A multi-channel internal wave flow early warning system and early warning measurement method

Abstract

The application belongs to the technical field of marine environment monitoring and early warning, and specifically relates to a multi-channel internal wave flow early warning system and an early warning measurement method, which comprises the following steps: a buoy body is fixedly arranged through a fixed anchor, and a marine profile comprehensive detection module synchronously collects multi-dimensional profile data such as flow velocity, temperature and salinity. A main control unit is embedded with a mixed driving early warning model, a phase velocity of an internal wave is solved based on a physical model, a nonlinear correction is performed through a machine learning residual correction module, a high-precision internal wave phase velocity is obtained, and then early warning time, shear flow and displacement amplitude are calculated and an alarm level is determined. Alarm information is redundantly sent to a shore station receiving system through a ground mobile communication link and a satellite communication link of a multi-channel communication module, and the shore station receiving system regularly updates the machine learning model by taking measured data as incremental samples. The application realizes high-precision and low-delay early warning of internal wave flow, has high communication reliability, and is suitable for marine engineering, underwater vehicles and safety guarantee of offshore operations.

Description

Technical Field

[0001] This invention belongs to the field of marine environmental monitoring and early warning technology, specifically a multi-channel internal wave current early warning system and early warning measurement method. Background Technology

[0002] Internal wave currents pose a serious threat to marine engineering activities. In deep-water drilling operations, internal wave currents can cause platform displacement, riser damage, and subsea wellhead instability. In operations on semi-submersible drilling platforms and anchored oil depots, the sudden, strong shear currents generated by internal wave currents can cause anchor chain breakage and platform drift, potentially resulting in casualties and property damage. During subsea pipeline laying and reconnection operations, internal wave currents can interfere with the precise positioning of pipelines, increasing operational risks and construction difficulty. Furthermore, the presence of internal waves can alter the propagation characteristics of ocean sound waves, severely affecting sonar functionality and interfering with submarine concealment and monitoring technologies.

[0003] Currently, the main shortcomings of monitoring and early warning technologies for internal wave currents are as follows:

[0004] (1) Single monitoring dimension and neglect of parameter coupling: Existing internal wave monitoring mainly relies on a single type of sensor, such as acoustic Doppler velocity profiler (ADCP) or temperature-salinity-depth meter (CTD). A single sensor can only acquire limited parameters such as velocity or temperature-salinity, making it difficult to comprehensively characterize the hydrodynamic features of internal wave flow. ADCP can provide high temporal resolution velocity profile data, but it is not sensitive to changes in density field—while the essence of internal waves is fluctuation in density-stratified fluids, and velocity information alone cannot accurately determine the amplitude and energy of internal waves. CTD can accurately measure the vertical structure of temperature-salinity, but it cannot acquire velocity information in real time. The coupling relationship between the velocity field and the density field is ignored, resulting in a fundamental error in the inversion of internal wave parameters.

[0005] (2) The early warning models are too simplified and lack nonlinear characterization capabilities: Existing internal wave early warning methods are mostly based on simplified theoretical models (such as the KdV equation, eKdV equation, etc.). These models have good predictive capabilities under weak nonlinear conditions, but internal waves in the actual marine environment often exhibit strong nonlinear characteristics, especially in shallow sea shelf areas and internal wave breaking zones. A single theoretical model is difficult to accurately describe the entire process of internal wave generation, propagation, and evolution. Some studies have attempted to use methods such as Gaussian function models to calculate internal solitary wave parameters, but there is still considerable room for improvement in areas such as noise processing of measured data, separation of background flow fields, and characterization of nonlinear effects.

[0006] (3) Poor timeliness of early warning and inability to achieve rolling prediction: Traditional internal wave early warning mostly relies on satellite remote sensing (long revisit period) or numerical prediction models (long calculation time), which is difficult to meet the safety requirements of real-time maritime operations. Moreover, most existing systems are "threshold triggered" type alarms, which can only tell "whether the threshold has been exceeded" and cannot provide prediction information such as "when the internal wave will arrive in the future and how strong it will be".

[0007] (4) Insufficient flexibility of fixed monitoring platforms: Most existing internal wave monitoring systems use fixed buoys or underwater moorings, which can only provide fixed-point data and cannot flexibly adjust the monitoring position according to changes in the marine environment or engineering operation requirements. In offshore engineering operations, the operation point changes continuously with the progress of the project, and fixed monitoring systems can hardly provide real-time internal wave current information for the operation point. Summary of the Invention

[0008] The purpose of this invention is to provide a multi-channel internal wave flow early warning system and its measurement method, which overcomes the shortcomings of existing technologies such as "few information dimensions, low accuracy of early warning models, lack of predictive ability, and inflexible deployment". Through three major innovations, namely flow velocity-density coordinated observation, hybrid driven early warning model, and mobile-fixed network observation, it realizes real-time, accurate, and predictive early warning of internal wave flow.

