Seabed turbidity current multi-parameter comprehensive monitoring method and system

Abstract

The invention discloses a seabed turbidity current multi-parameter comprehensive monitoring method and system, and relates to the technical field of ocean observation. According to the method, a synergistic interaction architecture composed of an equipment layer, an intelligent algorithm layer and a digital twinborn layer is constructed. The system performs multi-parameter data acquisition by deploying an equipment cluster, and realizes working state scheduling of equipment, online calibration and multi-source fusion of data, and inversion and evolution prediction of a turbidity current state based on an intelligent algorithm. The master device is dynamically awakened through an intelligent trigger mechanism, and the contradiction between low event capture rate and high energy consumption is solved; single-point observation is improved to real-time inversion and prediction of a three-dimensional space structure of the turbidity current through a multi-source data assimilation and digital twinning technology; and finally, starting graded early warning and equipment linkage based on the turbidity current intensity index to form a monitoring-early warning-response closed loop.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ocean observation, in particular to a submarine turbidity current multi-parameter comprehensive monitoring method and system. BACKGROUND

[0002] Submarine turbidity current is an important geological force that shapes submarine topography and transports sediments, and also poses a serious threat to key infrastructure such as submarine cables and pipelines. Due to its suddenness, high frequency and strong destructiveness, effective monitoring of turbidity currents has been a major challenge in the field of ocean science and engineering.

[0003] In the prior art, a submarine turbidity current multi-parameter comprehensive monitoring device is disclosed in Chinese patent CN109142172B. This device integrates cement pile hole pressure monitoring, turbidity optical monitoring, floating ball flow rate monitoring, and sediment sampling functions, enabling simultaneous observation of multiple parameters of turbidity current at a single fixed point, avoiding the problem of inconsistent observation and sampling positions caused by instrument movement with the ship. However, this technical solution has several inherent defects that limit its monitoring effectiveness and application value. First, this device is essentially a static, passive single-point observation system, with its observation range limited to the location of the device itself, unable to perceive and invert the three-dimensional structure and lateral evolution of turbidity current in space, resulting in great uncertainty in assessing the scale, path, and impact of turbidity current. Second, its triggering and sampling mechanism relies on a pre-set, fixed flow rate threshold to pull the floating ball and start sampling. This mechanical triggering method has low sensitivity and poor flexibility, cannot adaptively optimize for turbidity events of different intensities and properties, and is prone to miss weak initial-stage turbidity events or events with flow rates below the pre-set threshold. Finally, this device focuses on in-situ data collection and storage, lacks real-time intelligent processing, multi-source information fusion, and deep mining capabilities, and its data processing and analysis heavily rely on post-recovery, making it unable to provide effective early warning time and decision support for submarine engineering disasters. SUMMARY

[0004] (I) Technical problems solved To address the shortcomings of the prior art, the present application provides a submarine turbidity current multi-parameter comprehensive monitoring method and system, which solves the problems raised in the background art.

[0005] (II) Technical solutions To achieve the above objectives, the present application is implemented through the following technical solutions: a submarine turbidity current multi-parameter comprehensive monitoring method and system, comprising the following steps: S1, construct a monitoring architecture including a device layer, an intelligent algorithm layer and a digital twin layer; wherein the device layer is used for collecting data and executing control instructions, the intelligent algorithm layer is used for processing data of the device layer and generating device scheduling instructions and early warning information, and the digital twin layer receives data of the intelligent algorithm layer to update a model and feeds back a model prediction result to the intelligent algorithm layer; S2, multi-parameter data collection is performed by a device cluster deployed at a seabed monitoring point, and working state scheduling of the device, online calibration and multi-source fusion of data, and inversion and evolution prediction of the turbidity current state are realized based on the intelligent algorithm layer; The working state scheduling of the device includes: continuously monitoring by a low-power trigger sensor, and when the monitoring data meet preset turbidity current triggering conditions, awakening a high-power main device group in a standby state to perform high-frequency sampling.

