also fall within the protection scope of the claims of this invention. The above description of the disclosed embodiments enables those skilled in the art to implement or use this invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of this invention. Therefore, this invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for anomaly detection and localization of underwater engineering structures based on a hydrophone array of a submerged buoy, characterized in that, Includes the following steps: S1. Based on the external environmental parameters of the underwater engineering structure and the operating status information of the underwater engineering structure, an equivalent coupled acoustic propagation model is established. S2. Deploy a mooring system in the water area above the underwater engineering structure. The mooring system is equipped with an acoustic receiving array including several scalar hydrophones. Perform time synchronization, spatial position calibration of array elements, and channel gain calibration on the acoustic receiving array. S3. The acoustic receiving array continuously collects multi-channel acoustic signals from the water surrounding the underwater engineering structure. When an abnormal acoustic event with spatially consistent propagation characteristics is detected, the acoustic receiving array enters the enhancement working mode and outputs multi-channel enhanced acoustic signals. S4. In enhanced working mode, output the judgment result of whether an abnormal acoustic event has occurred, and provide the probability of the abnormal event and the confidence level of the judgment. S5. After determining that an abnormal acoustic event has occurred, solve for the estimated position of the abnormal event along the structural axis and the corresponding position information interval. S6. Perform event-driven online correction and update of the equivalent sound velocity and equivalent coupling delay parameters in the equivalent coupled sound propagation model; The specific implementation process of step S4 is as follows: S4.1 Extraction of time-domain features of the enhanced output signal from the acoustic receiving array; In the enhanced working mode of step S3, a sliding time window with length is selected, and time-domain feature extraction is performed on the acoustic receiving array enhanced output signal In the first time window, the array time-domain energy feature is defined as: ， wherein, denotes the start time of the sliding time window, which covers a time interval of ] ; S4.2 Extraction of frequency domain features of the enhanced output signal from the acoustic receiver array: Within the same time window as step S4.1 Within the acoustic receiver array, frequency domain analysis is performed on the enhanced output signal. Acoustic signal for each channel Perform a short-time Fourier transform to obtain its spectral representation. Based on the typical spectral distribution characteristics of the target's abnormal acoustic events, the target frequency band is pre-set. And construct the average frequency domain energy characteristics of the array. : ； S4.3 Construction of spatial correlation characteristics of the enhanced output signal of the acoustic receiving array; The first [channel signal] was calculated based on the multi-channel signal of the acoustic receiving array. Acoustic receiver array spatial consistency index within a time window and will As a spatial correlation feature of the acoustic receiving array, it is directly introduced into the abnormal event judgment process to distinguish sound source events with clear spatial propagation characteristics from spatially unrelated random background noise; For the first The spatial consistency index of the acoustic receiver array within each time window is smoothed within adjacent time windows to obtain the smoothed array spatial consistency characteristics. : ， in, Indicates the number of time windows used for smoothing; S4.4, Characteristic Temporal Evolution Analysis and Joint Decision: Let the first Within a time window, a joint feature vector is constructed, consisting of array time-domain energy characteristics, array frequency-domain energy characteristics, and array spatial consistency characteristics. : ， By comparison The changes within adjacent time windows are analyzed to determine their temporal evolution trend, and a characteristic rate of change index is constructed. : ， When the joint feature vector simultaneously satisfies the following conditions within multiple consecutive windows: the improvement of both the array time-domain energy feature and the array frequency-domain energy feature relative to the background level is not less than 3 dB, preferably 3-10 dB; the spatial consistency index is consistently higher than the consistency decision threshold of 0.5-0.8; and the rate of change of the joint feature within adjacent time windows... If the amplitude does not exceed 20%-50% of the previous window's characteristic value, then the current acoustic event is determined to meet the typical characteristics of the target abnormal acoustic event; S4.5 Output of abnormal event judgment results: When the above joint decision conditions are met, the judgment result of the abnormal acoustic event is output, and the joint feature vector within the corresponding time window is output. Characteristic rate of change index This serves as the foundational data for assessing the probability of abnormal events and the confidence level of decisions.

