In-situ suspended sediment concentration and flux prediction system based on conductivity data

Abstract

The application discloses a system for predicting in-situ suspended sediment concentration and flux based on conductivity data, and relates to the technical field of in-situ monitoring of marine environment, comprising a data processing and visualization platform, which is in communication connection with the following modules, wherein: a data acquisition module is used for extracting multi-dimensional water body data collected in an in-situ environment from a database. The application couples the corrected suspended sediment concentration data and the flow velocity data to generate a continuous vertical concentration-flow velocity matrix, and combines a flux prediction model established in combination with historical data to analyze the nonlinear relationship between the sediment transport flux and the flow velocity-concentration, adopts a random forest algorithm, takes the flow velocity, the conductivity interval depth and the corrected suspended sediment concentration as input variables, and outputs the sediment flux prediction value, so that the static and dynamic comprehensive prediction of the sediment flux at different moments and different positions is realized, and a quantitative analysis tool is provided for marine sediment dynamic research and mineral environment assessment.

Description

Technical Field

[0001] This invention relates to the field of in-situ monitoring technology of marine environment, specifically to an in-situ suspended sediment concentration and flux prediction system based on conductivity data. Background Technology

[0002] In-situ monitoring of marine suspended sediment concentration and flux is one of the core technologies for assessing the environmental effects of deep-sea mineral resources. It is of great significance for studying sediment transport dynamics and the environmental impact of resource extraction. Traditional field observation methods, such as multi-channel water samplers, have significant limitations: the sampling process disturbs the seabed environment and cannot obtain accurate data at different depths near the bottom in real time; reliance on filtration and weighing leads to large errors in flux calculation; in addition, acoustic and optical instruments (such as turbidimeters and suspended sediment profilers) are easily affected by sediment concentration, particle size and current thickness, have limited range, and are difficult to cope with sediment transport monitoring under extreme conditions such as high concentrations in rivers flowing into the sea or storm surges.

[0003] To address the shortcomings of existing technologies, this invention proposes an in-situ suspended sediment concentration and flux prediction system based on conductivity data. By using a mobile towed cable chain for monitoring, monitoring data of the study area is acquired, and the relationship between conductivity and volumetric sediment concentration is analyzed to establish a flux prediction model. The relationship between sediment transport flux and flow velocity-sediment concentration is analyzed to achieve comprehensive prediction of both static and dynamic in-situ suspended sediment concentration and flux, providing reliable data support for marine sedimentary dynamics research and mineral environment assessment. Summary of the Invention

[0004] The purpose of this invention is to provide an in-situ suspended sediment concentration and flux prediction system based on conductivity data, so as to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0006] An in-situ suspended sediment concentration and flux prediction system based on conductivity data includes a data processing and visualization platform. The data processing and visualization platform is communicatively connected to the following modules:

[0007] The data acquisition module is used to extract multi-dimensional water body data collected in the in-situ environment from the database and preprocess it to form a multi-dimensional dataset. The multi-dimensional water body data includes water conductivity, flow velocity and water depth data.

[0008] The data quality control module is used to receive pre-processed multi-dimensional water body data, perform quality control and Archie's Law inversion, correct suspended sediment concentration data and flow velocity data, and eliminate environmental interference.

[0009] The spatial reconstruction and flux prediction engine is used to couple the corrected suspended sediment concentration data and velocity data to generate a continuous vertical suspended sediment concentration-velocity matrix. Combined with a pre-established flux prediction model, it performs single-point prediction of sediment flux to obtain sediment flux at different times and locations.

[0010] The cross-sectional flux prediction module is used to integrate the single-point prediction results of all sediment fluxes and predict the total sediment flux of each layer in the monitoring section through integration.

[0011] The anomaly early warning module is used to compare the predicted results of the total sediment flux of each layer in the monitoring section with the pre-set sediment flux threshold to determine whether there are abnormal events in the monitoring area, and to provide risk management support for deep-sea mining and waterway maintenance.

[0012] A further improvement to the technical solution of this invention lies in the following: the process of forming a multi-dimensional dataset in the data acquisition module is as follows:

[0013] Multi-dimensional water body data collected in the in-situ monitoring environment are retrieved and extracted from the preset database, including water body conductivity, flow velocity and water depth data. The water body conductivity data needs to be pre-compensated for temperature and salinity.

[0014] The extracted multi-dimensional water body data is classified and sorted, and initially organized according to timestamps and spatial location parameters to ensure data integrity and continuity. The multi-dimensional water body data is then cleaned to remove outliers or erroneous records, and missing data is imputed. Finally, after standardization, a multi-dimensional dataset with a unified format is formed.

[0015] A further improvement of the technical solution of the present invention is that the data quality control module includes a conductivity-concentration conversion module and a turbulence disturbance correction module;

[0016] The conductivity-concentration conversion module converts water conductivity data into suspended sediment concentration based on Archie's Law, introduces adjustment factors to eliminate environmental interference, and outputs a high-precision suspended sediment concentration profile, covering a range from low to high concentration.

[0017] The turbulence disturbance correction module is used to smooth turbulence noise through high-frequency sampling, analyze the relationship between drag speed and suspension amount by combining the pre-conducted bed disturbance experiment, subtract the secondary suspension background value, correct the suspended sediment concentration, and reduce the measurement deviation caused by uneven mixing and drag disturbance.

[0018] A further improvement to the technical solution of this invention lies in the following: the process of outputting the suspended sediment concentration profile in the conductivity-concentration conversion module is as follows:

[0019] Historical water conductivity data were pre-analyzed, quality checks were performed, outliers were removed, and fitting analysis was conducted on sediment volume concentration and the conductivity of sediment-seawater mixture and seawater conductivity at different temperatures. The historical water conductivity data were collected and stored over a long period of time by conductivity sensors before the system was deployed, and included water conductivity measurement data under different field collection and experimental measurement environments.

[0020] Based on the fitting analysis results, adjustment factors were introduced, and Archie's law was applied to compare the conductivity of sediment-seawater mixtures with that of pure seawater. The water conductivity data were converted into suspended sediment concentration, covering a range from low to high concentration.

[0021] By integrating suspended sediment concentration data from various depth layers, a vertical suspended sediment concentration profile is generated. The inversion accuracy is evaluated by comparing the profile with laboratory calibration data, and a standardized concentration dataset is output.

[0022] A further improvement to the technical solution of this invention lies in the following: the process of correcting the suspended sediment concentration in the turbulence disturbance correction module is as follows:

[0023] For the raw conductivity and flow velocity data obtained by high-frequency sampling of 1-50Hz, the moving average is used to eliminate short-term turbulent fluctuations, retain the true signal of sediment transport, ensure data smoothness, and avoid dynamic response distortion caused by excessive filtering.

[0024] Based on the prior bed disturbance experiment, the concentration of additional suspended sediment generated by bed disturbance under different towing speeds was measured experimentally. The functional relationship between towing speed and background concentration was fitted, and a quantitative relationship model between towing speed and secondary suspended concentration was established to quantify the background contribution of towing disturbance to the measured values.

[0025] By combining real-time flow velocity measurements, the background value of secondary suspension under the current operating conditions is calculated using a quantitative relationship model between drag velocity and secondary suspension concentration. This background value is then subtracted from the original measured concentration to correct measurement deviations caused by uneven mixing or drag disturbances, and a high-precision flow velocity-concentration joint dataset corrected for turbulence disturbances is output.

[0026] A further improvement of the technical solution of the present invention is that the spatial reconstruction and flux prediction engine includes a vertical concentration field reconstruction unit and a single-point flux prediction unit.

