Device and method for measuring concentration distribution of fine particles in water based on multi-source information fusion

Abstract

The application relates to a water fine particle concentration distribution measuring device based on multi-source information fusion, a water sampling and turbidity measuring device is used to obtain actual measured turbidity values of water samples at the point, as a first calibration data source; a mud pan volume difference measuring device uses a 3D scanning technology to scan a surface shape of sediments, calculates a volume difference of bottom sediments before and after experiments, and serves as a second calibration data source; a mud pan mass difference measuring device measures a mass difference of bottom fine particles before and after disturbance, converts the mass difference into a suspended particle mass, and serves as a third calibration data source; a laser turbidity measuring device obtains turbidity data from a mapping relationship of light intensity data, as a fourth calibration data source; and an image processing mechanism trains an image recognition model based on a mapping relationship between space discrete point turbidity monitoring data obtained by an experimental source data acquisition mechanism and image pixel information collected by a camera device, so that fine particle soil concentration distribution changes of a target region are captured and reconstructed.

Description

Technical Field

[0001] This invention relates to a device and method for measuring the concentration distribution of fine particles in water based on multi-source information fusion, belonging to the field of water particulate matter concentration measurement technology. Background Technology

[0002] As a key approach to alleviating the supply and demand imbalance of terrestrial mineral resources, the development and utilization of deep-sea mineral resources has gradually become a focus of attention for countries worldwide. However, despite increasing emphasis on technological research and development in deep-sea mining, commercial-scale mining has yet to be achieved. Besides technological factors, the degree of disturbance to the seabed ecosystem is a crucial limiting factor distinguishing deep-sea mining from terrestrial mining. Regarding deep-sea mining activities, the United Nations Convention on the Law of the Sea (UNCLOS) outlines basic principles and specific procedures for marine environmental protection. Consequently, an increasing number of countries require mining applicants to submit Environmental Impact Assessments (EIS) and environmental management plans in their mining applications. Low environmental disturbance has become a prerequisite for deep-sea mining operations. Early environmental impact studies have shown that sediment disturbance is one of the most significant and longest-lasting environmental impact issues in deep-sea mining. Developing accurate, efficient, and reliable technologies for identifying and measuring sediment disturbance and diffusion is crucial for further research and development of high-performance, low-environmental-disturbance deep-sea mining technologies.

[0003] The shortcomings of existing technologies and their causes:

[0004] 1) Traditional methods for measuring the concentration distribution of fine particles in water typically involve setting up certain filtration points, extracting water samples from these points after the experiment, and then using a turbidimeter to read the turbidity data. However, because fine particles easily adhere to the pipe wall during filtration, non-destructive sampling is impossible, making it difficult to guarantee the accuracy of the data.

[0005] 2) The method of using a limited number of sensors to measure the suspended diffusion data of fine particles in water avoids the error caused by fine particles adhering to the pipe wall during the filtration process. However, due to the discrete distribution of measurement points and the fact that too many measurement devices can easily interfere with the surrounding flow field, it is also difficult to accurately obtain the spatiotemporal distribution of fine particles in the entire experimental area.

[0006] 3) Most existing devices for measuring the concentration distribution of fine particles in water lack versatility and universality, have high design and construction costs, poor economic efficiency, and the accuracy of measurement data needs to be improved.

[0007] Comparison of patent document list:

[0008] CN116486564A Smoke monitoring and alarm device and method;

[0009] CN116403381A A smoke monitoring method and device, a smoke alarm method and system;

[0010] CN115717671A A three-dimensional monitoring device for fine particulate matter in the atmosphere;

[0011] CN112068141A An environmental monitoring device for sediments from deep-sea polymetallic nodule mining;

[0012] CN116380741A A device and method for detecting the composition of boiler combustion flue gas;

[0013] This invention takes the identification and measurement of disturbed sediment diffusion as its starting point and further extends it to enable high-precision, multi-scale, long-period measurement of the concentration distribution of fine particles in various types of water. Summary of the Invention

[0014] The purpose of this invention is to provide a device and method for measuring the concentration distribution of fine particles in water based on multi-source information fusion, in order to solve the problems of low accuracy and poor adaptability of existing fine particle concentration distribution measurement technologies in water.