2. The method for detecting and locating anomalies in underwater engineering structures based on a submersible buoy acoustic array according to claim 1, characterized in that, The specific implementation process of step S1 is as follows: S1.1 Acquisition of environmental parameters and operating status: External environmental parameters include the temperature of the seawater surrounding the underwater engineering structure. ,salinity and water depth ; Engineering operation status information includes the pressure of the medium inside or near the underwater linear engineering structure. medium temperature Relative velocity and the proportion parameters of the medium components The engineering operation status information is described using interval format: ； S1.2 Calculation of initial value of sound velocity in external seawater; Based on the obtained seawater temperature ,salinity and water depth information Calculate the speed of sound in external seawater The initial estimate is: ， in, An empirical model function representing the relationship between seawater sound speed and environmental parameters; S1.3 Initialization of acoustic prior parameters and setting of initial equivalent sound velocity for engineering operation status information: Pressure of the medium inside or near the underwater linear engineering structure medium temperature Relative velocity and the proportion parameters of the medium components The set of prior acoustic parameters of the constituent medium: ， Initial value of external seawater sound speed For reference, the initial value of the equivalent speed of sound Configure settings: ， in, and This represents the dimensionless empirical boundary coefficient, used to define the initial value of the equivalent speed of sound. relative to the initial value of the external seawater sound speed The feasible range; S1.4 Establishment of the equivalent coupled sound propagation model: Assume the spatial orientation or equivalent centerline of the underwater linear engineering structure is determined by the parametric equations. It means that among them This represents a parameter along the axial direction of an underwater linear engineering structure. Let the effective axial length of the underwater linear engineering structure be denoted by ; let the in the acoustic receiving array be denoted by . The spatial location of each scalar hydrophone is The abnormal sound source is located at the position of the structural axis parameter. When the propagation distance to the array element is expressed as: ， An equivalent coupled acoustic propagation model is used to represent the arrival time of the anomalous acoustic signal at the first... Arrival time of each scalar hydrophone: ， in, This represents the initial value of the equivalent speed of sound; This represents the initial value of the equivalent coupling delay, used to characterize the equivalent time offset introduced by factors such as structural coupling, near-field radiation effects, and system fixed delay. Based on the above equivalent coupled sound propagation model, we can obtain any two scalar hydrophones. and The predicted arrival time difference is: 。 3. The method for detecting and locating anomalies in underwater engineering structures based on a submersible acoustic array according to claim 1, characterized in that, The specific implementation process of step S2 is as follows: S2.1 Deployment of the underwater buoy system and local acoustic receiving array structure: A mooring system is deployed in the water area above the underwater linear engineering structure. A rigid mounting frame is set on the top of the mooring system, and four scalar hydrophones are arranged on the mounting frame to form an acoustic receiving array. The acoustic receiving array is a four-element non-coplanar tetrahedral array. S2.2 Time Synchronization and Channel Delay Calibration: Let the first The discrete signals collected by each scalar hydrophone are The first scalar hydrophone is selected as the reference channel. Calibration signals for each channel are acquired using a calibration sound source at a known location. For any given... Normalized cross-correlation function between the calculated channel and the reference channel: ， Based on the peak position of the cross-correlation function, determine the first... Delay estimate of the channel relative to the reference channel: ， Based on this, time delay compensation is performed on the channel signal to align it with the reference channel in time. ， in, Indicates the number after time delay calibration Channel signal; Indicates the first The raw discrete-time acoustic signals acquired from each channel; This indicates the first value estimated through cross-correlation analysis. The time offset of each channel relative to the reference channel; This indicates that the original number will be... The signal sequence after the channel signal is shifted on the discrete time axis; S2.3, Scalar