[0027] The vertical concentration field reconstruction unit is used to perform spatial interpolation on sparse multidimensional water body data using the Kriging interpolation method, and to generate a continuous vertical suspended sediment concentration-velocity matrix by combining it with the ADCP velocity field, so as to solve the problem of insufficient upper layer resolution and capture the concentration abrupt change layer.

[0028] The single-point flux prediction unit is used to combine a multi-dimensional dataset with a flux prediction model pre-established using historical data to analyze the relationship between sediment transport flux and flow velocity-sediment concentration, thereby predicting sediment flux at different times and locations.

[0029] A further improvement to the technical solution of this invention lies in the following: the process of generating a continuous vertical suspended sediment concentration-velocity matrix in the vertical concentration field reconstruction unit is as follows:

[0030] We analyzed sparse, multi-dimensional water body data, extracted discrete layer concentration data and synchronous ADCP velocity profiles, constructed a three-dimensional spatial dataset, and used variogram analysis to determine the semivariogram model parameters of Kriging interpolation to characterize the continuity of concentration variation with depth.

[0031] The configured kriging interpolation model is used to make the optimal unbiased estimate of the concentration of unmonitored layers. The ADCP velocity field is introduced as a covariate in the interpolation process. The vertical resolution is enhanced by co-kriging to ensure accurate capture of concentration abrupt change layers. A continuous vertical concentration profile matrix is ​​generated based on the interpolation results.

[0032] Cross-validation of the interpolation results is performed, the root mean square error is calculated to assess the reconstruction accuracy, outlier interpolation points exceeding the confidence interval are removed, and a standardized vertical suspended sediment concentration-velocity matrix is ​​output to solve the problem of insufficient resolution caused by the sparsity of upper-layer data.

[0033] A further improvement to the technical solution of this invention lies in the following: the process of predicting sediment flux at different times and locations in the single-point flux prediction unit is as follows:

[0034] The flow velocity and water depth data of the target points are extracted from the multi-dimensional dataset, the interval depth of the conductivity sensor is determined, and the corrected suspended sediment concentration data is extracted, aligned by time series, and integrated to form a standardized input dataset.

[0035] Historical data on sediment flux, including interval depth data from flow velocity and conductivity sensors, suspended sediment concentration data, and sediment flux, were integrated and divided into training and testing sets. Based on the random forest model, the training set was input into the random forest model to train the flux prediction model, optimize the model parameters, and the trained model was tested using the testing set to obtain the completed flux prediction model.

[0036] Based on the pre-trained flux prediction model, the input dataset is used as the model input of the flux prediction model. The relationship between sediment transport flux and flow velocity-sediment concentration is analyzed to perform sediment flux prediction at a single point and output the sediment flux at different times and locations.

[0037] A further improvement to the technical solution of this invention lies in the following: In the cross-sectional flux prediction module, the prediction process for the total sediment flux of each layer of the monitoring section is as follows:

[0038] Collect the sediment flux prediction results of each individual point, check the data integrity and accuracy, and according to the stratification information of the monitoring section, map the prediction results of each individual point to the corresponding stratum. Use the integration method to calculate the total sediment flux of each stratum based on the individual point prediction, and obtain the stratified flux data.

[0039] The sediment fluxes calculated from each stratum are summarized to obtain the total sediment flux of each stratum in the monitoring section. The prediction results are then output in a specified format and presented on a data processing and visualization platform.

[0040] A further improvement to the technical solution of the present invention is that the process of determining whether there is an abnormal event in the monitoring area in the anomaly early warning module is as follows:

[0041] By loading a pre-set sediment flux threshold, the total flux prediction results of each layer in the monitoring section are compared with the corresponding threshold in real time to identify potential abnormal layers that exceed the threshold. The sediment flux threshold is a safe range set based on environmental standards.

[0042] Based on the comparison results, the layers exceeding the limit are identified, it is determined whether there are regional abnormal events, and the abnormalities are classified according to the extent of exceeding the limit, and the risk levels are clarified as mild risk level, moderate risk level and severe risk level.

[0043] The system automatically triggers early warning signals for confirmed abnormal events, generates reports containing abnormal layers, flux deviations, and recommended measures, and pushes them to the data processing and visualization platform to provide real-time risk management basis for decisions on deep-sea mining operations adjustments or waterway maintenance.

[0044] Due to the adoption of the above technical solution, the technical progress achieved by this invention compared to the prior art is as follows:

[0045] This invention provides an in-situ suspended sediment concentration and flux prediction system based on conductivity data. By coupling corrected suspended sediment concentration data with flow velocity data, a continuous vertical concentration-flow velocity matrix is ​​generated. Combined with a flux prediction model established using historical data, the nonlinear relationship between sediment transport flux and flow velocity-concentration is analyzed. Using a random forest algorithm, with flow velocity, conductivity interval depth, and corrected suspended sediment concentration as input variables, the system outputs predicted sediment flux values. This achieves comprehensive static and dynamic prediction of sediment flux at different times and locations, providing a quantitative analysis tool for marine sedimentary dynamics research and mineral environment assessment.

[0046] This invention provides an in-situ suspended sediment concentration and flux prediction system based on conductivity data. By using a single-site flux prediction unit and combining a multi-dimensional dataset with a flux prediction model, it reveals the complex influence of the coupling effect of flow velocity and suspended sediment concentration on flux. At the same time, the contribution of concentration to flux also exhibits nonlinear characteristics with the change of flow velocity. The nonlinear relationship is automatically captured by a random forest model, and the flux at a single site is calculated time-by-time, providing theoretical support for understanding sediment transport mechanisms.

[0047] This invention provides an in-situ suspended sediment concentration and flux prediction system based on conductivity data. Through a cable-chain-deployed sensor network, it achieves accurate acquisition of multi-dimensional data such as water conductivity, flow velocity, and water depth. At the same time, it adopts high-frequency sampling combined with moving average filtering technology to effectively smooth turbulence noise and retain the true signal of sediment transport. It uses Archie's law to convert conductivity data into suspended sediment concentration and introduces secondary disturbance of the bottom bed to correct for environmental interference, outputting a high-precision vertical concentration profile. The concentration measurement covers the bottom bed to the sea surface, and the range covers low to high concentration range, significantly improving the temporal resolution and measurement accuracy of the data. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0049] Figure 1 This is a schematic diagram of the workflow of the in-situ suspended sediment concentration and flux prediction system based on conductivity data according to the present invention.

[0050] Figure 2 This is a correlation curve between electrical conductivity and suspended sediment concentration at a specific salinity level according to the present invention.

[0051] Figure 3 This is a correlation curve of conductivity versus temperature at a specific salinity level according to the present invention.

[0052] Figure 4 This is a correlation curve of conductivity and salinity at a specific concentration according to the present invention.

[0053] Figure 5 This is a graph showing the fitting data of sediment volume concentration and the conductivity of sediment-seawater mixture and seawater at different temperatures according to the present invention;

[0054] Figure 6 This is a schematic diagram illustrating the data acquisition and processing of the in-situ suspended sediment concentration and flux prediction system based on conductivity data according to the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1, such as Figure 1 , Figure 6 As shown, this invention provides an in-situ suspended sediment concentration and flux prediction system based on conductivity data, including a data processing and visualization platform. The data processing and visualization platform is communicatively connected to the following modules, wherein:

[0057] The data processing and visualization platform, serving as the system's user interface, displays monitoring data, analysis results, and early warning information, facilitating user interaction and data viewing. Presented in a graphical user interface (GUI), it integrates data display, analysis result visualization, and early warning information push functions. It receives data processed by various modules, including vertical concentration profiles, velocity matrices, flux predictions, and anomaly warning reports, and performs multi-dimensional visualization. The platform supports real-time refresh and historical data playback. It also provides a parameter configuration interface, allowing users to set flux thresholds, adjust display ranges, and export data reports. Early warning information is pushed in real-time using color coding and pop-up windows, assisting users in quickly identifying abnormal events and making informed decisions.