[0015] The present invention adopts the following technical solution:

[0016] A device for measuring the concentration distribution of fine particles in water based on multi-source information fusion includes a calibration water tank, an experimental source data acquisition mechanism, an image processing mechanism, and a laser turbidity measurement module. The calibration water tank has a water tank slide rail 2 and a slider that slides along the water tank slide rail 2. A telescopic slide rail is vertically fixed to the slider, and a telescopic rod is vertically fixed to the lower part of the telescopic slide rail. A clamping component is rotatably fixed to the telescopic rod and fixes the camera device 3. The telescopic slide rail, telescopic rod, and clamping component constitute a 360° rotatable support 4. The experimental data acquisition mechanism includes a water sampling turbidity measurement device, a mud pan volume difference measurement device, a mud pan mass difference measurement device, and a laser measurement module. A turbidity measuring device; the water sampling and turbidity measuring device has at least one set, each set including: two parallel water sampling slide rails 9, several carbon fiber rods 8, several water sampling pipes 7, and a turbidity measuring module; the water sampling slide rails 9 are embedded in the grooves on the side wall of the water tank and can slide along the water tank; the upper and lower ends of the carbon fiber rods 8 are embedded in the tracks on the water sampling slide rails 9 and can slide along the water sampling slide rails 9; the water sampling ports 7a of each water sampling pipe 7 are fixed at intervals on the carbon fiber rods 8, and the water sample at the location is collected by its respective turbidity measuring module to the water storage mechanism outside the calibration water tank. The turbidity measuring module uses the principle of laser scattering to obtain the actual measured turbidity value of the water sample at that location, which serves as the first calibration data source; the mud pan The volume difference measurement device uses 3D scanning technology to scan the surface shape of the sediment before and after the calibration experiment, and calculates the volume difference of the bottom sediment before and after the experiment, which serves as the second calibration data source. The mud pan mass difference measurement device includes a flat plate structure and a weighing device set at the bottom of the calibration tank. The fine particles are initially laid flat at the bottom of the tank. The water sampling slide rail 9 slides along the slide rail 9 to simulate the movement of the track of seabed mining equipment. The fine particles are suspended after being disturbed by the sliding of the water sampling slide rail 9. The mass difference of the bottom fine particles before and after the disturbance is measured by the weighing device, and the mass of the suspended particles is calculated. The mass of the suspended particles is positively correlated with the severity of the disturbance and the particle concentration, and serves as the third calibration data. The laser turbidity measurement device includes a laser emitter and receiver mounted on a slide rail in a water tank. Based on the different refraction and scattering degrees of laser light by fine particulate matter of varying concentrations, the device reads the light intensity data recorded by the laser receiver. Turbidity data is obtained from the mapping relationship of the light intensity data and used as the fourth calibration data source. The first, second, third, and fourth calibration data sources form spatial discrete-point turbidity monitoring data through mathematical operations. The image processing mechanism trains an image recognition model based on the mapping relationship between the spatial discrete-point turbidity monitoring data obtained by the experimental source data acquisition mechanism and the image pixel information acquired by the camera device 3, thereby capturing and reconstructing the changes in the concentration distribution of fine-particle soil in the target area. It should be noted that the first, second, third, and fourth calibration data sources serve as the sources for four mathematical operations. The specific mathematical operation methods are not a direct contribution of this application to the prior art, and there are multiple ways to choose from for the specific mathematical operations.

[0017] Preferably, the clamping component includes a pair of vertically parallel clamping rods, and the camera device 3 is rotatably connected to the pair of vertically parallel clamping rods.

[0018] Preferably, the target area is a seabed mineral collection area.