hydrophone spatial position calibration: Establish a local coordinate system for a four-element non-coplanar tetrahedral array. With the geometric center of the mounting bracket as the origin Let the fixed coordinates of the four scalar hydrophones in the local coordinate system be... , The three scalar hydrophones are located on the same horizontal plane. The top two sides form an equilateral triangle: ， The fourth scalar hydrophone is located above. Location: ， in, Indicates the horizontal scale of the array elements; Indicates vertical height, and and They are on the same order of magnitude; The position of the scalar hydrophone in the global coordinate system is denoted as When a mooring buoy experiences attitude angle changes due to ocean currents, the mapping relationship between the local and global coordinate systems can be expressed as a rigid body transformation: ， in, This indicates the position of the mounting bracket origin in the global coordinate system; This represents the rotation matrix measured by the attitude sensor IMU; S2.4 Channel Gain Calibration: Under the same calibration sound source conditions, let the first... The effective signal amplitude of the channel is estimated as follows: ， in, This represents the total number of discrete-time sampling points used for amplitude estimation under the action of a calibrated sound source; This represents the total number of discrete-time sampling points. If the first scalar hydrophone is selected as the reference channel, then the... The gain calibration coefficient for each channel is defined as follows: ， in, Indicates the effective signal amplitude of the reference channel; Based on this, the channel acoustic signal is subjected to amplitude normalization: 。 4. The method for detecting and locating anomalies in underwater engineering structures based on a submersible acoustic array according to claim 1, characterized in that, The specific implementation process of step S3 is as follows: S3.1 Multi-channel acoustic signal acquisition and preprocessing: After time synchronization and channel gain calibration, the four scalar hydrophones in the underwater buoy system output calibrated multi-channel acoustic signals, denoted as: ， in, Indicates the first Continuous or discrete acoustic signals of a scalar hydrophone under a global time reference; The above multi-channel signals are synchronously preprocessed, including DC component removal, bandpass filtering, and multi-channel adaptive noise reduction, to obtain multi-channel acoustic signals for subsequent event triggering and decision-making: ； S3.2 Construction of spatial consistency features of acoustic receiving array: Using scalar hydrophones Calculate the signal in the sliding time window using units of measurement. Cross-correlation function within: ， in, This represents the time delay between channels, i.e., a scalar hydrophone. With scalar hydrophone The relative time offset between received signals; Indicates the start time of the current sliding time window; And based on the cross-correlation function, a scalar hydrophone pair is defined. Normalized correlation coefficient: ， in, Indicates the first The autocorrelation function values ​​of the scalar hydrophone channel signals under zero-delay conditions; Indicates the first The autocorrelation function values ​​of the scalar hydrophone channel signals under zero-delay conditions; Indicates the first The and the first The scalar hydrophone channel signal has a time delay of Cross-correlation function values ​​under the given conditions; For an acoustic receiving array consisting of four scalar hydrophones, six different scalar hydrophone pairs can be formed. The normalized correlation coefficients of each scalar hydrophone pair are combined to construct a spatial consistency index for the acoustic receiving array. ； S3.3 Event Trigger Determination and Enhanced Mode Switching: When the acoustic receiver array spatial consistency index Higher than the preset consistency threshold And the duration of this state exceeds the minimum decision window. When the current acoustic event meets the spatial propagation characteristics of a potential abnormal event, the acoustic receiving array is triggered to switch from the conventional monitoring mode to the enhanced working mode. When the acoustic receiver array spatial consistency index Below the preset consistency threshold If this state lasts for more than the preset recovery time, the acoustic receiving array will automatically return from the enhanced working mode to the normal monitoring mode. S3.4, Multi-channel signal output in enhanced mode: In the enhanced working mode, the multi-channel acoustic signal preprocessed in step S3.1 is... As an acoustic receiving array, it enhances the output signal and simultaneously records the array spatial consistency index within the corresponding time window. And the relevant characteristic parameters of each scalar hydrophone.