[0009] The technical solution adopted by the present invention to achieve the above objectives is: a multi-channel internal wave current early warning system, comprising: a buoy body, a multi-channel communication module, a fixed anchorage, a comprehensive ocean profile detection module, a shore station receiving system, and a mobile observation device;

[0010] The buoy body serves as a surface support platform and is fixed at a fixed point at sea by a fixed anchorage.

[0011] The multi-channel communication module is fixed to the top of the buoy body and is used to achieve ground mobile communication via satellite communication.

[0012] The ocean profile integrated detection module is fixed on the buoy body, and there are no obstructions in the downward detection direction of the ocean profile integrated detection module, which is used to simultaneously collect multi-dimensional parameters of the underwater environment.

[0013] The shore-based receiving system is used to receive observation data collected by the multi-channel communication module through the ocean profile integrated detection module, and connect to the ground mobile communication network through the Internet interface to store and analyze the data received through the satellite communication link and the ground mobile communication link. At the same time, it can visualize and display early warning information and send remote control commands to the buoy through the multi-channel communication module.

[0014] The mobile observation device is a wave glider or unmanned surface vessel equipped with a miniature ADCP and a miniature temperature and salinity sensor; the mobile observation device and the buoy establish a master-slave data interaction link through a multi-channel communication module to perform encrypted observation, maneuvering replenishment or wide-area scanning tasks.

[0015] The integrated ocean profile detection module includes:

[0016] Acoustic Doppler current profiler is used to detect the horizontal velocity components of multiple ocean profiles. and vertical velocity component ;

[0017] A CTD array consists of multiple temperature and salinity sensors arranged at equal intervals along the cross-sectional direction to simultaneously acquire the temperature of each depth layer. and salinity data;

[0018] The data fusion unit, connected to the acoustic Doppler current profiler and CTD array, is used to perform joint inversion of velocity data and temperature-salinity data in the spatiotemporal domain.

[0019] The data fusion unit includes: a density calculation submodule and a floating frequency calculation submodule;

[0020] The density calculation submodule is used to calculate seawater density based on temperature T, salinity S, and pressure P, using the seawater state equation. ;

[0021] The buoyancy frequency calculation submodule is connected to the density calculation submodule and is used to calculate based on seawater density. Calculate the floating frequency .

[0022] The buoy body includes: a buoy body shell and a main control unit, a battery positioning module, buoyancy material, and a solar panel disposed inside the buoy body shell;

[0023] The main control unit is used to receive and process the data collected by the ocean profile integrated detection module, perform internal wave current identification and alarm level determination, and send the calculated alarm information results to the shore station receiving system through the multi-channel communication module.

[0024] The battery is used to provide working power for the main control unit, positioning module, multi-channel communication module and marine profile integrated detection module, and its output terminal is electrically connected to the power input terminal of the main control unit.

[0025] The solar panel is used to convert solar energy into electrical energy and charge the battery, and its electrical output terminal is electrically connected to the charging input terminal of the battery.

[0026] The positioning module is used to obtain the real-time geographical location information of the buoy, and its data output end is connected to the data acquisition end of the main control unit.

[0027] The buoyancy material is filled inside the outer shell of the buoy body to provide the buoy body with the buoy force required for floating on the water surface, and to enclose and fix the main control unit, battery and positioning module inside the outer shell of the buoy body.

[0028] The main control unit is embedded with a hybrid drive early warning model, which includes a physical model module and a machine learning residual correction module.

[0029] The physical model module is used to solve the linear wave characteristic equation of the internal wave to obtain the linear theoretical phase velocity based on the density profile and buoyancy frequency data obtained by the integrated ocean profile detection module (4).

[0030] The machine learning residual correction module is connected to the output of the physical model module. It is used to output the phase velocity residual prediction value based on the received real-time marine environment characteristic parameters and the pre-trained machine learning model, and to perform nonlinear correction on the linear theoretical phase velocity to obtain the corrected internal wave phase velocity.

[0031] The multi-channel communication module includes: a signal strength monitoring unit, a channel switching execution unit, and a redundant transmission control unit;

[0032] The signal strength monitoring unit is used to detect the signal strength of the terrestrial mobile communication link in real time;

[0033] The channel switching execution unit is connected to the signal strength monitoring unit and is used to automatically activate the satellite communication link when the ground mobile communication signal strength is lower than a preset threshold.

[0034] The redundant transmission control unit is connected to the main control unit and is used to control the alarm information to be transmitted in parallel through at least two different communication links.

[0035] A method for early warning measurement in a multi-channel internal wave current early warning system includes the following steps:

[0036] Step S1: Control the acoustic Doppler current profiler in the integrated ocean profile detection module to collect the horizontal velocity components of multiple ocean profiles. and vertical velocity component Simultaneously, the CTD array in the ocean profiling module is controlled to synchronously acquire the temperature of each depth layer. and salinity data;

[0037] Step S2: The data fusion unit in the ocean profile integrated detection module aligns the current velocity data with the temperature and salinity data in the time and spatial domains and performs joint inversion;

[0038] Step S3: The physical model module in the main control unit solves the linear wave characteristic equation of the internal wave based on the density profile and buoyancy frequency data obtained by inversion, and obtains the linear theoretical phase velocity;

[0039] Step S4: The machine learning residual correction module in the main control unit receives real-time marine environmental characteristic parameters, outputs the phase velocity residual prediction value based on the pre-trained machine learning model, and performs nonlinear correction on the linear theoretical phase velocity to obtain the corrected internal wave phase velocity; at the same time, the main control unit calculates the warning time, shear flow intensity and displacement amplitude A based on the corrected internal wave phase velocity, and determines the alarm level based on the shear flow intensity and displacement amplitude A.

