A method for monitoring the spread of seabed mining plumes based on acoustic-optical-electrical technology

Abstract

The present invention provides a method for a seabed mining plume diffusion monitoring device based on acoustic-optical-electrical technology. By combining multiple monitoring technologies such as acoustics, optics and electricity, multi-dimensional and high-precision monitoring of the deep-sea mining plume diffusion process is achieved. Specific objectives include: Multi-source data fusion: Effectively integrate acoustic, optical and electrical data to improve the accuracy and comprehensiveness of plume diffusion monitoring. Improved concentration conversion accuracy: Through multi-source data calibration, high-precision conversion of acoustic data to particle concentration data is achieved. Real-time monitoring capability: Provide real-time monitoring and data analysis capabilities to promptly reflect the dynamic changes of plume diffusion. Environmental impact assessment: Provide reliable data support for deep-sea mining environmental impact assessment and promote sustainable development.

Description

Technical Field

[0001] The present invention relates to the field of marine engineering and environmental monitoring technology, and in particular to a method for monitoring the diffusion of seabed mining plumes based on acoustic-optical-electrical technology. Background Art

[0002] With the development of deep-sea mining technology, seabed mining activities are becoming increasingly frequent, and the resulting environmental impacts are attracting considerable attention. Existing technologies primarily use a single monitoring method to monitor mining plumes. For example, patent document CN117491218 B discloses a method for monitoring the spread of deep-sea mining plumes using an acoustic Doppler current profiler (ADCP).

[0003] Existing technologies typically use single monitoring devices, such as the Acoustic Doppler Current Profiler (ADCP), which monitors the spread of a plume by emitting sound waves and receiving echo intensity, while also providing data on the velocity and direction of the water flow. Turbidimeters, which measure the concentration of suspended particles in water, typically provide single-point measurements. Electrical probes, which identify the sediment-seawater interface by monitoring the natural potential difference between the two. These devices operate independently, providing either hydrodynamic or particle concentration data, but a monitoring system that integrates data from multiple sources is lacking.

[0004] Existing technologies for monitoring the spread of deep-sea mining plumes have the following shortcomings:

[0005] In terms of data acquisition: Although ADCP can simultaneously obtain data on acoustic echo intensity and water velocity and direction, it is easily interfered by noise in complex marine environments, resulting in an inability to accurately identify concentration changes at the plume boundary layer. Electrical probes can effectively identify concentration changes at the plume boundary layer, but their accuracy is inferior to that of acoustic equipment, and they are mainly used to identify the interface between sediment and seawater. Turbidimeters are usually single-point measurements, which makes it difficult to cover the wide area of ​​plume diffusion, resulting in insufficient spatial distribution information. Traditional sampling methods require frequent manual intervention and equipment maintenance, consume a lot of manpower and material resources, and cannot achieve real-time monitoring.

[0006] In terms of data accuracy: In high-noise environments, the echo intensity of the ADCP acoustic signal may be disturbed, affecting the accurate identification of the plume boundary. Although electrical probes can monitor potential differences, their low sensitivity makes it difficult to provide high-precision particle concentration data.

[0007] In terms of data interpretation and processing: existing technologies lack a mechanism to effectively integrate multi-source data (acoustic, optical, and electrical), resulting in the inability to verify and supplement the data between various monitoring methods, affecting the accuracy and reliability of the overall monitoring results.

[0008] In summary, the limited deployment locations and monitoring range of a single device make it difficult to fully monitor the three-dimensional spread of a plume, especially in the boundary layer and long-range diffusion regions. Existing technologies have significant shortcomings in data acquisition diversity, monitoring accuracy, data fusion, and real-time performance. There is an urgent need for a plume dispersion monitoring method and device that can integrate multi-source data, improve monitoring accuracy, and achieve real-time monitoring. Summary of the Invention

[0009] In order to make up for the shortcomings of the existing technology, the present invention provides a method for a seabed mining plume diffusion monitoring device based on acoustic-optical-electrical technology, which realizes multi-dimensional and high-precision monitoring of the deep-sea mining plume diffusion process by combining multiple monitoring technologies such as acoustics, optics and electricity. Specific goals include: Multi-source data fusion: Effectively integrate acoustic, optical and electrical data to improve the accuracy and comprehensiveness of plume diffusion monitoring. Improved concentration conversion accuracy: Through multi-source data calibration, high-precision conversion of acoustic data to particle concentration data is achieved. Real-time monitoring capability: Provide real-time monitoring and data analysis capabilities to promptly reflect the dynamic changes of plume diffusion. Environmental impact assessment: Provide reliable data support for deep-sea mining environmental impact assessment and promote sustainable development.