[0058] The data acquisition module is used to extract multi-dimensional water body data collected in the in-situ environment from the database and preprocess it to form a multi-dimensional dataset. This multi-dimensional water body data includes water conductivity, flow velocity, and water depth data. The module retrieves and extracts multi-dimensional water body data collected in the in-situ monitoring environment from a pre-set database, including water conductivity, flow velocity, and water depth data. The water conductivity data requires pre-compensation for temperature and salinity. The water conductivity, flow velocity, and water depth data are obtained using a cable-chain-deployed conductivity sensor (CD), current meter (ADCP / ADP), and ultrasonic sensor. The depth sounder collects, classifies and sorts the extracted multi-dimensional water body data, and performs preliminary organization based on timestamps and spatial location parameters to ensure data integrity and continuity. It also cleans the multi-dimensional water body data to remove outliers or erroneous records, imputes missing data, and then performs standardization to form a multi-dimensional dataset in a unified format. In addition, the data acquisition module is presented as an integrated hardware and software entity, including conductivity sensors, current meters and ultrasonic depth sounders deployed on underwater cable chains, as well as supporting data acquisition and communication units.

[0059] The expression for temperature compensation calculation is as follows:

[0060] ;

[0061] In the formula: It is the electrical conductivity (mS / cm) at a certain temperature. Conductivity at standard temperature, where T is the temperature value. It is the correction temperature under standard temperature. The temperature compensation coefficient is calculated using a conductivity meter for automatic temperature compensation. A corrected temperature compensation coefficient is selected for seawater with a salinity of 34-35‰ at a standard temperature (25℃). Different temperatures are set, and the temperature compensation coefficient is calculated using a numerical fitting method. For every 1℃ change in temperature, the conductivity increases by approximately 1.8%-2.3%.

[0062] The expression for salinity compensation calculation is as follows:

[0063] ;

[0064] In the formula: It is the electrical conductivity (mS / cm) at a certain salinity. Conductivity at standard salinity, where S is the salinity value. It is the corrected salinity under standard salinity. The salinity compensation coefficient is calculated using a conductivity meter for automatic salinity compensation. A corrected salinity compensation coefficient is selected for seawater with a salinity of 34-35‰ at a standard temperature (25℃). Different salinities are set, and the salinity compensation coefficient is calculated using a numerical fitting method. On average, for every 5‰ decrease in salinity, the conductivity decreases by approximately 2.6%-3.2%.

[0065] The data acquisition module's specific tasks are as follows: It retrieves and extracts multi-dimensional water body data collected in the in-situ monitoring environment from a pre-set database, including water conductivity, flow velocity, and water depth data. This data is acquired in real-time using a cable-chain-deployed conductivity sensors (CD), current meters (ADCP / ADP), and ultrasonic depth sounders. The conductivity sensors are deployed at different depth intervals on the cable chain, arranged exponentially from bottom to top. Due to the high concentration of suspended sediment near the bottom, a denser deployment at the bottom and a sparser deployment at the top is possible, and the positions can be adjusted according to actual site requirements. The conductivity sensors monitor water conductivity and water depth data in real-time. The upper layer is equipped with current meters (ADCP or ADP) to monitor flow velocity, and the bottom layer is equipped with a weighted object with an ultrasonic depth sounder to ensure it is in contact with the bottom in real-time. The weight is a lead-ball type design with a cross-shaped opening in the center. In the intersecting space, ultrasonic sensors and conductivity sensors are vertically deployed at the bottom of the spherical structure, connected to a cable chain at the top. Horizontally, conductivity sensors are deployed at both ends for monitoring near-bottom sediments and bidirectional correction. The extracted multi-dimensional water body data is classified and sorted according to timestamps and spatial location parameters to ensure temporal continuity and spatial consistency. A time-space index structure is constructed by parsing time tags, latitude and longitude coordinates, and equipment identifiers in the multi-dimensional water body data. Simultaneously, the integrity of the data sequence is checked, missing segments due to equipment failure or communication interruption are identified, and outliers are marked. The classified data is cleaned, removing obviously abnormal or erroneous records, and missing data points are filled in using interpolation algorithms. Finally, the data is standardized to a unified format, including unit conversion, coordinate system alignment, and time synchronization processing, forming a structured multi-dimensional dataset.

[0066] In addition, in actual operation, the data acquisition module collects data in real time through a cable chain-deployed conductivity sensor, acoustic Doppler current meter, and ultrasonic depth sounder. The conductivity sensor is deployed along the cable chain in an exponential distribution pattern, with a higher sensor density near the bottom to capture high-concentration suspended sediment layers, while the upper sensors are relatively sparse. The data is automatically classified and sorted according to timestamps and spatial location parameters to construct a time-space index structure. All data undergoes unit unification, coordinate alignment, and time synchronization processing to form a structured multi-dimensional dataset, providing standardized input for the system.

[0067] The data quality control module receives preprocessed multi-dimensional water body data, performs quality control and Archie's Law inversion, corrects suspended sediment concentration and flow velocity data, and eliminates environmental interference. The data quality control module includes a conductivity-concentration conversion module and a turbulence disturbance correction module. The data quality control module exists as an algorithm-driven data processing engine, possessing two core functions: conductivity-concentration conversion and turbulence disturbance correction. It is responsible for quality control and inversion correction of the preprocessed multi-dimensional water body data, mainly comprising two sub-modules: conductivity-concentration conversion and turbulence disturbance correction. The conductivity-concentration conversion module is based on Archie's Law. The water conductivity data is converted into suspended sediment concentration. During the process, an adjustment factor is introduced to dynamically adjust the inversion formula to adapt to different sea areas and sediment types. Then, a quantitative relationship between conductivity and concentration is established through fitting analysis, and a high-precision vertical concentration profile is output. The turbulence disturbance correction module uses moving average filtering technology to smooth short-term turbulence fluctuations for high-frequency sampled raw data, retaining the real sediment transport signal. At the same time, combined with the drag velocity-background concentration function model established by the bottom disturbance experiment, the module calculates and subtracts the secondary suspended background value in real time, corrects the measurement deviation caused by uneven mixing or drag disturbance, and outputs a high-quality velocity-concentration joint dataset.

[0068] The conductivity-concentration conversion module converts water conductivity data into suspended sediment concentration based on Archie's Law. It introduces adjustment factors to eliminate environmental interference and outputs a high-precision suspended sediment concentration profile, covering a range from low to high concentrations. It pre-analyzes historical water conductivity data, performs quality checks, removes outliers, and performs fitting analyses of sediment volume concentration and sediment-seawater mixture conductivity with seawater conductivity at different temperatures. The historical water conductivity data was collected and stored long-term using conductivity sensors (CDs) before system deployment. The system includes water conductivity measurement data under different field collection and experimental conditions (different temperatures, salinity, water depths, and flow velocities, etc.). Based on the fitting analysis results, adjustment factors are introduced, and Archie's law is applied to compare the conductivity of sediment-seawater mixtures with that of pure seawater. The water conductivity data is converted into suspended sediment concentration, covering a low to high concentration range. Suspended sediment concentration data from various depth layers are integrated to generate a vertical suspended sediment concentration profile. The inversion accuracy is evaluated by comparing with laboratory calibration data, and a standardized concentration dataset is output.

[0069] ;

[0070] In the formula, The electrical conductivity of the sediment-seawater mixture, The electrical conductivity of pure seawater. The suspended sediment concentration is denoted by m, which is an adjustment factor ranging from 1.2 to 3.0.