[0040] Step S5: The main control unit will send the alarm information, which includes the alarm level and the warning time, through the terrestrial mobile communication link in the multi-channel communication module to send the complete alarm information data packet; at the same time, it will send the coded data containing the alarm level and the predicted arrival time to the shore station receiving system through the satellite communication link in the multi-channel communication module.

[0041] Step S6: The shore station receiving system uses the newly acquired measured flow velocity data, temperature and salinity data, and corresponding internal wave event records as incremental training samples to periodically update and train the machine learning model in the machine learning residual correction module.

[0042] In step S2, the data fusion unit performs joint inversion, specifically including the following steps:

[0043] Step S2-1: Based on temperature and salinity The density of seawater was calculated using the UNESCO International Equation of State for Seawater, along with the pressure P at the corresponding depth. for:

[0044]

[0045] Step S2-2: Based on the seawater density Calculate the floating frequency for:

[0046]

[0047] Where g is the acceleration due to gravity, and z is the depth coordinate, with vertical upward being defined as positive;

[0048] Step S2-3: Combine the vertical shear rate of the flow velocity measured by the acoustic Doppler current profiler with the seawater density. The vertical mode structure and horizontal wavenumber of the internal waves are inverted using the Taylor-Goldstein equation.

[0049] In step S3, the physical model module solves the characteristic equation of the internal wave linear wave, specifically as follows:

[0050] Floating frequency Substituting into the Taylor-Goldstein equation:

[0051]

[0052] Where c is the internal wave phase velocity, For horizontal wavenumber, The vertical velocity eigenfunction;

[0053] The eigenvalue problem is solved by combining preset boundary conditions, and the linear theoretical phase velocity of each mode is extracted.

[0054] Step S4 specifically includes the following sub-steps:

[0055] Step S4-1: Input the feature vector composed of the currently measured velocity profile, density profile, tidal phase, and background flow field into the pre-trained machine learning model. The machine learning model is a long short-term memory network model or a random forest model, and outputs the predicted phase velocity residual value. ;

[0056] Step S4-2: Calculate the corrected internal wave phase velocity : .

[0057] Step S4-3: Based on the corrected internal wave phase velocity And the known distance L between the current observation point and the protected target, calculate the warning time. for:

[0058]

[0059] in, This is the preset system processing delay time;

[0060] Step S4-4: Based on the vertical velocity eigenfunction obtained from the inversion... and horizontal wavenumber Combined with floating frequency and the corrected internal wave phase velocity The eigenfunction of the horizontal flow velocity is obtained through the linear internal wave polarization relation. and displacement eigenfunctions of isodense surfaces ;

[0061] Step S4-5: Estimate the current internal wave amplitude using the internal wave amplitude projection method. The predicted maximum shear flow is:

[0062]

[0063] in, This represents the current internal wave amplitude; For depth coordinates Maximum value operation; eigenfunctions of horizontal flow velocity The modulus of the derivative with respect to depth; It is the horizontal velocity distribution of the internal wave vertical mode, and its vertical gradient represents the velocity shear eigenfunction; The predicted displacement amplitude of the isodensity surface;

[0064] The displacement amplitude is then:

[0065] ;

[0066] in, Let be the magnitude of the displacement eigenfunction of the isodense surface;

[0067] Step S4-6: Based on the predicted maximum shear flow Based on the predicted displacement amplitude A of the isodense surface, the alarm level is determined by performing fuzzy logic judgment, namely:

[0068] when or At that time, it was determined to be a Level 1 concern;

[0069] when or At that time, it was determined to be a Level 2 warning;

[0070] when or At that time, it was determined to be a Level III emergency.

[0071] The present invention has the following beneficial effects and advantages:

[0072] 1. This invention achieves, for the first time, the synchronous acquisition and joint inversion of velocity and density fields through the coordinated observation of ADCP and CTD arrays, solving the fundamental defect of traditional methods that "only consider velocity and not density", and improving the accuracy of internal wave parameter inversion by more than 30%.

[0073] 2. This invention upgrades from "current threshold triggering" to "future prediction and early warning." Based on a hybrid driving model, it can predict the arrival time and intensity of internal waves 5-30 minutes in advance, gaining a valuable response window for offshore operations. Compared with a pure physical model, the phase velocity prediction error is reduced by 40%-60% after introducing machine learning residual correction.

[0074] 3. The physical model of this invention ensures the interpretability of the prediction results and the basic physical constraints. The machine learning model learns the nonlinear deformation characteristics unique to the local sea area. The combination of the two makes the model more robust under different seasons and tidal conditions, and reduces the false alarm rate to below 5%.