[0010] The present invention is achieved through the following technical solutions: a method for a seabed mining plume diffusion monitoring device based on acoustic-optical-electrical technology, the seabed mining plume diffusion monitoring device comprising an acoustic monitoring module, an acoustic monitoring module, and an optical monitoring module;

[0011] The acoustic monitoring module includes a downward-looking acoustic Doppler current profiler ADCP, a downward-looking ADCP connecting rod, and an acoustic transducer transmitting and receiving head, wherein the acoustic transducer transmitting and receiving head is installed at the end of the acoustic Doppler current profiler ADCP;

[0012] The acoustic monitoring module includes a device control terminal, a natural potential probe and a conical head connected from top to bottom. The acoustic Doppler current profiler ADCP is installed on the device control terminal through the downward-looking ADCP connecting rod. An electrode ring is set every 2 cm on the natural potential probe; the device control terminal 2-1 is used to set the working status of the device and locally store monitoring data to ensure that data is not lost. It can be connected to an external power supply.

[0013] The optical monitoring module includes an optical turbidity meter and a double clamp. The optical turbidity meter is installed on the natural potential probe rod through the double clamp.

[0014] The specific steps include:

[0015] Step S1: Deployment of monitoring equipment

[0016] Deploy seabed mining plume spread monitoring devices at predetermined locations on the seabed to ensure that each device covers areas where the plume may spread;

[0017] Step S2: Data acquisition

[0018] The Acoustic Doppler Current Profiler (ADCP) continuously transmits sound waves and receives echo intensity, recording the acoustic signals of plume diffusion and water flow data;

[0019] Optical turbidity meters measure the concentration of suspended particles in water in real time and record changes in particle concentration;

[0020] The natural potential probe regularly records the natural potential difference between sediment and seawater to monitor changes in sediment distribution;

[0021] Step S3: Data preprocessing

[0022] Filter and denoise the collected acoustic, optical and electrical data;

[0023] Step S4: Multi-source data fusion

[0024] A multi-source data fusion model was established to fuse the data from ADCP, optical turbidimeter, and spontaneous potential probe to improve the accuracy of monitoring. The specific steps include:

[0025] Step S4.1: Data synchronization and alignment

[0026] Align the data from the ADCP, optical turbidimeter, and spontaneous potential probe in time and space to ensure that data at the same time point corresponds to the same spatial location;

[0027] Step S4.2: Feature extraction

[0028] Extract key features from each source data:

[0029] ADCP data: acoustic wave echo intensity 𝐴, flow velocity 𝑉, flow direction 𝜃;

[0030] Electrical probe data: potential difference 𝐸;

[0031] Step S4.3: Construct training dataset

[0032] The training dataset is constructed using the laboratory-calibrated turbidity meter data as the label 𝐶 and the ADCP and electrical probe data as the input features 𝐴,𝑉,𝜃,𝐸;

[0033] Specifically expressed as:

[0034]

[0035] Where: Xi = [Ai, Vi, θi, Ei] is the input feature vector of the i-th sample; Ci is the particle concentration label of the i-th sample; N is the number of samples;

[0036] Step S4.4: Selecting a machine learning algorithm

[0037] Select the random forest algorithm to build a model for converting acoustic data to concentration data;

[0038] Step S4.5: Model training and optimization

[0039] Use the training dataset 𝐷 constructed in step S4.3 to train the random forest model;

[0040] Step S4.6: Model Validation

[0041] An independent validation dataset is used to validate the trained random forest model and evaluate its prediction performance under different conditions; validation indicators include mean square error (MSE) and coefficient of determination (𝑅) 2 ;

[0042] Step S4.7: Concentration Conversion Formula

[0043] Based on the trained random forest model, a conversion relationship from multi-source monitoring data to particle concentration is established; the specific formula is: C=f(A,V,θ,E)

[0044] Where 𝐶 is the predicted particle concentration vector, 𝐴 is the ADCP acoustic echo intensity data, 𝑉 is the water flow velocity data, 𝜃 is the water flow direction data, and 𝐸 is the electrical probe potential difference data;

[0045] Random forest prediction process:

[0046] Input feature vector: X=[A,V,θ,E]

[0047] For each decision tree 𝑡 in the random forest, its prediction result is:

[0048] C t =DecisionTreet t (X)

[0049] The final particle concentration prediction value 𝐶 is the average of all decision tree prediction results:

[0050]

[0051] Where 𝑇 is the number of decision trees in the random forest;

[0052] Step S4.8: Application of multi-source data fusion model

[0053] The trained random forest model is applied to the monitoring data, and the particle concentration of the current plume is predicted by inputting the data of the current ADCP and natural potential probe.