5. The method for detecting and locating anomalies in underwater engineering structures based on a submersible buoy acoustic array according to claim 1, characterized in that, The specific implementation process of step S5 is as follows: S5.1 Acquisition of the Time Difference of Arrival Observation: No. The scalar hydrophone and the first A scalar hydrophone pair composed of scalar hydrophones in the corresponding time window Internally calculate the cross-correlation function between the two channel signals: ， And based on the peak position of the cross-correlation function, determine the first... The scalar hydrophone and the first The relative time difference of arrival between individual scalar hydrophones : ， For an acoustic receiving array consisting of four scalar hydrophones, six sets of time difference of arrival (TDOA) observations for scalar hydrophone pairs can be obtained, and these observations can be combined to form a TDOA observation vector. : ； S5.2 Propagation model prediction based on structural axis constraints; Based on the equivalent coupled sound propagation model established in step S1, assuming the abnormal sound source is located on the structural axis or equivalent centerline of a known underwater linear engineering structure, at the equivalent sound velocity... Under these conditions, for any given axial position parameter Based on the equivalent propagation distance from the abnormal sound source to each array element, the corresponding predicted time difference of arrival is calculated, and a predicted time difference of arrival vector is constructed: ； S5.3 One-dimensional inversion localization based on axial prior constraints: By introducing prior structural constraints that the anomalous sound source is located on a known structural axis or equivalent centerline, a cost function is constructed with the deviation between the observed and predicted arrival time differences as the objective: ， in, Represents the L2 norm; Within a given axial range Internally, the cost function is optimized using one-dimensional search or numerical optimization methods. Solving for the axial position parameters that minimize the cost function yields the following: ， in, This represents an estimated value indicating the axial location of the abnormal sound source. S5.4, Location Confidence Interval and Stability Assessment: Axial position estimate Nearby, for the cost function The variation characteristics are analyzed based on the preset error tolerance parameters. Define the axial confidence interval for axial position. for: 。 6. The method for detecting and locating anomalies in underwater engineering structures based on a submersible buoy acoustic array according to claim 1, characterized in that, The specific implementation process of step S6 is as follows: S6.1 Selection of Updatable Parameters: Based on the equivalent coupled sound propagation model established in step S1, the model parameter vector selected to describe the propagation characteristics of abnormal sound sources is as follows: ， in, It represents the equivalent speed of sound and is used to comprehensively characterize the combined effects of external seawater propagation characteristics, structural coupling effects, and unexplicitly modeled factors on propagation speed. It represents the equivalent pipe wall acoustic coupling delay, used to characterize the equivalent time offset introduced by factors such as structural transmission / radiation, near-field effects, and system fixed delay; S6.2 Calculation of parameter correction based on observation-prediction bias: The estimated axial position of the abnormal sound source calculated in step S5 Obtain the time difference of arrival observation vector observed by the acoustic receiving array within the corresponding time window. ; In the current model parameters Under the given conditions, based on the equivalent coupled sound propagation model, the location of the abnormal sound source is calculated. The corresponding predicted arrival time difference vector ; The deviation vector between the observed time difference of arrival (TDOA) and the predicted time difference of arrival (TDOA) is calculated by comparing the two vectors. ； S6.3 Constrained Correction Update of Equivalent Sound Propagation Parameters: Define a cost function targeting the arrival time difference residual: ， By minimizing the cost function The parameters of the equivalent coupled sound propagation model are updated by inversion: ， The update process is simplified to the equivalent speed of sound One-dimensional or low-dimensional optimization, with constraints on the update magnitude: ， in, This indicates the preset maximum update step size, used to ensure the smoothness and engineering controllability of the parameter update process of the equivalent coupled acoustic propagation model; S6.4 Information Upload and Alarm Handling: After completing the parameter correction and update of the equivalent coupled acoustic propagation model, the judgment results of abnormal events, the estimated axial position, the corresponding axial confidence interval, the probability of occurrence of abnormal events and the decision confidence information, as well as the operating status information of the underwater glider system, are transmitted to the monitoring platform. When the information meets the preset alarm conditions, the monitoring platform triggers a leakage alarm and stores the relevant data to support subsequent operation and maintenance decisions and event retrospective analysis.

7. The method for detecting and locating anomalies in underwater engineering structures based on a submersible acoustic array according to claim 1, characterized in that, The correction and update of the equivalent coupled acoustic propagation model parameters adopts an event-driven triggering mechanism, which is executed when the following conditions are met simultaneously: the abnormal event judgment result in step S4 is true; the center localization result in step S5 is true. It is stable, and the positioning information interval corresponding to the positioning result is less than the preset threshold range.

Images