[0006] The method further includes a system self-organizing deployment phase preceding S1: after the device cluster is deployed, a monitoring network is formed by self-organizing through underwater acoustic communication between nodes, and based on a preset seabed digital elevation model, a spatial correlation relationship and a trigger logic topology between the trigger sensor and the main device group are dynamically optimized to ensure full coverage monitoring of the turbidity current path.

[0007] Preferably, the online calibration and multi-source fusion of data include: Periodically or when a turbidity jump is detected, an actual water sample is collected by controlling a pump suction water sampler, and a sediment image is photographed by controlling a benthic camera; Based on a reference suspended matter concentration obtained by analyzing the actual water sample, a calibration curve of the turbidity sensor is updated; The concentration calculated by the ADCP backscattering intensity and the concentration measured by the turbidimeter are compared with the reference suspended matter concentration respectively; when the deviation of any one exceeds a preset threshold, a re-calibration process of the device is triggered.

[0008] Preferably, the inversion and evolution prediction of the turbidity current state include: An LSTM model is trained by historical multi-parameter data collected by the monitoring system to generate a background field; Real-time observation data of a fixed lander, an autonomous underwater vehicle and a seismograph are assimilated into the background field by a set Kalman filtering algorithm; The thickness, sediment flux and turbulent dissipation rate of the turbidity current are inversed in real time by using the assimilated model, and an evolution process of the turbidity current to a key area downstream is predicted; The set Kalman filtering generates not less than 50 set members, each member has different erosion coefficients and Manning coefficients; and the identification error of the turbidity current thickness is not more than ±3m.

[0009] Preferably, the method further comprises a hierarchical early warning step: calculating the turbidity current intensity index , which is obtained by accumulating the product of the flow velocity and the concentration along the water depth direction, i.e. , wherein and are the concentration and flow velocity of the i-th layer, is the layer thickness; according to the value of the turbidity current intensity index , different levels of early warning are issued and corresponding equipment linkage responses are triggered, specifically: when < 100, it is a blue early warning, only data is recorded; when 100 ≤ < 500, it is a yellow early warning, 1-hour summary data is sent to the shore base; when ≥ 500, it is a red early warning, satellite real-time transmission is started and the operator is notified to start engineering protection measures, and the early warning lead time is 3 to 5 hours.

[0010] Preferably, the equipment cluster adopts a three-level deployment architecture, including: a warning layer deployed upstream, composed of intelligent buoys with satellite communication capability; a main measurement layer deployed in the core area, composed of seabed landers and autonomous underwater vehicles carrying multi-parameter sensors; and a verification layer deployed downstream, composed of pressure gauges and delayed data recovery devices, the warning layer is laid at least 5 kilometers upstream of the core area, and the autonomous underwater vehicle can automatically sail and conduct spatial encryption observation along the preset section after receiving the trigger signal.

[0011] Preferably, the data online calibration and multi-source fusion step further comprises dynamic data quality evaluation and weighted fusion: a data quality factor is calculated for each data source in real time, which is determined based on the recent health status of the sensor, the signal signal-to-noise ratio, and the consistency with the readings of adjacent sensors; when data assimilation is performed in the digital twin layer, a weighted fusion algorithm based on the data quality factor is adopted, wherein the weight of the data quality factor ; When the value of a certain data source continuously falls below a preset threshold, the intelligent algorithm layer marks this source as unusable and triggers the redundant sensors in the equipment cluster to perform replacement or start the aforementioned online calibration process, the value range of the data quality factor is 0 to 1, and when the value is less than 0.5, it is determined that the data source is unusable.