[0075] 4. This invention achieves a leap in observation capabilities from "single-point observation" to "cross-section + region" through the networked collaboration of mobile observation units and fixed buoys. The mobile unit can perform intensive observations, provide temporary backup, and conduct wide-area scanning, significantly improving the system's environmental adaptability and mission flexibility.

[0076] 5. The multi-channel adaptive switching + redundant transmission strategy of this invention ensures that alarm information can be reliably delivered with the lowest delay, whether at sea or near shore. Attached Figure Description

[0077] Figure 1 This is a schematic diagram of the overall structure of the present invention;

[0078] Figure 2 This is a schematic diagram of the structure of the buoy body of the present invention;

[0079] Figure 3 This is a schematic diagram of the shore station receiving system of the present invention;

[0080] Figure 4 This is a flowchart of the early warning measurement method of the present invention;

[0081] Among them, 1 is the buoy body, 2 is the multi-channel communication module, 3 is the fixed anchor, 4 is the ocean profile integrated detection module, 5 is the shore station receiving system, 101 is the main control unit, 102 is the battery, 103 is the positioning module, 104 is the buoyancy material, 105 is the solar panel, 501 is the satellite receiver, 502 is the network cable, and 503 is the server. Detailed Implementation

[0082] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0083] like Figure 1 As shown, this embodiment provides a multi-channel internal wave current early warning system, including: a buoy body 1, a multi-channel communication module 2, a fixed mooring 3, a marine profile integrated detection module 4, a shore station receiving system 5, and a mobile observation device (not shown in the figure).

[0084] Buoy 1 is fixed at a predetermined location in the sea area by a fixed anchorage 3. The length of the anchorage is designed according to the water depth to ensure that the horizontal displacement of the buoy does not exceed 5% of the water depth under strong current conditions. Buoy 1 serves as a surface platform and integrates a main control unit 101, a battery 102, a positioning module 103, buoyancy materials, and a solar panel 105. A multi-channel communication module 2 is installed on the mast at the top of the buoy and includes a 4G / 5G module and a BeiDou / Iridium satellite communication module, with signal strength monitoring and adaptive switching capabilities. An ocean profile detection module 4 is fixed to the bottom support of buoy 1, providing unobstructed downward detection coverage of the entire water layer from 2m below the surface to 10m above the seabed. The mobile observation device is a wave glider equipped with a miniature ADCP and a miniature temperature and salinity sensor, which establishes a master-slave data interaction link with buoy 1 through the multi-channel communication module 2.

[0085] 1. Integrated Ocean Profiling Module

[0086] This module consists of three parts:

[0087] Acoustic Doppler velocity profiler: using a 300kHz or 600kHz ADCP, set to 25 layers (bins), each layer thickness 2-5m, sampling interval 1min, outputting horizontal velocity components. and vertical velocity component .

[0088] CTD array: 8–16 miniature temperature and salinity sensors (using RBRsolo³ or Sea-BirdSBE37) are arranged at equal intervals along the cable to simultaneously collect temperature data at various depth levels. and salinity The sampling rate is aligned with ADCP.

[0089] Data fusion unit: Built into the main control unit 101, it includes a density calculation submodule and a floating frequency calculation submodule. Its execution steps are as follows:

[0090] 1-1) Calculate seawater density using the UNESCO International Equation of State for Seawater ;

[0091] 1-2) Calculate the floating frequency: ;

[0092] Combining ADCP with vertical velocity shear, the internal wave vertical mode structure is inverted using the Taylor-Goldstein equation. and horizontal wavenumber .

[0093] 2. Buoy body

[0094] like Figure 2The diagram shows the internal structure of the buoy body 1. The buoy body 1 includes a buoy body shell and a main control unit 101, a battery 102, a positioning module 103, a buoyancy material 104, and a solar panel 105 disposed on the shell.

[0095] The main control unit 101 is used to receive and process the data collected by the ocean profile integrated detection module 4, perform internal wave current identification and alarm level determination, and send the calculated alarm information results to the shore station receiving system 5 through the multi-channel communication module 2.

[0096] Battery 102 is used to provide working power to the main control unit 101, positioning module 103, multi-channel communication module 2 and ocean profile integrated detection module 4, and its output terminal is electrically connected to the power input terminal of the main control unit 101.

[0097] The solar panel 105 is used to convert solar energy into electrical energy and charge the battery 102, and its electrical output terminal is electrically connected to the charging input terminal of the battery 102.

[0098] The positioning module 103 is used to obtain the real-time geographical location information of the buoy body 1, and its data output end is connected to the data acquisition end of the main control unit 101;

[0099] Buoyancy material 104 is filled inside the buoy body shell to provide buoyancy required for the buoy body 1 to float on the water surface, and to enclose and fix the main control unit 101, battery 102 and positioning module 03 inside the buoy body shell.