[0019] Preferably, the external water storage mechanism is a beaker or a test tube.

[0020] Preferably, the disturbance source disposed in the calibration tank is a device capable of generating at least one of walking disturbance, emission disturbance, and impact disturbance. This patent does not elaborate on the disturbing object, as it is prior art.

[0021] A method for measuring the concentration distribution of fine particles in water based on multi-source information fusion is proposed. The method uses the aforementioned device for measuring the concentration distribution of fine particles in water based on multi-source information fusion. Before and after the calibration experiment, the surface shape of the sediment is scanned using 3D scanning. The volume difference and mass difference of the bottom sediment before and after the experiment are calculated. The suspended mass of fine soil particles is inferred from this and fused with the concentration distribution data obtained from image processing for verification.

[0022] Preferably, the calibration procedure is set up as follows: the calibration materials used include glass beads of different colors with diameters of 1-200 micrometers and quartz powder with diameters of 1-200 micrometers; before the formal experiment begins, the calibration materials are used to simulate sediment, laid at the bottom of the calibration tank and a calibration material perturbation experiment is carried out. The calibration material perturbation experiment is consistent with the formal experiment procedure. Compared with the actual sediment with low color recognition, the calibration materials can help train the image recognition model and improve the reliability and accuracy of the model.

[0023] Preferred algorithm for reconstructing fine particle concentration distribution field:

[0024] After obtaining the concentration field samples, the concentration field of the entire space can be recovered through spatial interpolation methods. This system uses inverse distance weighted interpolation to reconstruct the concentration field; the specific algorithm of inverse distance weighted interpolation is as follows:

[0025] Let x1, x2, ... be a series of observation points in the region, and Z(x1), Z(x2), ... be the corresponding observation values. The value Z(x0) at the interpolation point x0 can be estimated using a linear combination:

[0026] Among them, w i The weight is related to the value at point x. i The distance between x0 and x0 is inversely proportional. Assume the positions p of n discrete sampling points in a given space... i and its corresponding concentration value D(p) i If ), then the interpolation result at position r is:

[0027]

[0028] The weight function w is defined as follows:

[0029]

[0030] The parameter R represents the radius of the interpolation weighting range. That is, sampling points that are more than R away from the point to be interpolated are not considered. The value of R can be manually selected during the specific interpolation process. When selecting R, it is ensured that for each point r to be interpolated, there is at least one sampling point within the neighborhood of R.

[0031] The beneficial effects of this invention are as follows:

[0032] Advantage 1: Improved data accuracy

[0033] By combining a sophisticated water sampling mechanism with a non-contact high-speed camera, the data error problem caused by the limitation of the number of sampling points and the size of water pipes in the traditional filtration method is solved, and the accuracy of this device is significantly improved.

[0034] Advantage 2: [Improved Data Continuity and Comprehensiveness]

[0035] By using multi-camera, multi-angle, non-contact high-resolution, high-frame-rate video recording and reinforcement learning algorithms, it is possible to analyze fine particle diffusion profiles captured by high-speed cameras at any angle, enabling data analysis of fine particles disturbed in water across the entire time and spatial domains, and continuous, accurate, and comprehensive capture and reconstruction of the distribution changes of fine particle soil concentration in mining areas.

[0036] Advantage 3: [Increased system scalability and economic benefits]

[0037] This device integrates a water sampling mechanism, a calibration tank, and image processing technology. Each module connection is equipped with a sliding rail mechanism, which allows for arbitrary addition or removal of accessories and adjustment of accessory positions. Compared with traditional methods for measuring the concentration distribution of fine particles in water, this device has wider applicability, can achieve large-scale mass production, reduce the cost of related research, and improve economic efficiency. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the external shape of the water fine particle concentration distribution measurement device based on multi-source information fusion according to the present invention.

[0039] Figure 2 This is a perspective schematic diagram of the water fine particle concentration distribution measurement device based on multi-source information fusion according to the present invention.