[0071] The specific functions of the conductivity-concentration conversion module are as follows: First, it performs quality checks on historical water conductivity data, removing outliers caused by sensor drift, environmental interference, or communication anomalies. Second, it performs fitting analysis based on sediment volume concentration and sediment-seawater mixture conductivity data at different temperatures to establish a quantitative relationship between conductivity and concentration. Third, it determines adjustment factors through regression analysis to ensure adaptability to different environmental conditions. Fourth, based on the fitting analysis results, it applies Archie's law to convert water conductivity data into suspended sediment concentration, dynamically adjusting the m-value to adapt to different sea areas or sediment types, ensuring inversion accuracy across the low to high concentration range, and outputting standardized suspended sediment concentration data. Fifth, it integrates suspended sediment concentration data from various depth layers to generate a continuous vertical concentration profile. Sixth, it evaluates the inversion accuracy by comparing it with laboratory calibration data, identifies systematic errors, calculates the root mean square error and mean absolute error, and if the error exceeds a preset error threshold, it triggers a parameter re-optimization process, ultimately outputting a standardized concentration dataset.

[0072] The turbulence disturbance correction module is used to smooth turbulence noise through high-frequency sampling. Combined with pre-conducted bed disturbance experiments, it analyzes the drag velocity-suspension relationship, subtracts secondary suspension background values, corrects suspended sediment concentration, and reduces measurement bias caused by mixing inhomogeneity and drag disturbance. For raw conductivity and velocity data obtained through 1-50Hz high-frequency sampling, moving averages are used to eliminate short-term turbulence fluctuations, retaining the true signal of sediment transport and ensuring data smoothness. Simultaneously, it avoids dynamic response distortion caused by over-filtering. Based on pre-conducted bed disturbance experiments, the module... The concentration of additional suspended sediments generated by bed disturbance under different towing speeds was measured. The functional relationship between towing speed and background concentration was fitted, and a quantitative relationship model between towing speed and secondary suspended concentration was established to quantify the background contribution of towing disturbance to the measured values. Combined with real-time flow velocity measurements, the background value of secondary suspended sediment under the current operating conditions was calculated using the quantitative relationship model between towing speed and secondary suspended concentration. This background value was subtracted from the original measured concentration to correct the measurement deviation caused by uneven mixing or towing disturbance. A high-precision flow velocity-concentration joint dataset corrected for turbulence disturbance was output.

[0073] The specific functions of the turbulence disturbance correction module are as follows: Preprocessing the raw conductivity and velocity data sampled at high frequencies of 1-50Hz to eliminate the interference of short-term turbulent fluctuations on sediment transport signals; employing a moving average filter to locally average the high-frequency data sequence within a set time window; balancing signal smoothness and dynamic response requirements with the window length; uniform weighting of the filter weights to reduce the impact of boundary effects on data continuity; and dynamically adjusting the sliding window size based on the turbulent kinetic energy of the real-time flow velocity to ensure that a shorter window preserves details at low flow velocities and a longer window suppresses noise at high flow velocities, thereby outputting a smoothed conductivity and velocity time series; based on pre-conducted... Subsoil disturbance experiments were conducted to experimentally determine and fit the functional relationship between towing velocity and background concentration, and to establish a quantitative relationship model between towing velocity and secondary suspended concentration, quantifying the background contribution of instrument towing to the measured values. In the real-time measurement phase, the secondary suspended background value under the current operating conditions was calculated by combining the smoothed flow velocity data and the established quantitative relationship model. The difference between the secondary suspended concentration output by the quantitative relationship model and the suspended sediment concentration inverted from the original conductivity was calculated, and the measurement deviation caused by towing disturbance was deducted to obtain the corrected true concentration value. Finally, a flow velocity-concentration joint dataset corrected for turbulence disturbance was output, including timestamp, depth, flow velocity, original concentration, corrected concentration, and quality flag.

[0074] The expression for the functional relationship between drag speed and background density is:

[0075] ;

[0076] ;

[0077] In the formula, v is the dragging speed, covering common operation speeds, and the background density increases monotonically with the dragging speed; The background suspended sediment concentration corresponds to the drag velocity v. The increase is gradual at low speeds and significant at high speeds, which is affected by the nonlinearity of the bed shear force. The critical drag velocity is the threshold velocity at which bottom sediments transition from a steady state to a state of significant disturbance. , The proportionality coefficient was determined through experimental fitting. , ; , The power-law exponent reflects the sensitivity of concentration to velocity. It tends to be 1.2-1.8 (approximately linear in the low-speed region). It tends towards 2.0-3.0 (high-speed region nonlinear enhancement). This is the background concentration offset, ensuring the function... The state is continuous; as v approaches 0, Approaching 0 (no disturbance without drag), as v approaches the maximum operating speed. Approaching the maximum observed value in the experiment, in the low-speed region Fitting and The linear relationship, the slope is... The intercept is In the high-speed area Fitting using nonlinear least squares method , ;

[0078] The spatial reconstruction and flux prediction engine is used to couple the corrected suspended sediment concentration data and velocity data to generate a continuous vertical suspended sediment concentration-velocity matrix. Combined with a pre-established flux prediction model, it performs single-point prediction of sediment flux to obtain sediment flux at different times and locations.

[0079] The cross-sectional flux prediction module is used to integrate the single-point prediction results of all sediment fluxes and predict the total sediment flux of each layer in the monitoring section through integration.

[0080] The anomaly early warning module is used to compare the predicted results of the total sediment flux of each layer in the monitoring section with the pre-set sediment flux threshold to determine whether there are abnormal events in the monitoring area, and to provide risk management support for deep-sea mining and waterway maintenance.

[0081] It should be noted that the workflow of this in-situ suspended sediment concentration and flux prediction system based on conductivity data is as follows: The data acquisition module extracts multi-dimensional water body data collected in the in-situ monitoring environment from the database, including conductivity, flow velocity, and water depth data, and performs preprocessing to form a standardized dataset. Then, the data quality control module performs temperature and salinity compensation on the data, uses Archie's law to invert the suspended sediment concentration, and uses the turbulence disturbance correction module to subtract the secondary suspended background value, outputting a high-precision concentration profile. Then, combined with the spatial reconstruction and flux prediction engine, the Kriging interpolation method is used to generate a continuous vertical concentration-velocity matrix. Based on the random forest algorithm, a flux prediction model is built to achieve single-point sediment flux prediction. The cross-sectional flux prediction module summarizes the flux of each layer using the integration method and outputs the total cross-sectional flux. Finally, the anomaly early warning module compares the prediction results in real time based on preset thresholds, identifies abnormal events, and triggers early warnings.

[0082] Example 2, as Figure 1 As shown, based on Embodiment 1, the present invention provides a technical solution: preferably, the spatial reconstruction and flux prediction engine includes a vertical concentration field reconstruction unit and a single-point flux prediction unit;

[0083] The vertical concentration field reconstruction unit is used to spatially interpolate sparse multidimensional water body data using Kriging interpolation. Combined with ADCP velocity field, it generates a continuous vertical suspended sediment concentration-velocity matrix, addressing the insufficient resolution of the upper layers and capturing concentration abrupt change layers. It analyzes sparse multidimensional water body data, extracts discrete layer concentration data and synchronous ADCP velocity profiles, constructs a three-dimensional spatial dataset, and uses variogram analysis to determine the semivariogram model parameters of Kriging interpolation, characterizing the continuity of concentration changes with depth. Using a pre-configured Kriging interpolation model, it performs optimal unbiased estimation of concentration in unmonitored layers. The ADCP velocity field is introduced as a covariate during interpolation, and co-Kriging enhances vertical resolution, ensuring accurate capture of concentration abrupt change layers. A continuous vertical concentration profile matrix is ​​generated based on the interpolation results. The interpolation results are cross-validated, and the root mean square error is calculated to assess reconstruction accuracy. Abnormal interpolation points exceeding the confidence interval are removed, and a standardized vertical suspended sediment concentration-velocity matrix is ​​output, addressing the insufficient resolution problem caused by sparse upper-layer data.