[0054] As a preferred solution, the double clamp includes an optical turbidity meter clamp and a natural potential probe rod clamp. The optical turbidity meter clamp is fixedly installed on the outer wall of the natural potential probe rod clamp. The optical turbidity meter clamp and the natural potential probe rod clamp are parallel to each other. The optical turbidity meter is installed in the optical turbidity meter clamp, and the natural potential probe rod clamp is sleeved on the natural potential probe rod.

[0055] As a preferred solution, step S4.5 specifically includes the following steps:

[0056] Step S4.5.1 Decision Tree Construction: Randomly sample multiple subsamples from the training dataset with replacement, i.e., Bootstrap sampling. Build a decision tree on each subsample, and randomly select some features to select the optimal split point when splitting a node.

[0057] Step S4.5.2 Model parameter optimization: Determine the parameters of the random forest, including the number of trees 𝑇, the maximum depth of each tree 𝐷, and the maximum number of features considered at each split 𝐹; optimize these parameters through cross-validation and select the parameter combination that optimizes model performance.

[0058] Due to the adoption of the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0059] Multi-source data fusion: Combining acoustic, optical and electrical monitoring methods to achieve comprehensive monitoring of plume diffusion, overcoming the limitations of single-device monitoring.

[0060] High-precision concentration conversion: Through machine learning models, ADCP acoustic data is accurately converted into particle concentration data, significantly improving monitoring accuracy.

[0061] Real-time monitoring capability: The system can collect, process and transmit data in real time, promptly reflecting the dynamic changes of plume diffusion, and is suitable for rapid response environmental monitoring needs.

[0062] Data verification and reliability improvement: Utilize data from turbidimeters and electrical probes to calibrate and verify ADCP data, ensuring the reliability and accuracy of monitoring results.

[0063] Broad application prospects: It is not only suitable for monitoring deep-sea mining plumes, but can also be extended to monitor the diffusion of suspended particles in other marine engineering environments, and has broad market potential.

[0064] Reduce manpower and material costs: Compared with traditional sampling methods, this device realizes automation and real-time monitoring, reduces manual intervention and equipment maintenance, and reduces manpower and material costs.

[0065] Additional aspects and advantages of the invention will become apparent from the description which follows, or may be learned by practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which:

[0067] Figure 1 Schematic diagram of the three-dimensional structure of the device of the present invention;

[0068] Figure 2 It is a schematic diagram of the top view of the structure of the present invention;

[0069] Figure 3 It is a bottom view structural schematic diagram of the present invention;

[0070] Figure 4 It is a schematic diagram of the three-dimensional structure of the upper part of the device of the present invention;

[0071] Figure 5 It is a schematic diagram of the three-dimensional structure of the middle part of the device of the present invention;

[0072] Figure 6 This is a time series diagram of vertical sediment diffusion in the polymetallic manganese nodule mining area in the western Pacific Ocean derived from acoustic inversion;

[0073] Figure 7 The concentration change of suspended particles at a distance of 0.74m from the bottom was obtained by an electrical probe. DETAILED DESCRIPTION

[0074] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0075] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0076] The following combination Figures 1 to 7 The method of the seabed mining plume diffusion monitoring device based on the acoustic-optical-electrical technology according to the embodiment of the present invention is described in detail.

[0077] like Figure 1As shown, the present invention proposes a method for monitoring the spread of seabed mining plumes based on acoustic-optical-electrical technology, wherein the seabed mining plume diffusion monitoring device includes an acoustic monitoring module, an acoustic monitoring module, and an optical monitoring module;

[0078] The acoustic monitoring module includes a downward-looking acoustic Doppler current profiler ADCP1-1, a downward-looking ADCP connecting rod 1-2, and an acoustic transducer transmitting and receiving head 1-3. The acoustic transducer transmitting and receiving head 1-3 is installed at the end of the acoustic Doppler current profiler ADCP1-1; the transducer transmitting and receiving head is used to transmit sound waves and receive echo intensity, monitor the diffusion process of the plume in real time, and obtain water flow velocity and direction data at the same time.