[0100] 3. Main control unit 101 and hybrid drive early warning model

[0101] The main control unit uses an ARM embedded industrial control board (in this embodiment, the main frequency is 1.5GHz and the memory is 4GB), with a built-in Linux real-time operating system. Its embedded hybrid driver early warning model includes:

[0102] Physical Model Module: Solve the eigenvalue problem of the Taylor-Goldstein equations to obtain the linear theoretical phase velocity. The boundary conditions are set to: sea surface. ,seabed (Rigid boundary).

[0103] Machine learning residual correction module: Employs a Long Short-Term Memory (LSTM) network. The number of input layer nodes corresponds to the feature vectors (including velocity profile, density profile, tidal phase, and background flow field), and the output layer is the phase velocity residual. The model is pre-trained using historical internal wave event data from this sea area spanning more than one year, with the loss function being the root mean square error (RMSE).

[0104] 4. Multi-channel communication module

[0105] This module includes:

[0106] Signal strength monitoring unit: Detects 4G / 5G signal strength (RSRP value) every 10 seconds.

[0107] Channel switching execution unit: When RSRP < -110dBm for 30 seconds, automatically switch to BeiDou short message or Iridium link.

[0108] Redundant transmission control unit: Alarm information (including alarm level, predicted arrival time, and location) is transmitted simultaneously via 4G and satellite links. Ordinary observation data is transmitted only via 4G; the satellite link is used solely for alarms and remote control commands.

[0109] 5. Shore station receiving system

[0110] like Figure 3 The diagram shows the structure of the shore-based receiving system 5. The shore-based receiving system 5 includes a satellite receiver 501, a network cable 502, and a server 503. The satellite receiver 501 receives satellite link data from the multi-channel communication module 2 and transmits it to the server 503 via the network cable 502. The server 503 also connects to the terrestrial mobile communication network via an internet interface, receives 4G / 5G link data, stores, analyzes, and visualizes it, and can send remote control commands to the buoy 1 via the multi-channel communication module 2.

[0111] 6. Mobile observation device

[0112] The wave glider is equipped with a miniature ADCP (1MHz, 16 layers) and a miniature temperature and salinity sensor (sampling rate 0.5Hz). Its coordination mode with the buoy is remotely set by the shore station.

[0113] Encrypted observation mode: When the hybrid drive model predicts that the internal wave will arrive within 30 minutes, the shore station sends a command to move the observation device to the predicted path and perform an encrypted scan of the 500m cross section.

[0114] Mobile replacement mode: If the main control unit of the buoy detects an ADCP abnormality during self-test, it will automatically request a mobile observation device to temporarily replace it.

[0115] Wide-area scanning mode: During periods without internal wave warnings, the mobile observation device scans the temperature, salinity, and velocity fields within a 10km radius along a triangular cruising route, and the data is used to update the LSTM model.

[0116] The following section uses a typical internal wave early warning event to illustrate the working steps of this embodiment in detail.

[0117] like Figure 4The diagram shown is a flowchart of the method of the present invention. This embodiment of an early warning measurement method for a multi-channel internal wave current early warning system includes the following steps:

[0118] Step S1: Control the acoustic Doppler current profiler in the ocean profile integrated detection module 4 to collect the horizontal current velocity components of multiple ocean profiles. and vertical velocity component Simultaneously, the CTD array in the ocean profile integrated detection module 4 is controlled to synchronously acquire the temperature of each depth layer. and salinity data;

[0119] Step S2: The data fusion unit in the ocean profile integrated detection module 4 aligns the current velocity data with the temperature and salinity data in the time and spatial domains and performs joint inversion;

[0120] Step S3: The physical model module in the main control unit 101 solves the linear wave characteristic equation of the internal wave based on the density profile and buoyancy frequency data obtained by inversion, and obtains the linear theoretical phase velocity;

[0121] Step S4: The machine learning residual correction module in the main control unit 101 receives real-time marine environmental characteristic parameters, outputs the phase velocity residual prediction value based on the pre-trained machine learning model, and performs nonlinear correction on the linear theoretical phase velocity to obtain the corrected internal wave phase velocity; at the same time, the main control unit 101 calculates the warning time, shear flow intensity and displacement amplitude A based on the corrected internal wave phase velocity, and determines the alarm level based on the shear flow intensity and displacement amplitude A.

[0122] Step S5: The main control unit 101 sends the alarm information, which includes the alarm level and the warning time, to the ground mobile communication link in the multi-channel communication module 2 first; at the same time, it sends the coded data containing the alarm level and the predicted arrival time to the shore station receiving system 5 through the satellite communication link in the multi-channel communication module 2.

[0123] Step S6: The shore station receiving system 5 uses the newly acquired measured flow velocity data, temperature and salinity data and corresponding internal wave event records as incremental training samples to periodically update and train the machine learning model in the machine learning residual correction module.

[0124] Example 1

[0125] Step S1: Synchronous data acquisition in this embodiment

[0126] In this embodiment, it is assumed that an internal isolated wave occurs in a certain sea area, and the ADCP and CTD arrays on buoy 1 begin synchronous data acquisition at 10:00:00 on August 15, 2025. The ADCP outputs current velocity data for 30 layers, and the CTD array outputs temperature and salinity at the corresponding depths. The main control unit 101 timestamps and caches the data.