[0012] A seabed turbidity current multi-parameter comprehensive monitoring system applied to the seabed turbidity current multi-parameter comprehensive monitoring method, characterized in that it comprises: a device cluster comprising intelligent buoys, seabed landers, autonomous underwater vehicles, pressure gauges and data recovery devices arranged at different positions on the seabed; an edge computing node integrated in the seabed lander or autonomous underwater vehicle, in communication connection with the device cluster, configured to perform processing and scheduling functions of the intelligent algorithm layer; a digital twin platform deployed on a cloud or a shore-based server, connected to the edge computing node through a communication module, configured to perform data assimilation, inversion and prediction functions; wherein the devices in the device cluster are coordinated according to the instructions of the edge computing node or the digital twin platform; The system further comprises a solar water surface buoy for wireless charging of the seabed lander, and the communication module supports a store-carry-forward mechanism for unloading data by the autonomous underwater vehicle which periodically rises to the surface.

[0013] Preferably, the edge computing node is further configured to perform dynamic resource scheduling based on task and energy state: real-time monitoring of the remaining energy E_remaining and the computing load L_current of each device in the device cluster; receiving turbidity evolution prediction information fed back by the digital twin layer to estimate the monitoring task load L_forecast in a future period of time T; based on the remaining energy and the predicted load, dynamically adjusting the working mode of the device, including: when E_remaining / L_forecast is lower than a first threshold value, reducing the sampling frequency of non-critical devices or entering sleep mode; when E_remaining / L_forecast is lower than a second lower threshold value, the high-energy-consumption task originally to be performed by a single device is disassembled and distributedly scheduled to multiple devices with better energy state in the cluster for collaborative completion, the first threshold value is an energy load ratio corresponding to the endurance time set by the system initially, and the second threshold value is 60% of the first threshold value.

[0014] (Three) beneficial effects The present application provides a seabed turbidity current multi-parameter comprehensive monitoring method and system, which has the following beneficial effects: 1. The present application realizes the technical leap from static single-point observation to dynamic spatial perception. Through the cooperation of multi-node device clusters and digital twin platforms, discrete observation data is real-time assimilated and inverted into the three-dimensional spatial structure of turbidity currents, breaking through the technical bottleneck of low spatial resolution of traditional monitoring devices and the inability to reconstruct the three-dimensional motion form of turbidity currents, and extending the early warning time to 3-5 hours.

[0015] 2. The present application solves the problems of low event capture rate and long-term data drift of traditional monitoring through intelligent triggering-device linkage and online self-calibration mechanism. The system dynamically wakes up the device through algorithm, increases the event capture rate from about 35% to more than 90%, and combines water sample calibration and image verification to control the turbidity measurement error in the whole life cycle within ±5%, ensuring the accuracy and reliability of the data.

[0016] 3. The present application constructs a monitoring-inversion-early warning-response closed-loop system, which directly converts monitoring data into decision instructions. Based on the real-time calculation of turbidity current intensity index, the hierarchical early warning and device linkage are started, which upgrades the system value from scientific research support to active prevention of engineering disasters, effectively safeguards the safety of submarine infrastructure, and avoids major economic losses. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The present application is a system schematic diagram. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0019] Embodiment: As shown in the drawings, Figure 1 The present application provides a submarine turbidity current multi-parameter comprehensive monitoring method system, and the device cluster is arranged in an active upstream area of a submarine canyon.

[0020] After the system is deployed, it first enters the self-organizing deployment phase. The submarine lander (main measurement layer) deployed in the core area establishes communication links with the intelligent buoy (early warning layer) 5 kilometers upstream and the pressure gauge array (verification layer) 3 kilometers downstream through the underwater acoustic communication module. Each node shares its GPS / Beidou positioning information, and dynamically confirms the relative positions of each other based on the high-precision digital elevation model of the area pre-installed in the submarine lander. The intelligent algorithm layer optimizes the triggering logic topology accordingly, and sets the highest confidence when both the upstream buoy and the core lander detect abnormal signals.