[0079] The acoustic monitoring module includes a device control terminal 2-1, a natural potential probe rod 2-2 and a conical head 2-3 connected from top to bottom. The acoustic Doppler current profiler ADCP1-1 is installed on the device control terminal 2-1 through the downward-looking ADCP connecting rod 1-2. An electrode ring is set every 2 cm on the natural potential probe rod 2-2 to monitor the natural potential difference between seawater and sediment and identify the interface between sediment and seawater; the device control terminal 2-1 is used to set the working status of the device and locally store monitoring data to ensure data is not lost. It can be connected to an external power supply.

[0080] The optical monitoring module includes an optical turbidity meter 3-1 and a double clamp. The optical turbidity meter 3-1 is installed on the natural potential probe rod 2-2 through the double clamp. The optical turbidity meter 3-1 and the acoustic Doppler current profiler ADCP1-1 are deployed synchronously to measure the concentration of suspended particles in the water body as standard data; the double clamp includes an optical turbidity meter clamp 3-2 and a natural potential probe rod clamp 3-3. The optical turbidity meter clamp 3-2 is fixedly installed on the outer wall of the natural potential probe rod clamp 3-3. The optical turbidity meter clamp 3-2 and the natural potential probe rod clamp 3-3 are parallel to each other. The optical turbidity meter 3-1 is installed in the optical turbidity meter clamp 3-2, and the natural potential probe rod clamp 3-3 is mounted on the natural potential probe rod 2-2.

[0081] The specific steps include:

[0082] Step S1: Deployment of monitoring equipment

[0083] according to Figures 1 to 5 As shown, the seabed mining plume diffusion monitoring devices are deployed on the seabed according to predetermined positions to ensure that each device can cover the area where the plume may spread;

[0084] Step S2: Data acquisition

[0085] The acoustic Doppler current profiler ADCP1-1 continuously transmits sound waves and receives echo intensity, recording the acoustic signals of plume diffusion and water flow data;

[0086] Optical turbidity meter 3-1 measures the concentration of suspended particles in water in real time and records the changes in particle concentration;

[0087] The natural potential probe 2-2 regularly records the natural potential difference between sediment and seawater to monitor changes in sediment distribution;

[0088] Step S3: Data preprocessing

[0089] Filter and denoise the collected acoustic, optical and electrical data to ensure data quality;

[0090] Step S4: Multi-source data fusion

[0091] A multi-source data fusion model is established to fuse the data from ADCP1-1, optical turbidity meter 3-1, and natural potential probe 2-2 to improve the accuracy of monitoring. Specifically, the following steps are included:

[0092] Step S4.1: Data synchronization and alignment

[0093] The data from ADCP1-1, optical turbidimeter 3-1 and natural potential probe 2-2 are synchronized in time and space to ensure that the data at the same time point corresponds to the same spatial position;

[0094] Step S4.2: Feature extraction

[0095] Extract key features from each source data:

[0096] ADCP1-1 data: acoustic echo intensity 𝐴, flow velocity 𝑉, flow direction 𝜃;

[0097] Electrical probe data: potential difference 𝐸;

[0098] Step S4.3: Construct training dataset

[0099] The training dataset is constructed using the laboratory-calibrated turbidity meter data as the label 𝐶 and the ADCP and electrical probe data as the input features 𝐴,𝑉,𝜃,𝐸;

[0100] Specifically expressed as:

[0101]

[0102] Where: Xi = [Ai, Vi, θi, Ei] is the input feature vector of the i-th sample; Ci is the particle concentration label of the i-th sample; N is the number of samples;

[0103] Step S4.4: Selecting a machine learning algorithm

[0104] The random forest algorithm was selected to establish a conversion model from acoustic data to concentration data. By constructing multiple decision trees and integrating their prediction results, the random forest algorithm can effectively process high-dimensional data and improve the accuracy and robustness of predictions.

[0105] Step S4.5: Model training and optimization

[0106] The random forest model is trained using the training dataset 𝐷 constructed in step S4.3. Specifically, the following steps are included:

[0107] Step S4.5.1 Decision Tree Construction: Randomly sample multiple subsamples from the training dataset with replacement, i.e., Bootstrap sampling. Build a decision tree on each subsample. When splitting nodes, randomly select some features to select the optimal split point to increase the diversity of the model.