[0040] Figure 3 yes Figure 2 A magnified view of the details on the left.

[0041] Figure 4 yes Figure 2 A magnified view of the details in the middle section.

[0042] Figure 5 This is an isometric view of the water sampling device.

[0043] Figure 6 This is a schematic diagram of a laser transmitter and a laser receiver.

[0044] Figure 7 This is a bottom view of the weighing plate.

[0045] Figure 8 This is a technical flowchart of the present invention, taking the measurement of sediment concentration distribution data in a mineral collection experiment as an example.

[0046] Figure 9 This is a logic block diagram illustrating the principle of the water fine particle concentration distribution measurement method based on multi-source information fusion, as described in this invention.

[0047] In the diagram, 1. Calibration tank; 2. Tank slide rail; 3. High-speed camera; 4. 360° rotatable bracket; 5. Laser emitter; 6. Laser receiver; 7. Water sampling pipe; 8. Carbon fiber rod; 9. Water sampling slide rail; 10. Weighing plate; 11. Computer; 12. Movable slider; 7a. Water sampling port. Detailed Implementation

[0048] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0049] To address the issues of low accuracy and poor adaptability in existing water fine particle concentration distribution measurement technologies, this embodiment introduces a water fine particle concentration distribution measurement device and method based on multi-source information fusion.

[0050] It should be noted that the disturbance source in the calibration tank is not shown in the attached figure. The main purpose of this invention is to simulate disturbances during the seabed mining process, disturbances during the movement of tracked mechanisms on the seabed, disturbances during the seabed dredging process, and disturbances during the seabed dredging process. However, it can also simulate other forms of disturbance, such as disturbances from the discharge of sediments.

[0051] Technical solutions and research methods:

[0052] Taking the measurement of sediment concentration distribution data in a mineral collection experiment as an example, the technical process of this invention is as follows: Figure 4 As shown.

[0053] The device for measuring the concentration distribution of fine particles in water based on multi-source information fusion includes a calibration tank, an experimental source data acquisition mechanism, and an image processing mechanism.

[0054]

Calibrating the water tank mechanism

[0055] In this embodiment, the dimensions of the calibration tank can be customized according to user needs. The top of the tank is equipped with a sliding rail, which can be configured with a movable cantilever high-speed camera and water sampling mechanism according to user requirements.

[0056] [Water intake module]

[0057] This device can be configured with a data acquisition mechanism of a certain scale according to user needs. One acquisition mechanism consists of two upper and lower slide rails, several carbon fiber rods, and several water sampling and turbidity measurement modules. The slide rails are used to position the acquisition plane within the calibration tank; the upper and lower ends of the carbon fiber rods are embedded in the slide rails and can move along the slide rails within the acquisition plane; the water sampling devices are fixed by the carbon fiber rods, and the number and spacing are configured according to user needs. The water sampling mechanism uses a custom-made thin, flexible tube to collect water samples from the specified location to a water storage device (such as a beaker or test tube) outside the tank. Simultaneously, the water sample passes through the turbidity measurement module, which uses the principle of laser scattering to obtain the actual turbidity value of the water sample at that location, thus obtaining the calibration data source.

[0058] Laser Turbidity Measurement Module

[0059] See Figure 2 and Figure 6 This device is equipped with a laser turbidity measurement module. A laser transmitter and receiver are installed on the slide rail of the water tank. Based on the different degrees of refraction and scattering of laser light by fine particulate matter of different concentrations, the light intensity data recorded by the laser receiver is read, and the turbidity data is further obtained from the mapping relationship.

[0060] [Weighing Module]

[0061] See Figure 2 and Figure 7 A flat-plate weighing device is installed at the bottom of the calibration tank. Fine particulate matter initially lies flat at the bottom of the tank, becoming suspended after disturbance. The weighing device measures the mass difference of the fine particulate matter at the bottom before and after disturbance, calculating the mass of the suspended particulate matter. A larger suspended mass indicates severe disturbance and a correspondingly higher concentration of suspended particles; a smaller suspended mass indicates mild disturbance and a correspondingly lower concentration of suspended particles. Combining multi-source data collected by the water sampling module, laser measurement module, and mass difference measurement module further enhances the accuracy of the device's measurement results.