[0021] The system enters normal monitoring mode. The low-power trigger sensors (differential pressure gauge and turbidity gauge with 1W power consumption) deployed on the seafloor lander continuously monitor at 1Hz frequency. One day, a turbidity current front arrives due to a distant storm. The trigger sensors detect that the rate of pressure change dP / dt reaches 0.5kPa / min within 1 minute, while the turbidity jumps over 15. This signal meets the preset "and" trigger condition. The intelligent algorithm layer immediately sends a wake-up signal to the high-power main devices (total power consumption about 50W) such as ADCP, CTD in standby state through the internal communication bus. The main devices complete initialization within 0.5 seconds and enter a high-frequency sampling mode of 10Hz. At the same time, the command is sent to the AUV in sleep state through underwater acoustic communication, and the AUV immediately sets sail, scans back and forth along the preset section perpendicular to the turbidity current direction, and performs spatial encryption observation.

[0022] In the turbidity current event, the system detects that there is a turbidity jump layer with dNTU / dz>7NTU / m at 10 meters above the seafloor. The intelligent algorithm layer immediately starts the online calibration process: controls the pump suction water sampler to collect 5 liters of water sample at the turbidity peak depth and seals it; at the same time, instructs the benthic camera to continuously shoot the bottom bed, and the image shows that a large amount of medium and fine sand (estimated D50 about 0.25mm) is resuspended. After that, the AUV is lifted to the water surface, and the water sample analysis result (laboratory measured TSM concentration is 450mg / L) is transmitted back to the digital twin platform through satellite link. The platform fits this reference concentration with the NTU value recorded by the OBS turbidity gauge during the event, and automatically updates the calibration curve from C=k_old·NTU+b_old to C=k_new·NTU+b_new. At the same time, during the entire event, the volume concentration Cv calculated by the ADCP backscattering intensity and the concentration measured by the OBS turbidity gauge are both compared with the final reference concentration, and the deviation is less than 15%, which does not trigger the recalibration.

[0023] The digital twin platform receives the multi-source data transmitted by the edge computing node after preprocessing. The platform calls the background field generated by the LSTM model trained using the data of the system in the past year, and starts the ensemble Kalman filter algorithm. This algorithm generates 50 ensemble members, each with different bed erosion coefficients and Manning coefficients. By assimilating the vertical flow profile provided by the fixed lander, the three-dimensional spatial concentration data provided by the AUV, and the low-frequency vibration signals recorded by the seismometer, the model quickly converges, and real-time inversion of the main body thickness of the turbidity current is 85 meters (consistent with the actual measurement value of 87 meters by the AUV later, with an error of <3m), and it is predicted that it will reach the location of the submarine pipeline downstream after 3.5 hours.

[0024] Based on this, the edge computing node calculates the turbidity current intensity index every second During the peak period of the event, The value reached 580. According to the hierarchical warning rule, the system immediately issued a red warning. The warning information was transmitted in real time to the shore-based control center and the oil company operation platform through the satellite communication module integrated on the intelligent buoy. After receiving the red warning, the operator had sufficient time to start the emergency plan, close the pipeline valve and start the protection program, successfully avoiding the possible pipeline blockage or damage accident.

[0025] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications, changes, variations, substitutions, and equivalents will occur to those of ordinary skill in the art without departing from the spirit and scope of the application as defined by the following claims and their equivalents.

Claims

1. A method for multi-parameter comprehensive monitoring of submarine turbidity currents, characterized in that, The method comprises the following steps: S1, constructing a monitoring architecture comprising a device layer, an intelligent algorithm layer, and a digital twin layer; wherein the device layer is used for collecting data and executing control instructions, the intelligent algorithm layer is used for processing the data of the device layer and generating device scheduling instructions and early warning information, and the digital twin layer receives the data of the intelligent algorithm layer to update the model and feeds back the model prediction result to the intelligent algorithm layer; S2, collecting multi-parameter data by a device cluster deployed at a seabed monitoring point, and realizing device working state scheduling, online calibration and multi-source fusion of data, and inversion and evolution prediction of turbidity current state based on the intelligent algorithm layer; The device working state scheduling comprises: continuously monitoring by a low-power trigger sensor, and when the monitoring data meet preset turbidity current triggering conditions, waking up a high-power main device group in a standby state to perform high-frequency sampling.