[0084] The specific tasks of the vertical concentration field reconstruction unit are as follows: Preprocessing sparse, multi-dimensional water body data; integrating suspended sediment concentration data from discrete layers with synchronous ADCP velocity profiles; constructing a three-dimensional spatial dataset containing time, depth, concentration, and velocity; analyzing the spatial correlation of vertical concentration distribution using variograms to address its non-uniformity; quantifying the continuity of vertical concentration variation by calculating the semivariogram values ​​at different depth intervals; fitting a spherical theoretical model based on the experimental variogram to determine the semivariogram model parameters for Kriging interpolation, including nugget value, sill value, and range, describing the spatial autocorrelation range and structural strength of concentration in the vertical direction; and configuring a Kriging interpolation model based on the variogram analysis results to perform optimal unbiased estimation of concentration in unmonitored layers. The ADCP velocity field is introduced as a covariate, and the spatial distribution characteristics of velocity are incorporated into the concentration interpolation process using the co-kriging method. By establishing a cross-covariance function between concentration and velocity, the vertical coupling relationship between the two is captured, enhancing the ability to identify concentration abrupt change layers. During the interpolation process, the weighting coefficients are automatically optimized to ensure that the estimated values ​​are statistically unbiased and have the minimum variance, generating a continuous high-resolution vertical concentration profile matrix. After the interpolation is completed, the reconstruction accuracy is evaluated through cross-validation, and the root mean square error between the measured values ​​and the interpolation is calculated to quantify the reliability of the interpolation results. Abnormal interpolation points that exceed the preset confidence interval are removed or corrected. Finally, a standardized vertical suspended sediment concentration-velocity matrix is ​​output, which includes the concentration estimate, corresponding velocity, and root mean square error for each depth layer.

[0085] The single-site flux prediction unit combines a multi-dimensional dataset with a flux prediction model pre-built using historical data to analyze the relationship between sediment transport flux and velocity-sediment concentration, thereby predicting sediment flux at different times and locations. It extracts velocity and water depth data for the target site from the multi-dimensional dataset, determines the interval depth of the conductivity sensor, and extracts corrected suspended sediment concentration data. These are aligned by time series and integrated to form a standardized input dataset. The dataset integrates historical data including velocity, conductivity sensor interval depth, suspended sediment concentration data, and sediment flux, dividing it into training and testing sets. Using a random forest model as the basic architecture, the training set is input into the random forest model to train the flux prediction model, optimize model parameters, and the training model is tested using the test set to obtain the trained flux prediction model. Based on the pre-trained flux prediction model, the input dataset is used as the model input for the flux prediction model to analyze the relationship between sediment transport flux and velocity-sediment concentration, perform single-site sediment flux prediction, and output sediment flux at different times and locations.

[0086] The formulas for calculating sediment flux at different times and locations are as follows:

[0087] ;

[0088] In the formula, Let be the sediment flux at time t with a water depth of h. Let be the flow velocity at time t when the water depth is h. Let be the concentration of suspended sediment at a water depth of h at time t, and D be the interval depth of the conductivity sensor.

[0089] The specific tasks of the single-point flux prediction unit are as follows: Extract velocity and water depth data for the target point from a multi-dimensional dataset; determine the interval depth based on the vertical distribution characteristics of the conductivity sensor; simultaneously extract suspended sediment concentration data corrected for temperature and salinity to eliminate interference from environmental factors in concentration measurement; align all data according to time series; use cubic spline interpolation to unify velocity, concentration, and conductivity data to the same time resolution; remove outliers; and integrate them into a standardized input dataset containing timestamps, water depth, velocity, conductivity interval depth, and corrected concentration. Divide historical data covering velocity, conductivity interval depth, suspended sediment concentration, and sediment flux observations into training and testing sets, and use stratified sampling to ensure data distribution consistency. A flux prediction model is constructed using a random forest model as its core architecture. Input variables include flow velocity, conductivity interval depth, and corrected suspended sediment concentration. The output is sediment flux. During model training, hyperparameters are optimized through grid search, 5-fold cross-validation is used to reduce the risk of overfitting, and redundant information is removed to improve model efficiency. The prediction accuracy is evaluated using root mean square error and coefficient of determination on a test set to ensure the model's generalization ability on independent data. Finally, the trained flux prediction model is output. Based on the pre-trained flux prediction model, a standardized input dataset is used as the model input. The model automatically captures the nonlinear influence of the flow velocity-concentration coupling relationship on flux, calculates the sediment flux at a single point time step, and outputs sediment flux values ​​containing time series.

[0090] In the cross-sectional flux prediction module, the prediction process for the total sediment flux at each layer of the monitored cross section is as follows:

[0091] Collect the sediment flux prediction results of each individual point, check the data integrity and accuracy, and according to the stratification information of the monitoring section, map the prediction results of each individual point to the corresponding stratum. Use the integration method to calculate the total sediment flux of each stratum based on the prediction of the individual point, obtain the stratified flux data, summarize the sediment flux calculated by each stratum to obtain the total sediment flux of each stratum of the monitoring section, and output the prediction results in the specified format and present them on the data processing and visualization platform.

[0092] The formula for calculating the total sediment flux at each stratum of the monitoring section is as follows:

[0093] ;

[0094] In the formula, Q represents the total sediment flux at each stratum of the monitoring section. Let f be the sediment flux at a water depth of h at time t;

[0095] The specific tasks of the cross-sectional flux prediction module are as follows: Collect the sediment flux prediction results from all individual points, check the completeness and accuracy of the data to ensure there are no missing or outliers, map the prediction results of each individual point to the corresponding strata based on the vertical stratification information of the monitoring cross-section, and use linear interpolation to supplement data for strata not directly covered by the cross-section, ensuring that each stratum has a flux value. Simultaneously, record the data source and quality control indicators for each stratum. Using an integral method, calculate the total sediment flux for each stratum based on the prediction results of its corresponding individual points. After completing the flux calculation for each stratum, further summarize the results for each stratum to obtain the total sediment flux for each stratum of the monitoring cross-section. Then, organize and output the calculated prediction results according to a specified format and present them on a data processing and visualization platform.

[0096] In the anomaly warning module, the process of determining whether there are abnormal events in the monitoring area is as follows:

[0097] The system loads pre-set sediment flux thresholds and compares the total flux prediction results of each layer in the monitoring section with the corresponding threshold in real time to identify potential abnormal layers that exceed the thresholds. The sediment flux thresholds are safety ranges set based on environmental standards. The system identifies layers that exceed the limits based on the comparison results, determines whether there are regional abnormal events, and classifies the anomalies according to the extent of exceeding the limits, clarifying the risk levels as mild, moderate, and severe. For confirmed abnormal events, the system automatically triggers early warning signals, generates reports containing abnormal layers, flux deviations, and recommended measures, and pushes them to the data processing and visualization platform to provide real-time risk management basis for deep-sea mining operation adjustments or waterway maintenance decisions.