[0062] Image Processing

[0063] A sliding rail 2 is configured at the top of the calibration tank. Multiple 360° rotatable cantilever brackets 4 can be installed on the sliding rail 2 during the calibration process, according to user needs. A high-speed camera is installed at the other end of the bracket 4. Through multi-angle, non-contact, high-resolution, high-frame-rate imaging, the distribution profile of fine-grained sediments at any time and spatial location within the tank is obtained. Analyzing the cross-sectional images captured by the high-speed camera, clear water without sediment is recorded as white (255), and the point with the highest sediment concentration is recorded as black (0). Based on the mapping relationship between the spatial discrete point monitoring data obtained by the water sampling mechanism and the image pixel information, an image recognition model is trained to achieve continuous, accurate, and comprehensive capture and reconstruction of the distribution changes of fine-grained soil concentration in the mining area.

[0064] 3D scanning of sediment layers

[0065] Before and after the calibration experiment, the surface shape of the sediment was scanned using 3D scanning technology. The volume and mass differences of the bottom sediments before and after the experiment were calculated, thereby inferring the suspended mass of fine-particle soil. This data was then fused and corroborated with the concentration distribution data obtained from image processing, further improving the accuracy of the measurement system.

[0066] Calibration and calibration module

[0067] To further improve the accuracy of the device, a calibration module was set up. The calibration materials consist of quartz powder of different sizes and colored glass beads. Before the formal experiment, the calibration materials can be used to simulate sediment, laid at the bottom of the calibration tank, and a calibration material perturbation experiment is conducted, following the same procedure as the formal experiment. Compared to actual sediment, which has lower color recognition, the calibration materials can assist in training the image recognition model, improving the model's reliability and accuracy.

[0068] Research Methods:

[0069] Algorithm for reconstructing fine particle concentration distribution field

[0070] After obtaining the concentration field samples, the concentration field of the entire space can be reconstructed using spatial interpolation methods. This system employs inverse distance weighted interpolation to reconstruct the concentration field. The specific algorithm of inverse distance weighted interpolation is as follows:

[0071] Let x1, x2, ... be a series of observation points in the region, and Z(x1), Z(x2), ... be the corresponding observation values. The value Z(x0) at the interpolation point x0 can be estimated using a linear combination:

[0072]

[0073] Among them, w i The weight is related to the value at point x. i The distance between x0 and x0 is inversely proportional. Assume the positions p of n discrete sampling points in a given space...i and its corresponding concentration value D(p) i If ), then the interpolation result at position r is:

[0074]

[0075] The weight function w is defined as follows:

[0076]

[0077] The parameter R represents the radius of the interpolation weighting range. Sampling points that are more than R away from the point to be interpolated are not considered. The value of R can be manually selected during the interpolation process. It is important to note that when selecting R, it should be ensured that for each point r to be interpolated, there is at least one sampling point within the neighborhood of R.

[0078] Innovation points:

[0079] Innovation Point 1: Underwater dynamic multi-camera, multi-angle image capture:

[0080] In this invention, the top edge of the calibration tank is equipped with a sliding rail mechanism, allowing any number of high-speed cameras to be installed at any position on the rail according to user needs. The rotatable cantilever bracket enables 360° omnidirectional angle fixation. Together, these mechanisms ensure that the high-speed cameras can capture images from multiple positions and angles. The high-speed cameras can slide freely along the rails, enabling dynamic scanning of underwater profiles, expanding the capture range of the high-speed cameras and achieving coupling between camera movement and sediment diffusion and evolution. Furthermore, through the airtight design of the high-speed cameras, this device can be further applied to actual sea trials.