2. The method according to claim 1, characterized in that, The online calibration and multi-source fusion of data comprise: Periodically or when detecting a turbidity jump, controlling a pump suction water sample collector to collect actual water samples, and controlling a benthic camera to take sediment images; Based on a reference suspended matter concentration obtained by analyzing the actual water samples, updating a calibration curve of the turbidity sensor; Comparing the concentration calculated by the ADCP backscattering intensity and the concentration measured by the turbidimeter with the reference suspended matter concentration respectively; when the deviation of any one exceeds a preset threshold, triggering a re-calibration process of the device.

3. The method according to claim 1, characterized in that, The inversion and evolution prediction of turbidity current state comprise: Training an LSTM model using historical multi-parameter data collected by the monitoring system to generate a background field; Using the ensemble Kalman filter algorithm, assimilating real-time observation data of fixed landers, autonomous underwater vehicles and seismographs into the background field; Using the assimilated model, inversing the thickness of the turbidity current, the sediment flux and the turbulent dissipation rate in real time, and predicting the evolution process to the key areas downstream.

4. The method according to claim 1, characterized in that: The method further comprises a hierarchical early warning step: Calculating turbidity current intensity index which is obtained by accumulating the product of the flow velocity square and the concentration in the water depth direction, i.e. ， wherein and is the concentration and flow rate of the i-th layer, is the layer thickness; According to the value of the turbidity current intensity index Different levels of early warning are issued and corresponding equipment linkage responses are triggered.

5. The method according to claim 1, characterized in that, The device cluster adopts a three-level deployment architecture, comprising: a warning layer deployed upstream, composed of intelligent buoys with satellite communication capability; a main measurement layer deployed in the core area, composed of seabed landers and autonomous underwater vehicles carrying multi-parameter sensors; and a verification layer deployed downstream, composed of pressure gauges and delayed data recovery devices.

6. The method according to claim 2 or 3, characterized in that, The online calibration and multi-source fusion of the data also comprises a dynamic evaluation of data quality and weighted fusion: a data quality factor is calculated in real time for each data source , which is determined jointly based on the recent health status of the sensor, the signal signal-to-noise ratio and the consistency with the readings of neighboring sensors; when data assimilation is performed at the digital twin layer, a weighted fusion algorithm based on the data quality factor is used, wherein the weight of the data quality factor ; When the values of a certain data source continuously fall below a preset threshold, the intelligent algorithm layer flags this source as unavailable and triggers a redundant sensor in the device cluster to take over or initiates the aforementioned online calibration process.

7. A turbidity current multi-parameter comprehensive monitoring system applied to the turbidity current multi-parameter comprehensive monitoring method of any one of claims 1-6, characterized in that, Comprise: A device cluster comprising intelligent buoys, seabed landers, autonomous underwater vehicles, pressure gauges and data recovery devices arranged at different positions on the seabed; An edge computing node integrated in the seabed lander or autonomous underwater vehicle, in communication connection with the device cluster, configured to perform processing and scheduling functions of the intelligent algorithm layer; A digital twin platform deployed on a cloud or a shore-based server, connected to the edge computing node through a communication module, configured to perform data assimilation, inversion and prediction functions; Wherein, the devices in the device cluster are coordinated according to the instructions of the edge computing node or the digital twin platform.

8. The method according to claim 1, characterized in that, The edge computing node is further configured to perform dynamic resource scheduling based on task and energy state: Real-time monitoring of the remaining energy E_remaining and the current computing load L_current of each device in the device cluster; Receiving the turbidity evolution prediction information fed back by the digital twin layer to estimate the monitoring task load L_forecast in a future period of time T; Based on the remaining energy and the predicted load, dynamically adjusting the working mode of the device, including: When E_remaining / L_forecast is lower than a first threshold, reducing the sampling frequency of non-critical devices or entering sleep; When E_remaining / L_forecast is lower than a lower second threshold, splitting and distributing a high-energy-consumption task originally to be executed by a single device to multiple devices with better energy states in the cluster for collaborative completion.

Images