[0098] The specific functions of the anomaly early warning module are as follows: It loads a pre-set sediment flux threshold, which is determined based on the safety range of environmental standards and is divided by stratigraphic level to accommodate differences in vertical flux distribution. It receives the total flux prediction results for each stratigraphic level at the monitoring section in real time and compares them with the corresponding threshold. During the comparison, a sliding time window is used to analyze flux trends to ensure that short-term fluctuations do not falsely trigger warnings. If the flux value of a certain stratigraphic level exceeds the sediment flux threshold for three consecutive time steps, that stratigraphic level is determined to be a potential anomaly area. Simultaneously, the magnitude and duration of the exceedance are recorded. Based on the magnitude and spatial distribution of flux exceedance... The system classifies anomalous events into three risk levels: mild, moderate, and severe. Combining the vertical location of the anomalous stratum with flux changes in adjacent strata, it determines whether the anomalous event exhibits regional characteristics. After risk level determination, an anomalous event label is generated and associated with possible triggers. Upon confirmation of an anomalous event, an early warning signal is automatically triggered, generating a structured report. The report includes the anomalous stratum, flux deviation, and recommended measures. The early warning signal is pushed in real-time through a data processing and visualization platform, supporting multi-terminal access and notification to relevant personnel via SMS. Simultaneously, the anomalous event is recorded in the database.

[0099] In addition, it should be noted that this system can be applied to the following scenarios: 1. Near-shore and deep-sea mining sediment plume diffusion, enabling mobile towed cable chain monitoring of sediment transport concentration and flux, achieving three-dimensional real-time monitoring of sediment plumes in seawater; 2. Near-shore estuaries and coastal marine sediment dynamics monitoring, enabling fixed-point shipborne cable chain monitoring of sediment transport concentration and flux; 3. Sediment transport processes in estuaries, deep-sea plains, and submarine canyons, enabling fixed-point mooring in-situ monitoring of sediment transport concentration and flux in submarine channels over a time series.

[0100] For the application scenario of "monitoring the diffusion of near- and deep-sea mining sediment plumes", the monitoring method adopts a mobile towed chain system. Monitoring is carried out in the near- and deep-sea mining area and surrounding potential diffusion areas during and after mining operations. In the early stage, topographic and hydrological data of the mining area are collected. Conductivity sensors are deployed at different intervals and depths on the towed chain. The conductivity sensors are distributed exponentially from bottom to top. The upper end of the chain is fixed to the ship with a current meter (ADCP or ADP) to monitor the current velocity. The weight at the bottom of the chain is a lead ball design with a cross-shaped space in the middle. Ultrasonic sensors and conductivity sensors are vertically deployed at the bottom of the ball and connected to the chain at the top. Conductivity sensors are deployed at both ends of the chain for monitoring near-bottom sediments and bidirectional correction. The ultrasonic depth sounder at the bottom of the weight ensures that it is in contact with the bottom in real time. The monitoring scheme is designed to cover the plume source area, diffusion path and background area. The depth gradient of the chain and the speed of the ship are set to construct a three-dimensional monitoring grid. During on-site operation, the cable chain is fixed to the monitoring vessel, which then sets sail at a constant speed along a preset route. The underway ADCP transmits the velocity at different water depths in real time. Based on the conductivity data collected by the conductivity sensor, the mathematical model of the in-situ suspended sediment concentration and flux prediction system is used to calculate and transmit concentration and velocity data in real time. The vessel position, water depth, and environmental parameters are recorded simultaneously. By matching the coordinates of the three-dimensional monitoring points, the concentration is converted and the flux is calculated. After removing outliers, the three-dimensional diffusion and transport flux distribution results of the plume are generated.

[0101] For the application scenario of "sedimentary dynamic monitoring in nearshore estuaries and coastal areas", the monitoring method is based on GIS technology and water depth data. The monitoring area is delineated according to monitoring needs, and the number and location of monitoring points are determined according to deployment principles. The mooring system design includes: conductivity sensors are deployed at different depth intervals on the towing chain in an exponential distribution from bottom to top; the upper end of the chain is connected to a current meter (ADCP or ADP), fixed to the vessel to monitor current velocity; a lead sphere-shaped weight is placed at the bottom, with a cross-shaped internal structure. An ultrasonic sensor and conductivity sensor are vertically installed at the bottom, and conductivity sensors are deployed at both ends laterally for near-bottom sediment monitoring and bidirectional calibration. An ultrasonic depth sounder at the bottom of the sphere can detect the bottom contact status in real time. During deployment, a suitable vessel is selected to transport the system to the monitoring point, and the mooring system is slowly lowered after the vessel has stabilized. First, ensure the anchor is firmly anchored to the bottom. Then, release the wire rope and monitoring instrument in sequence, keeping the instrument vertical and adjusting the ADCP transducer downwards. After deployment, accurately record information such as latitude, longitude, water depth, and time. During monitoring, the in-situ suspended sediment concentration and flux prediction system based on conductivity data is used to view the equipment status and collected data in real time. The system integrates real-time data uploaded by conductivity sensors at different layers, combined with current velocity information obtained from the current meter, to calculate the suspended sediment concentration and flux at each depth in real time. Simultaneously, meteorological and hydrological data such as wind speed, wind direction, and wave height are recorded. The monitoring duration should cover the entire tidal cycle. In areas with significant seasonality, monitoring for no less than 25 hours should be carried out seasonally. During retrieval, slowly lift the wire rope and gradually return the equipment to the deck. Immediately afterward, check the appearance of the instrument and the data storage status. After confirming that everything is correct, clean and maintain it.

[0102] For applications involving sediment transport processes in river estuaries, deep-sea plains, and submarine canyons, the monitoring method involves first defining the monitoring area using GIS adjustment factors and water depth data, then determining monitoring points based on deployment principles. The mooring system includes conductivity sensors deployed at different depth intervals in an exponential, bottom-up distribution. The upper part of the system is fixed to the vessel via an ADCP / ADP current meter for monitoring current velocity; the bottom is a lead-ball-shaped weight with ultrasonic and conductivity sensors vertically installed in a cross-shaped space, and conductivity sensors deployed at both ends laterally to achieve near-bottom sediment monitoring and bidirectional correction. The ultrasonic sensor at the bottom ensures real-time bottom contact. When deploying, select a suitable vessel, and after stabilizing, slowly lower the anchor mooring to ensure the anchor is firmly on the bottom. Release the wire rope and instruments in sequence, keeping the ADCP transducer facing downwards, and record latitude, longitude, water depth, and time information. During monitoring, the equipment status and data are viewed in real time through the in-situ prediction system. Integrate conductivity and flow velocity data from different layers, calculate and store suspended sediment concentration and flux in real time, and record meteorological and hydrological parameters such as wind speed and waves simultaneously. Monitoring should cover the complete tidal cycle, and seasonal monitoring should be conducted for no less than 25 hours in seasonal areas. When recovering, use GIS and Beidou positioning to confirm the location, slowly lift the equipment, and immediately check the instruments and data after recovery, and perform cleaning and maintenance.

[0103] Example 3, as Figures 2 to 5 As shown in Examples 1-2, this invention provides a feasibility analysis of an indoor experimental method for determining conductivity to predict suspended sediment concentration and an exploration of its influencing factors, as detailed below:

[0104] 1. Experimental Materials and Methods

[0105] Sediment samples used in the experiment: Two types of sediments were used in the experiment: sediment samples were collected from the Yellow River Delta mudflats with a median particle size of 0.03 mm. The soil samples were dried in an electric heating dryer at 105 ℃, and then the soil samples were ground and passed through a 0.5 mm sieve.

[0106] To quantify the relationship between conductivity and sediment concentration, a laboratory experiment was conducted to measure the changes in conductivity of seawater and sediment mixtures under different combinations of sediment concentration and temperature conditions. The experiment was first carried out at an ambient temperature of 10 °C. Salt was dissolved in 10 L of tap water to prepare salt solutions with concentrations of 5, 10, 15, 20, 25, 30, 35, and 40‰. Sediment of different weights (dry weight) was then added to the salt solutions and vigorously stirred to create uniformly mixed sediment solutions with concentrations of 0, 25, 50, 75, 100, 125, 150, 175, 200, and 225 g / L. The incremental method for measuring the conductivity of the sediment-water mixture is as follows:

[0107] 1) Before testing, the sediment solution must be stirred evenly to ensure that the sediment and water are completely mixed;

[0108] 2) Use a conductivity (CT) sensor to measure the salinity, conductivity, and temperature of solutions with different concentrations in the container. The test starts with the sample with the lowest concentration, and the first sample has the lowest concentration and the highest conductivity value.