[0081] Innovation Point 2: Reconstruction of the three-dimensional spatial distribution of fine particle concentration:

[0082] Compared to the discrete point data values ​​obtained by traditional filtration methods and pure sensor measurement methods, this invention introduces image recognition technology to construct a mapping relationship between multi-source data obtained by the water sampling module, laser measurement module, and weighing module and image information captured by high-speed cameras in the full time and space domains. By using non-contact, high-resolution, high-frame-rate cameras and reinforcement learning algorithms, the concentration distribution of fine-grained soil in the mining area can be continuously, accurately, and comprehensively captured and reconstructed.

[0083] Innovation Point 3: Low-flow-field disturbance water sampling:

[0084] In traditional filtration and pure sensor measurement methods, the number of sampling points and the size of the filtration tube directly affect the accuracy of the results. In this invention, the water sampling device is equipped with sliding rail mechanisms at both ends, allowing users to easily add or remove accessories and arbitrarily position them as needed. The water sampling device is fixed by carbon fiber rods, which possess characteristics such as high temperature resistance, friction resistance, thermal conductivity, and corrosion resistance. Compared to traditional fixed supports, the carbon fiber rods have a smaller diameter, greatly reducing disturbance to calibration experiments and improving the accuracy of the device.

[0085] Beneficial effects and advantages:

[0086] Advantage 1: Improved data accuracy

[0087] By combining a sophisticated water sampling mechanism, a laser turbidity measurement module, a weighing module, and a non-contact high-speed camera, the data error problem caused by the limitation of the number of sampling points and the size of the water pipe in the traditional filtration method is solved, and the accuracy of this device is significantly improved.

[0088] Advantage 2: [Improved Data Continuity and Comprehensiveness]

[0089] By using multi-camera, multi-angle, non-contact high-resolution, high-frame-rate video recording and reinforcement learning algorithms, it is possible to analyze fine particle diffusion profiles captured by high-speed cameras at any angle, enabling data analysis of fine particles disturbed in water across the entire time and spatial domains, and continuous, accurate, and comprehensive capture and reconstruction of the distribution changes of fine particle soil concentration in mining areas.

[0090] Advantage 3: [Increased system scalability and economic benefits]

[0091] This device integrates a water sampling mechanism, a calibration tank, and image processing technology. Each module connection is equipped with a sliding rail mechanism, which allows for arbitrary addition or removal of accessories and adjustment of accessory positions. Compared with traditional methods for measuring the concentration distribution of fine particles in water, this device has wider applicability, can achieve large-scale mass production, reduce the cost of related research, and improve economic efficiency.

[0092] Image recognition technology for the spatiotemporal evolution of disturbed suspension diffusion of soft sediments typically employs a limited number of sensors to measure the spatiotemporal distribution of disturbed suspension diffusion of fine-grained soil. However, due to the discrete distribution of measurement points and the interference of excessive measurement equipment on the surrounding flow field, it is difficult to accurately obtain the spatiotemporal distribution of fine-grained soil throughout the entire mining area. This invention introduces image recognition and reinforcement learning methods, innovatively proposing an image recognition technology for the spatiotemporal distribution of disturbed suspension diffusion of fine-grained soil. Based on the mapping relationship between multi-source monitoring data such as sediment concentration at spatially discrete points and linear or surface laser refractive index and image information, a reinforcement learning model is designed. Through multi-camera, multi-angle, non-contact, high-resolution, high-frame-rate photography and reinforcement learning algorithms, continuous, accurate, and comprehensive capture and reconstruction of the fine-grained soil concentration distribution changes in the mining area can be achieved. This provides a new technical means for studying and analyzing the spatiotemporal evolution of sediment suspension diffusion during the mining process.