[0108] Step S4.5.2 Model parameter optimization: Determine the parameters of the random forest, including the number of trees 𝑇, the maximum depth of each tree 𝐷, and the maximum number of features considered at each split 𝐹; optimize these parameters through cross-validation and select the parameter combination that optimizes model performance.

[0109] Step S4.6: Model Validation

[0110] An independent validation dataset is used to validate the trained random forest model and evaluate its prediction performance under different conditions; validation indicators include mean square error (MSE) and coefficient of determination (𝑅) 2 To ensure the generalization ability and prediction accuracy of the model;

[0111] Step S4.7: Concentration Conversion Formula

[0112] Based on the trained random forest model, a conversion relationship from multi-source monitoring data to particle concentration is established; the specific formula is: C=f(A,V,θ,E)

[0113] Where 𝐶 is the predicted particle concentration vector, 𝐴 is the ADCP acoustic echo intensity data, 𝑉 is the water flow velocity data, 𝜃 is the water flow direction data, and 𝐸 is the electrical probe potential difference data;

[0114] Random forest prediction process:

[0115] Input feature vector: X=[A,V,θ,E]

[0116] For each decision tree 𝑡 in the random forest, its prediction result is:

[0117] C t =DecisionTreet t (X)

[0118] The final particle concentration prediction value 𝐶 is the average of all decision tree prediction results:

[0119]

[0120] Where 𝑇 is the number of decision trees in the random forest. Through the above method, the random forest model can comprehensively consider the characteristics of multi-source data and improve the accuracy and reliability of particle concentration prediction.

[0121] Step S4.8: Application of multi-source data fusion model

[0122] The trained random forest model is applied to the monitoring data. By inputting the data of the current ADCP (1-1) and the natural potential probe (2-2), the particle concentration of the current plume is predicted, achieving high-precision real-time monitoring.

[0123] Acoustic signal data acquired by acoustic nodes installed at different heights were inverted to generate a time series diagram of vertical sediment diffusion. An external force disturbed the seafloor at 09:00 and ceased at 09:01. It can be seen that after the disturbance ended, the suspended sediment reached a height of 5 meters. Over time, the sediment gradually settled, and after about 5 minutes, most of the sediment was redeposited on the seabed.