[0109] 3) After each sample test, the sensor must be washed with clean water. This concentration-increasing test results in the last sample having the highest sediment concentration but the lowest conductivity, thus minimizing the test error.

[0110] Pour a small amount of the brine from bucket 2 back into bucket 1 and stir vigorously until a uniform suspension is formed. At the same time, continuously measure the conductivity and temperature, and take samples to determine the sediment concentration.

[0111] 4) Collect the above sediment-water mixture samples, place the samples in a water bath, heat the samples, and repeat the first two steps until all sample temperatures of 10, 15, 20, 25, 30, and 35°C have been tested.

[0112] Considering that the effect of sediment content in seawater on the conductivity of the mixture depends on the ratio of sediment particle conductivity to seawater conductivity, and that the conductivity of sediment particles is always several orders of magnitude smaller than that of seawater, the mineralogical properties of sediments can only affect the absolute conductivity of sediment particles, and have a limited effect on the change in the conductivity of the mixture.

[0113] 2. Experimental Setup and Results

[0114] The experimental setup and results of the conductivity test of sediment soil samples mixed with water under different salinity, temperature and sediment content were summarized in Table 1 through a series of quantitative analyses.

[0115] Table 1 Experimental setup and experimental data results

[0116]

[0117] 3. Analysis and Discussion

[0118] Based on the above experimental data, a correlation curve between electrical conductivity and suspended sediment concentration at a specific salinity was plotted. Figure 2The standard curve representing the relationship between sediment content and the electrical conductivity of the mixed water body shows that as the sediment content increases, the electrical conductivity of the mixed water body exhibits a significant decreasing trend, and the two show a good linear negative correlation. The main reason for this phenomenon is that the electrical conductivity of soil particles themselves is extremely low (close to zero). When sediment is added to the water body, these particles with extremely low conductivity become an important part of the mixed system, directly causing the overall electrical conductivity of the water-soil mixed system to be significantly lower than that of the original seawater.

[0119] Analysis of the standard curves showing the relationship between sediment content and the conductivity of the mixed water body reveals that the conductivity follows a functional relationship with the sediment content:

[0120] ;

[0121] In the formula: It is the electrical conductivity of a suspension mixture at a certain temperature and salinity. It is the electrical conductivity of seawater at a certain temperature and salinity. It is the electrical conductivity of suspended particulate matter at a certain temperature and salinity. It is the volume concentration of suspended particles in the mixture, i.e., the suspended sediment concentration. It is a function of salinity S and temperature T. b and c are factor parameters;

[0122] The data relationship curve is fitted using a multivariate nonlinear regression model (high-order polynomial or neural network): =-6.297454+1.556916×S+0.014036× -0.010135×T-0.031990× -0.000009× +0.000855× -0.001136×S× +0.029718×S×T-0.000452× ×T+0.000465× +0.000000× -0.000019× .

[0123] Under constant salinity conditions Figure 3 The correlation curves of conductivity and temperature at a specific salinity are shown. The slopes of the multiple standard curves exhibit a high degree of consistency and can be considered constant values. Based on this, Kriging interpolation can be used to establish a quantitative relationship model between sediment content, temperature, and conductivity at a specific salinity. This model is useful for exploring the effect of temperature on conductivity when salinity and particle size are the same. Figure 3Conductivity data points (randomly selected) were selected at salinities of 5, 15, 25, and 35, and sediment contents of 0 g / L and 100 g / L. The results showed that, under constant salinity, temperature is also a significant factor affecting conductivity, exhibiting a monotonically increasing conductivity with increasing temperature. Further analysis of the experimental data revealed that the temperature dependence of conductivity (i.e., the rate of increase) slightly weakened with rising temperature, reflected in a gradual decrease in the slope of the curve. It is speculated that this phenomenon may be related to the effect of increasing temperature on the behavior of conductive ions in the solution.

[0124] Figure 4 The correlation curves of conductivity and salinity at specific concentrations show the relationship between sediment content and conductivity under constant temperature and different salinity conditions. The analysis shows that there is a good linear correlation between the two, that is, the conductivity of water increases approximately linearly with the increase of salinity. Based on this linear relationship and the corresponding standard curve, a quantitative prediction model of sediment content and conductivity under specific temperature and salinity combinations can be constructed using the Kriging interpolation method. This model supports the quantitative inversion of sediment content under the premise of known environmental parameters (temperature, salinity) and measured conductivity values.

[0125] Experimental data analysis shows that ( Figures 2 to 4 Temperature and salinity are controlling factors affecting the electrical conductivity of water bodies. Under constant particle size, the electrical conductivity of mixed systems at different temperatures and salinities exhibits a significant gradient change: increased salinity increases electrical conductivity. Specifically, at the same suspended sediment concentration, the higher the salinity and temperature, the greater the absolute value of the slope of the standard curve for electrical conductivity. This mechanism stems from the solution conductivity mechanism: high salinity environments enhance the spatial hindrance effect of suspended particles on ion migration by increasing ion concentration and activity. Increasing temperature also increases particle activity, thus increasing the value of ionic electrical conductivity. The study also found that: 1. The electrical conductivity differences between high-salt and high-temperature systems are significant, with relatively small measurement and calculation errors; 2. The salinity sensitivity of electrical conductivity decreases as temperature decreases. These characteristics lead to increased measurement uncertainty and calculation accuracy errors in low-salt areas. In addition, high salinity conditions may cause an increase in the resistivity of soil particle interfaces, which is related to changes in the interaction of charged particles on the particle surface.

[0126] Studies have found that very high sediment content leads to a decrease in electrical conductivity, because the electrical conductivity of sediment particles is several orders of magnitude lower than that of seawater.

[0127] 4. Conclusion

[0128] Using electrical conductivity to determine suspended sediment content is feasible and exhibits good correlation. By fitting a large amount of experimental data into a multivariate nonlinear function relationship, a formula for calculating the suspended sediment content and electrical conductivity in a given area can be derived. If the seawater salinity is constant and known (35‰) during the monitoring period using the conductivity system, the aforementioned standard formula can be used to calculate the conductivity of the surrounding seawater in the turbidity current. Then, the ratio between the measured conductivity and the conductivity of the ambient seawater can be used to calculate the conductivity using the formula... Accurately predict the concentration of suspended sediments at different depths in the marine environment.

[0129] The program developed based on the experimental data will be made into software and placed in the instrument of the electrical in-situ monitoring system currently under development to realize the on-site monitoring and forecasting of suspended sediment content during storm sea conditions or other erosion processes, thereby obtaining the suspended sediment flux of the process. This system method lays the foundation for further research on suspended sediment and the development of instruments.

[0130] In addition, the conductivity monitoring equipment has been applied to indoor water tank experiments, using conductivity methods to monitor the concentration of suspended sediments at different layers during sediment transport.