[0127] Step S2: Joint Inversion

[0128] Data fusion unit execution:

[0129] S2-1: Calculate the seawater density at each depth and time. .

[0130] This embodiment uses the UNESCO International Equation of State for Seawater to calculate seawater density. for:

[0131]

[0132] S2-2: Calculation with a difference interval of 2m And smooth it out.

[0133] Calculate the floating frequency for:

[0134]

[0135] Where g is the acceleration due to gravity, and z is the depth coordinate, with vertical upward being defined as positive;

[0136] S2-3: Will Substituting the velocity shear rate into the Taylor-Goldstein equations, the vertical velocity eigenfunctions are obtained using the target-shooting method. and horizontal wavenumber (Corresponding wavelength approximately 524m).

[0137] Step S3: Physical Model Baseline

[0138] The physical model module solves the eigenvalue equations:

[0139]

[0140] Where c is the internal wave phase velocity, For horizontal wavenumber, The vertical velocity eigenfunction;

[0141] Given the boundary conditions, extract the linear theoretical phase velocity of the first mode. .

[0142] Step S4: Machine learning correction and alarm level determination

[0143] S4-1: Construct the current feature vector: velocity profile (30-dimensional), density profile (30-dimensional), tidal phase (calculated from astronomical calculations, 3 days after spring tide), and background flow field (average flow velocity of 0.1 m / s eastward over the past hour). Input the pre-trained LSTM model and output the residual. .

[0144] S4-2: Correction of phase velocity .

[0145] S4-3: The distance from the known observation point to a certain drilling platform System processing delay The warning time is as follows:

[0146]

[0147] S4-4: Calculation of the eigenfunction magnitude of horizontal flow velocity based on polarization relation: and the magnitude of displacement eigenfunctions of isodense surfaces .

[0148] S4-5: Estimating the current internal wave amplitude using the amplitude projection method ,and then:

[0149]

[0150]

[0151] S4-6: Determine the alarm level. Because... Falling Scope, at the same time Falling The range triggered a level 2 warning.

[0152] Step S5: Multi-channel alarm transmission

[0153] The main control unit 101 generates an alarm data packet, including: timestamp, location (22.5°N, 119.5°E), alarm level (Level 2), predicted arrival time (71.4 minutes later), predicted maximum shear flow (0.288 s⁻¹), and predicted amplitude (19.2 m). The complete data packet (approximately 2KB) is sent via the 4G link first; simultaneously, a simplified code "LV2,71min,0.288,19.2" is sent via BeiDou short message service. The shore station receiving system 5 receives both messages simultaneously within 2 seconds and verifies their accuracy.

[0154] Step S6: Incremental Model Update

[0155] Based on the early warning, the drilling platform shut down the blowout preventer and evacuated personnel in advance. The internal wave actually arrived 71 minutes later, with a measured maximum shear flow of 0.295 s⁻¹ and an amplitude of 20.1 m. The shore-based server added this event as an incremental sample to the training set, and fine-tuned the LSTM model using the most recent 30 internal wave events in the early morning of the following day (learning rate 0.001, training for 3 epochs). The updated model was then deployed to the buoy's master control unit.

[0156] Example 2:

[0157] Twenty minutes before the aforementioned warning event (at 10:00), the shore station, based on a hybrid drive model, predicted that the internal wave would propagate in a southeast-northwest direction and immediately sent a command to the wave glider for intensive observation. The wave glider, from its standby position (2 km from the buoy), headed towards the predicted path, arriving at the designated section at 10:15 and beginning a lateral traverse at a speed of 1 m / s, continuously collecting ADCP and temperature-salinity data from six profiles. After these data were transmitted back, the shore station system deduced the horizontal spatial scale of the internal wave (approximately 800 m) and corrected the arrival time, further improving the accuracy of the warning.

[0158] After six months of comparative testing at sea, this system outperforms the traditional single ADCP threshold alarm method in the following ways:

[0159] The prediction error for internal wave arrival time was reduced from ±15 minutes to ±4 minutes;

[0160] The amplitude prediction error was reduced from ±8m to ±2.5m;

[0161] The false alarm rate decreased from 18% to 3.5%;

[0162] Communication availability (offshore area) increased from 82% to 99.7% (due to automatic switching of redundant channels).

[0163] In summary, this invention achieves accurate inversion of internal wave current parameters, nonlinear correction of phase velocity, dynamic calculation of warning time, and fuzzy logic determination of alarm levels by networking and coordinating fixed buoys and mobile observation devices with a comprehensive ocean profile detection module, and utilizing an embedded hybrid-driven early warning model (physical model + machine learning residual correction). Simultaneously, the adaptive switching and redundant transmission strategy of the multi-channel communication module ensures reliable transmission of early warning information in open seas and under severe weather conditions. The shore-based receiving system continuously optimizes the machine learning model through incremental learning, enabling the system to adapt to environmental changes in different seasons and sea areas. Compared with existing technologies, this invention significantly improves the timeliness, accuracy, and communication reliability of internal wave early warning, providing effective safety guarantees for marine engineering activities such as offshore oil drilling platforms, subsea pipelines, and underwater vehicles.