[0124] In the description of the present invention, the term "plurality" refers to two or more than two. Unless otherwise expressly defined, the orientations or positional relationships indicated by the terms "upper" and "lower" are based on the orientations or positional relationships shown in the accompanying drawings. They are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or components referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they should not be understood as limitations on the present invention. The terms "connect," "install," and "fix" should be understood in a broad sense. For example, "connection" can mean a fixed connection, a detachable connection, or an integral connection; it can be a direct connection or an indirect connection through an intermediate medium. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0125] Throughout this specification, terms such as "one embodiment," "some embodiments," and "specific embodiments" mean that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0126] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for monitoring the spread of seabed mining plumes based on acoustic-optical-electrical technology, characterized in that , the seabed mining plume diffusion monitoring device includes an acoustic monitoring module, an acoustic monitoring module, and an optical monitoring module; The acoustic monitoring module comprises a downward-looking acoustic Doppler current profiler ADCP (1-1), a downward-looking ADCP connecting rod (1-2) and an acoustic transducer transmitting and receiving head (1-3), wherein the acoustic transducer transmitting and receiving head (1-3) is installed at the end of the acoustic Doppler current profiler ADCP (1-1); The acoustic monitoring module comprises a device control terminal (2-1), a natural potential probe (2-2) and a conical head (2-3) connected from top to bottom, an acoustic Doppler current profiler ADCP (1-1) is mounted on the device control terminal (2-1) via a downward-looking ADCP connecting rod (1-2), and an electrode ring is provided on the natural potential probe (2-2) every 2 centimeters; the device control terminal 2-1 is used to set the working state of the device and locally store monitoring data to ensure that data is not lost, and can be connected to an external power supply; The optical monitoring module comprises an optical turbidity meter (3-1) and a double clamp, and the optical turbidity meter (3-1) is installed on the natural potential probe rod (2-2) via the double clamp; The specific steps include: Step S1: Deployment of monitoring equipment Deploy seabed mining plume spread monitoring devices at predetermined locations on the seabed to ensure that each device covers areas where the plume may spread; Step S2: Data acquisition The acoustic Doppler current profiler ADCP (1-1) continuously transmits sound waves and receives echo intensity, recording the acoustic signals of plume diffusion and water flow data; Optical turbidity meter (3-1) measures the concentration of suspended particles in water in real time and records the changes in particle concentration; The natural potential probe (2-2) regularly records the natural potential difference between sediment and seawater and monitors changes in sediment distribution; Step S3: Data preprocessing Filter and denoise the collected acoustic, optical and electrical data; Step S4: Multi-source data fusion A multi-source data fusion model was established to fuse the data from ADCP (1-1), optical turbidity meter (3-1) and natural potential probe (2-2) to improve the accuracy of monitoring. The specific steps include: Step S4.1: Data synchronization and alignment The data from ADCP (1-1), optical turbidimeter (3-1) and natural potential probe (2-2) are synchronized in time and space to ensure that the data at the same time point correspond to the same spatial position; Step S4.2: Feature extraction Extract key features from each source data: ADCP (1-1) data: acoustic echo intensity 𝐴, flow velocity 𝑉, flow direction 𝜃; Electrical probe data: potential difference 𝐸; Step S4.3: Construct training dataset The training dataset is constructed using the laboratory-calibrated turbidity meter data as the label 𝐶 and the ADCP and electrical probe data as the input features 𝐴,𝑉,𝜃,𝐸; Specifically expressed as: Where: Xi = [Ai, Vi, θi, Ei] is the input feature vector of the i-th sample; Ci is the particle concentration label of the i-th sample; N is the number of samples; Step S4.4: Selecting a machine learning algorithm Select the random forest algorithm to build a model for converting acoustic data to concentration data; Step S4.5: Model training and optimization Use the training dataset 𝐷 constructed in step S4.3 to train the random forest model; Step S4.6: Model Validation An independent validation dataset is used to validate the trained random forest model and evaluate its prediction performance under different conditions; validation indicators include mean square error (MSE) and coefficient of determination (𝑅) 2 ; Step S4.7: Concentration Conversion Formula Based on the trained random forest model, a conversion relationship from multi-source monitoring data to particle concentration is established; the specific formula is: C=f(A,V,θ,E) Where 𝐶 is the predicted particle concentration vector, 𝐴 is the ADCP acoustic echo intensity data, 𝑉 is the water flow velocity data, 𝜃 is the water flow direction data, and 𝐸 is the electrical probe potential difference data; Random forest prediction process: Input feature vector: X=[A,V,θ,E] For each decision tree 𝑡 in the random forest, its prediction result is: C t =DecisionTreet t (X) The final particle concentration prediction value 𝐶 is the average of all decision tree prediction results: Where 𝑇 is the number of decision trees in the random forest; Step S4.8: Application of multi-source data fusion model The trained random forest model is applied to the monitoring data, and the particle concentration of the current plume is predicted by inputting the data of the current ADCP (1-1) and the natural potential probe (2-2).

2. The method of the device for monitoring the spread of seabed mining plumes based on acoustic-optical-electrical technology according to claim 1 is characterized in that The double clamp includes an optical turbidity meter clamp (3-2) and a natural potential probe rod clamp (3-3), the optical turbidity meter clamp (3-2) is fixedly installed on the outer wall of the natural potential probe rod clamp (3-3), the optical turbidity meter clamp (3-2) and the natural potential probe rod clamp (3-3) are parallel to each other, the optical turbidity meter (3-1) is installed in the optical turbidity meter clamp (3-2), and the natural potential probe rod clamp (3-3) is sleeved on the natural potential probe rod (2-2).

3. The method of the device for monitoring the spread of seabed mining plumes based on acoustic-optical-electrical technology according to claim 1 is characterized in that , the step S4.5 specifically includes the following steps: Step S4.5.1 Decision Tree Construction: Randomly sample multiple subsamples from the training dataset with replacement, i.e., Bootstrap sampling. Build a decision tree on each subsample, and randomly select some features to select the optimal split point when splitting a node. Step S4.5.2 Model parameter optimization: Determine the parameters of the random forest, including the number of trees 𝑇, the maximum depth of each tree 𝐷, and the maximum number of features considered at each split 𝐹; optimize these parameters through cross-validation and select the parameter combination that optimizes model performance.

Images