[0131] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An in-situ suspended sediment concentration and flux prediction system based on conductivity data, including a data processing and visualization platform, characterized in that, The data processing and visualization platform has the following communication modules, among which: The data acquisition module is used to extract multi-dimensional water body data collected in the in-situ environment from the database and preprocess it to form a multi-dimensional dataset. The multi-dimensional water body data includes water conductivity, flow velocity and water depth data. The data quality control module is used to receive preprocessed multi-dimensional water body data, perform quality control and Archie's Law inversion, and correct suspended sediment concentration data and flow velocity data. The spatial reconstruction and flux prediction engine is used to couple the corrected suspended sediment concentration data and velocity data to generate a continuous vertical suspended sediment concentration-velocity matrix. Combined with a pre-established flux prediction model, it performs single-point prediction of sediment flux to obtain sediment flux at different times and locations. The spatial reconstruction and flux prediction engine includes a vertical concentration field reconstruction unit and a single-point flux prediction unit. The vertical concentration field reconstruction unit is used to perform spatial interpolation on sparse multidimensional water body data using the Kriging interpolation method, and to generate a continuous vertical suspended sediment concentration-velocity matrix by combining it with the ADCP velocity field. The single-site flux prediction unit is used to combine a multi-dimensional dataset with a flux prediction model pre-established using historical data to analyze the relationship between sediment transport flux and flow velocity-sediment concentration, thereby predicting sediment flux at different times and locations. Specifically, it includes: The flow velocity and water depth data of the target points are extracted from the multi-dimensional dataset, the interval depth of the conductivity sensor is determined, and the corrected suspended sediment concentration data is extracted, aligned by time series, and integrated to form a standardized input dataset. Historical data on sediment flux, including interval depth data from flow velocity and conductivity sensors, suspended sediment concentration data, and sediment flux, were integrated and divided into training and testing sets. Based on the random forest model, the training set was input into the random forest model to train the flux prediction model, optimize the model parameters, and the trained model was tested using the testing set to obtain the completed flux prediction model. Based on the pre-trained flux prediction model, the input dataset is used as the model input of the flux prediction model. The relationship between sediment transport flux and flow velocity-sediment concentration is analyzed, and sediment flux at a single point is predicted. The sediment flux at different times and locations is output. The cross-sectional flux prediction module is used to integrate the single-point prediction results of all sediment fluxes and predict the total sediment flux of each layer in the monitoring section through integration. The anomaly warning module is used to compare the predicted results of the total sediment flux of each layer in the monitoring section with the pre-set sediment flux threshold to determine whether there are any abnormal events in the monitoring area.

2. The in-situ suspended sediment concentration and flux prediction system based on conductivity data according to claim 1, characterized in that: In the data acquisition module, the process of forming a multi-dimensional dataset is as follows: Multi-dimensional water body data collected in the in-situ monitoring environment are retrieved and extracted from the preset database, including water body conductivity, flow velocity and water depth data. The water body conductivity data needs to be pre-compensated for temperature and salinity. The extracted multi-dimensional water body data is classified and sorted, and initially organized according to timestamps and spatial location parameters. The multi-dimensional water body data is then cleaned to remove outliers or erroneous records, and missing data is imputed. Finally, after standardization, a multi-dimensional dataset with a unified format is formed.

3. The in-situ suspended sediment concentration and flux prediction system based on conductivity data according to claim 1, characterized in that: The data quality control module includes a conductivity-concentration conversion module and a turbulence disturbance correction module; The conductivity-concentration conversion module converts water conductivity data into suspended sediment concentration based on Archie's Law, and introduces an adjustment factor to output a suspended sediment concentration profile. The turbulence disturbance correction module is used to smooth turbulence noise through high-frequency sampling, analyze the relationship between towing speed and suspension amount in combination with the pre-conducted bed disturbance experiment, subtract the secondary suspension background value, and correct the suspended sediment concentration.

4. The in-situ suspended sediment concentration and flux prediction system based on conductivity data according to claim 3, characterized in that: The process of outputting the suspended sediment concentration profile in the conductivity-concentration conversion module is as follows: Historical water conductivity data were pre-analyzed, quality checks were performed, outliers were removed, and fitting analysis was conducted on sediment volume concentration and the conductivity of sediment-seawater mixture and seawater conductivity at different temperatures. The historical water conductivity data were collected and stored over a long period of time by conductivity sensors before the system was deployed, and included water conductivity measurement data under different field collection and experimental measurement environments. Based on the fitting analysis results, adjustment factors were introduced, and Archie's law was applied to compare the conductivity of sediment-seawater mixtures with that of pure seawater, converting the water conductivity data into suspended sediment concentration. By integrating suspended sediment concentration data from various depth layers, a vertical suspended sediment concentration profile is generated. The inversion accuracy is evaluated by comparing the profile with laboratory calibration data, and a standardized concentration dataset is output.

5. The in-situ suspended sediment concentration and flux prediction system based on conductivity data according to claim 3, characterized in that: In the turbulence disturbance correction module, the process of correcting the suspended sediment concentration is as follows: For the original conductivity and flow velocity data obtained by high-frequency sampling of 1-50Hz, a moving average is used to eliminate short-term turbulent fluctuations and retain the true signal of sediment transport. Based on the prior bed disturbance experiment, the concentration of additional suspended sediment generated by bed disturbance under different towing speeds was measured experimentally. The functional relationship between towing speed and background concentration was fitted, and a quantitative relationship model between towing speed and secondary suspended concentration was established to quantify the background contribution of towing disturbance to the measured values. By combining real-time flow velocity measurements, the background value of secondary suspension under the current operating conditions is calculated using a quantitative relationship model between drag velocity and secondary suspension concentration. This background value is then subtracted from the original measured concentration to correct measurement deviations caused by uneven mixing or drag disturbances, and a high-precision flow velocity-concentration joint dataset corrected for turbulence disturbances is output.

6. The in-situ suspended sediment concentration and flux prediction system based on conductivity data according to claim 1, characterized in that: In the vertical concentration field reconstruction unit, the process of generating a continuous vertical suspended sediment concentration-velocity matrix is ​​as follows: We analyzed sparse, multi-dimensional water body data, extracted discrete layer concentration data and synchronous ADCP velocity profiles, constructed a three-dimensional spatial dataset, and used variogram analysis to determine the semivariogram model parameters of Kriging interpolation to characterize the continuity of concentration variation with depth. The optimal unbiased estimate of the concentration in unmonitored layers is obtained by using a pre-configured Kriging interpolation model. The ADCP velocity field is introduced as a covariate in the interpolation process, and a continuous vertical concentration profile matrix is ​​generated based on the interpolation results. Cross-validation of the interpolation results is performed, the root mean square error is calculated to assess the reconstruction accuracy, outlier interpolation points exceeding the confidence interval are removed, and a standardized vertical suspended sediment concentration-velocity matrix is ​​output.

7. The in-situ suspended sediment concentration and flux prediction system based on conductivity data according to claim 6, characterized in that: In the cross-sectional flux prediction module, the prediction process for the total sediment flux at each layer of the monitoring section is as follows: Collect the sediment flux prediction results of each individual point, check the data integrity and accuracy, and according to the stratification information of the monitoring section, map the prediction results of each individual point to the corresponding stratum. Use the integration method to calculate the total sediment flux of each stratum based on the individual point prediction, and obtain the stratified flux data. The sediment fluxes calculated from each stratum are summarized to obtain the total sediment flux of each stratum in the monitoring section. The prediction results are then output in a specified format and presented on a data processing and visualization platform.

8. The in-situ suspended sediment concentration and flux prediction system based on conductivity data according to claim 1, characterized in that: In the anomaly early warning module, the process of determining whether there are abnormal events in the monitoring area is as follows: By loading a pre-set sediment flux threshold, the total flux prediction results of each layer in the monitoring section are compared with the corresponding threshold in real time to identify potential abnormal layers that exceed the threshold. The sediment flux threshold is a safe range set based on environmental standards. Based on the comparison results, the layers exceeding the limit are identified, it is determined whether there are regional abnormal events, and the abnormalities are classified according to the extent of exceeding the limit, and the risk levels are clarified as mild risk level, moderate risk level and severe risk level. For confirmed abnormal events, an early warning signal is automatically triggered, a report is generated including the abnormal layer, flux deviation and recommended measures, and pushed to the data processing and visualization platform.

Images