[0164] Those skilled in the art will understand that the above description is merely a preferred embodiment of the present invention, and the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. This is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0165] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A multi-channel internal wave current early warning system, characterized in that, include: The system includes a buoy body (1), a multi-channel communication module (2), a fixed anchorage (3), a comprehensive ocean profile detection module (4), a shore-based receiving system (5), and a mobile observation device. The buoy body (1) serves as a surface support platform and is fixed at a fixed point at sea by a fixed anchorage (3); The multi-channel communication module (2) is fixed to the top of the buoy body (1) and is used to achieve ground mobile communication via satellite communication. The ocean profile integrated detection module (4) is fixed on the buoy body (1), and there are no obstructions in the downward detection direction of the ocean profile integrated detection module (4), which is used to synchronously collect multi-dimensional parameters of the underwater environment; The shore station receiving system (5) is used to receive the observation data collected by the multi-channel communication module (2) through the ocean profile integrated detection module (4), and connect to the ground mobile communication network through the Internet interface to store and analyze the data received through the satellite communication link and the ground mobile communication link. At the same time, it visualizes the early warning information and sends remote control commands to the buoy body (1) through the multi-channel communication module (2). The mobile observation device is a wave glider or unmanned surface vessel equipped with a miniature ADCP and a miniature temperature and salinity sensor; the mobile observation device and the buoy (1) establish a master-slave data interaction link through a multi-channel communication module (2) to perform encrypted observation, maneuvering replenishment or wide-area scanning tasks.

2. The multi-channel internal wave flow early warning system according to claim 1, characterized in that, The integrated ocean profile detection module (4) includes: Acoustic Doppler current profiler is used to detect the horizontal velocity components of multiple ocean profiles. and vertical velocity component ; A CTD array consists of multiple temperature and salinity sensors arranged at equal intervals along the cross-sectional direction to simultaneously acquire the temperature of each depth layer. and salinity data; The data fusion unit, connected to the acoustic Doppler current profiler and CTD array, is used to perform joint inversion of velocity data and temperature-salinity data in the spatiotemporal domain.

3. The multi-channel internal wave flow early warning system according to claim 2, characterized in that, The data fusion unit includes: a density calculation submodule and a floating frequency calculation submodule; The density calculation submodule is used to calculate seawater density based on temperature T, salinity S, and pressure P, using the seawater state equation. ; The buoyancy frequency calculation submodule is connected to the density calculation submodule and is used to calculate based on seawater density. Calculate the floating frequency .

4. The multi-channel internal wave flow early warning system according to claim 1, characterized in that, The buoy body (1) includes: a buoy body shell and a main control unit (101), a battery (102), a positioning module (103), a buoyancy material (104), and a solar panel (105) disposed inside the buoy body shell. The main control unit (101) is used to receive and process the data collected by the ocean profile integrated detection module (4), perform internal wave current identification and alarm level determination, and send the calculated alarm information results to the shore station receiving system (5) through the multi-channel communication module (2); The battery (102) is used to provide working power for the main control unit (101), positioning module (103), multi-channel communication module (2) and marine profile integrated detection module (4), and its output terminal is electrically connected to the power input terminal of the main control unit (101). The solar panel (105) is used to convert solar energy into electrical energy and charge the battery (102), and its electrical energy output terminal is electrically connected to the charging input terminal of the battery (102). The positioning module (103) is used to obtain the real-time geographical location information of the buoy (1), and its data output end is connected to the data acquisition end of the main control unit (101); The buoyancy material (104) is filled inside the outer shell of the buoy body to provide the buoy body (1) with the buoy force required for floating on the water surface, and to wrap and fix the main control unit (101), battery (102) and positioning module (103) inside the outer shell of the buoy body.

5. A multi-channel internal wave flow early warning system according to claim 4, characterized in that, The main control unit (101) is embedded with a hybrid drive early warning model, which includes: a physical model module and a machine learning residual correction module; The physical model module is used to solve the linear wave characteristic equation of the internal wave to obtain the linear theoretical phase velocity based on the density profile and buoyancy frequency data obtained by the integrated ocean profile detection module (4). The machine learning residual correction module is connected to the output of the physical model module. It is used to output the phase velocity residual prediction value based on the received real-time marine environment characteristic parameters and the pre-trained machine learning model, and to perform nonlinear correction on the linear theoretical phase velocity to obtain the corrected internal wave phase velocity.

6. A multi-channel internal wave flow early warning system according to claim 1, characterized in that, The multi-channel communication module (2) includes: a signal strength monitoring unit, a channel switching execution unit, and a redundant transmission control unit; The signal strength monitoring unit is used to detect the signal strength of the terrestrial mobile communication link in real time; The channel switching execution unit is connected to the signal strength monitoring unit and is used to automatically activate the satellite communication link when the ground mobile communication signal strength is lower than a preset threshold. The redundant transmission control unit is connected to the main control unit (101) and is used to control the alarm information to be transmitted in parallel through at least two different communication links.