[0093] In this invention, the measurement results (multi-source data) of the following five parts are intrinsically related: ① mud pan volume change (measured by a 3D scanner), ② mud pan underwater weight change (measured by a high-precision mechanical sensor), ③ turbidity of water samples drawn using a thin flexible tube (measured in air using a high-precision turbidimeter), ④ underwater turbidity measurement method based on laser attenuation characteristics (laser transmitter and laser receiver signals), and ⑤ underwater image pixel information (multi-angle cameras). ① and ② measure the source of disturbance (the root cause of fine-particle soil suspension and diffusion due to jet breaking), while ③, ④, and ⑤ measure the diffusion area of ​​fine-particle soil. Combining the mass conservation of fine-particle soil and utilizing the intrinsic relationship, a conversion formula is established between these five measurement results, ultimately achieving a significant improvement in the accuracy of measuring the concentration distribution of fine particles in water. A new, highly reliable, and highly accurate method and device for measuring the temporal and spatial variations of fine-particle concentration in water are proposed.

[0094] In the water sampling process of this invention, the interference with the original flow field is minimized, and the sampling is currently the most accurate. This is because the placement of the suction head in the pool inevitably causes some disturbance. The laser measurement method is an advanced non-contact measurement method, but there will be interference from different profiles in the width direction of the pool. This needs to be solved by inverting the algorithm and constructing a new mapping relationship. When measuring the distributed concentration of fine particles, the volume change of the mud disc and the underwater mass change before the disturbance are measured first. This is the source of the disturbance and is directly related to the subsequent suspension and diffusion of fine particles. It is equivalent to providing a new solution condition for the mapping relationship, which can make the comprehensive method of fine particle concentration measurement based on multi-source data more accurate and reliable.

[0095] The above are preferred embodiments of the present invention. Those skilled in the art can make their own modifications or improvements based on this, and such modifications or improvements should fall within the scope of protection claimed by the present invention without departing from the overall concept of the present invention.

Claims

1. A device for measuring the concentration distribution of fine particles in water based on multi-source information fusion, characterized in that: Includes a calibration tank, an experimental source data acquisition mechanism, an image processing mechanism, and a laser turbidity measurement module; The calibration water tank is provided with a water tank slide rail (2) and a slider that slides along the water tank slide rail (2). The telescopic slide rail is vertically fixed on the slider, and the telescopic rod is vertically fixed at the lower part of the telescopic slide rail. The clamping component is rotated and fixed on the telescopic rod in a left-right rotation manner, and fixes the camera device (3). The telescopic slide rail, telescopic rod, and clamping component constitute a 360° rotatable bracket (4). The experimental data source acquisition mechanism includes a water sampling turbidity measurement device, a mud disc volume difference measurement device, a mud disc mass difference measurement device, and a laser turbidity measurement device. The water sampling and turbidity measurement device has at least one set, each set including: two parallel water sampling slide rails (9) at the top and bottom, several carbon fiber rods (8), several water sampling pipes (7) and a turbidity measurement module; the water sampling slide rails (9) are embedded in the grooves on the side wall of the water tank and can slide along the water tank; the upper and lower ends of the carbon fiber rods (8) are embedded in the tracks on the water sampling slide rails (9) and can slide along the water sampling slide rails (9); the water sampling ports (7a) of each water sampling pipe (7) are fixed at intervals on the carbon fiber rods (8) to collect water samples at the location through their respective turbidity measurement modules to the water storage mechanism outside the calibration water tank; the turbidity measurement module uses the principle of laser scattering to obtain the actual measured turbidity value of the water sample at the location, which serves as the first calibration data source; The mud disc volume difference measuring device uses 3D scanning technology to scan the surface shape of the sediment before and after the calibration experiment, and calculates the volume difference of the bottom sediment before and after the experiment, which serves as the second calibration data source. The mud disc mass difference measuring device includes a flat plate structure and a weighing device installed at the bottom of the calibration tank. The fine particles are initially spread flat at the bottom of the tank. After the disturbance source disturbs the sediment, the sediment becomes suspended. The weighing device measures the mass difference of the bottom fine particles before and after the disturbance, and calculates the mass of the suspended particles. The mass of the suspended particles is positively correlated with the severity of the disturbance and the particle concentration, and serves as the third calibration data source. The laser turbidity measurement device includes a laser emitter and a receiver installed on the slide rail of the water tank. Based on the different degrees of refraction and scattering of laser light by fine particulate matter of different concentrations, the light intensity data recorded by the laser receiver is read, and the turbidity data is obtained from the mapping relationship of the light intensity data as the fourth calibration data source. The first, second, third, and fourth calibration data sources are used to generate spatially discrete point turbidity monitoring data through mathematical operations; The image processing mechanism trains an image recognition model based on the mapping relationship between the spatial discrete point turbidity monitoring data obtained by the experimental source data acquisition mechanism and the image pixel information acquired by the camera device (3), so as to capture and reconstruct the distribution changes of fine-grained soil concentration in the target area.