7. A warning measurement method for a multi-channel internal wave current warning system according to any one of claims 1-6, characterized in that, Includes the following steps: Step S1: Control the acoustic Doppler current profiler in the integrated ocean profile detection module (4) to collect the horizontal velocity components of multiple ocean profiles. and vertical velocity component Simultaneously, the CTD array in the ocean profile integrated detection module (4) is controlled to synchronously collect the temperature of each depth layer. and salinity data; Step S2: The data fusion unit in the ocean profile integrated detection module (4) aligns the velocity data with the temperature and salinity data in the time and spatial domains and performs joint inversion; Step S3: The physical model module in the main control unit (101) solves the linear wave characteristic equation of the internal wave based on the density profile and buoyancy frequency data obtained by inversion, and obtains the linear theoretical phase velocity; Step S4: The machine learning residual correction module in the main control unit (101) receives real-time marine environmental characteristic parameters, outputs the phase velocity residual prediction value based on the pre-trained machine learning model, and performs nonlinear correction on the linear theoretical phase velocity to obtain the corrected internal wave phase velocity; at the same time, the main control unit (101) calculates the warning time, shear flow intensity and displacement amplitude A based on the corrected internal wave phase velocity, and determines the alarm level according to the shear flow intensity and displacement amplitude A. Step S5: The main control unit (101) sends the alarm information, which includes the alarm level and the warning time, through the ground mobile communication link in the multi-channel communication module (2) to send the complete alarm information data packet; at the same time, it sends the encoded data containing the alarm level and the predicted arrival time to the shore station receiving system (5) through the satellite communication link in the multi-channel communication module (2). Step S6: The shore station receiving system (5) uses the newly collected measured flow velocity data, temperature and salinity data and corresponding internal wave event records as incremental training samples, and periodically updates and trains the machine learning model in the machine learning residual correction module.

8. The early warning measurement method for a multi-channel internal wave current early warning system according to claim 7, characterized in that, In step S2, the data fusion unit performs joint inversion, specifically including the following steps: Step S2-1: Based on temperature and salinity The density of seawater was calculated using the UNESCO International Equation of State for Seawater, along with the pressure P at the corresponding depth. for: ； Step S2-2: Based on the seawater density Calculate the floating frequency for: ； Where g is the acceleration due to gravity, and z is the depth coordinate, with vertical upward being defined as positive; Step S2-3: Combine the vertical shear rate of the flow velocity measured by the acoustic Doppler current profiler with the seawater density. The vertical mode structure and horizontal wavenumber of the internal waves are inverted using the Taylor-Goldstein equation.

9. The early warning measurement method for a multi-channel internal wave current early warning system according to claim 7, characterized in that, In step S3, the physical model module solves the characteristic equation of the internal wave linear wave, specifically as follows: Floating frequency Substituting into the Taylor-Goldstein equation: ； Where c is the internal wave phase velocity, For horizontal wavenumber, The vertical velocity eigenfunction; The eigenvalue problem is solved by combining preset boundary conditions, and the linear theoretical phase velocity of each mode is extracted.

10. The early warning measurement method for a multi-channel internal wave current early warning system according to claim 7, characterized in that, Step S4 specifically includes the following sub-steps: Step S4-1: Input the feature vector composed of the currently measured velocity profile, density profile, tidal phase, and background flow field into the pre-trained machine learning model. The machine learning model is a long short-term memory network model or a random forest model, and outputs the predicted phase velocity residual value. ; Step S4-2: Calculate the corrected internal wave phase velocity : ; Step S4-3: Based on the corrected internal wave phase velocity And the known distance L between the current observation point and the protected target, calculate the warning time. for: ； in, This is the preset system processing delay time; Step S4-4: Based on the vertical velocity eigenfunction obtained from the inversion... and horizontal wavenumber Combined with floating frequency and the corrected internal wave phase velocity The eigenfunction of the horizontal flow velocity is obtained through the linear internal wave polarization relation. and displacement eigenfunctions of isodense surfaces ; Step S4-5: Estimate the current internal wave amplitude using the internal wave amplitude projection method. The predicted maximum shear flow is: ； in, This represents the current internal wave amplitude; For depth coordinates Maximum value operation; eigenfunctions of horizontal flow velocity The modulus of the derivative with respect to depth; It is the horizontal velocity distribution of the internal wave vertical mode, and its vertical gradient represents the velocity shear eigenfunction; The predicted displacement amplitude of the isodensity surface; The displacement amplitude is then: ； in, Let be the magnitude of the displacement eigenfunction of the isodense surface; Step S4-6: Based on the predicted maximum shear flow Based on the predicted displacement amplitude A of the isodense surface, the alarm level is determined by performing fuzzy logic judgment, namely: when or At that time, it was determined to be a Level 1 concern; when or At that time, it was determined to be a Level 2 warning; when or At that time, it was determined to be a Level III emergency.

Images