2. The water fine particle concentration distribution measurement device based on multi-source information fusion as described in claim 1, characterized in that: The clamping component includes a pair of vertically parallel clamping rods, and the camera device (3) is rotatably connected to the pair of vertically parallel clamping rods.

3. The water fine particle concentration distribution measurement device based on multi-source information fusion as described in claim 1, characterized in that: The target area is a seabed mineral collection area.

4. The water fine particle concentration distribution measurement device based on multi-source information fusion as described in claim 1, characterized in that: The external water storage mechanism is a beaker or a test tube.

5. The water fine particle concentration distribution measurement device based on multi-source information fusion as described in claim 1, characterized in that: The disturbance source set in the calibration tank is a device capable of realizing at least one of walking disturbance, emission disturbance, and impact disturbance.

6. A method for measuring the concentration distribution of fine particles in water based on multi-source information fusion, characterized in that: The water fine particle concentration distribution measurement device based on multi-source information fusion as described in any one of claims 1-4 is used; before and after the calibration experiment, the surface shape of the sediment is scanned by 3D scanning, the volume difference and mass difference of the bottom sediment before and after the experiment are calculated, and the suspended mass of fine soil particles is inferred from this, which is fused and verified with the concentration distribution data obtained by image processing.

7. The method for measuring the concentration distribution of fine particles in water based on multi-source information fusion as described in claim 5, characterized in that: The calibration procedure is as follows: The calibration materials used include glass beads of different colors with diameters ranging from 1 to 200 micrometers and quartz powder with diameters ranging from 1 to 200 micrometers. Before the formal experiment begins, the calibration materials are used to simulate sediments, which are laid at the bottom of the calibration tank and a calibration material perturbation experiment is carried out. The calibration material perturbation experiment is consistent with the formal experiment procedure. Compared with actual sediments with lower color recognition, the calibration materials can help train the image recognition model and improve the reliability and accuracy of the model.

8. The method for measuring the concentration distribution of fine particles in water based on multi-source information fusion as described in claim 5, characterized in that: Fine particle concentration distribution field reconstruction algorithm: After obtaining the concentration field samples, the concentration field of the entire space can be recovered through spatial interpolation methods. This system uses inverse distance weighted interpolation to reconstruct the concentration field; the specific algorithm of inverse distance weighted interpolation is as follows: Let x1, x2, ... be a series of observation points in the region, and Z(x1), Z(x2), ... be the corresponding observation values. The value Z(x0) at the interpolation point x0 can be estimated using a linear combination: Among them, w i The weight is related to the value at point x. i The distance between x0 and x0 is inversely proportional. Assume the positions p of n discrete sampling points in a given space... i and its corresponding concentration value D(p) i If ), then the interpolation result at position r is: The weight function w is defined as follows: The parameter R represents the radius of the interpolation weighting range. That is, sampling points that are more than R away from the point to be interpolated are not considered. The value of R can be manually selected during the specific interpolation process. When selecting R, it is ensured that for each point r to be interpolated, there is at least one sampling point within the neighborhood of R.

Images