A real-time abundance detection system and method for deep-sea polymetallic nodule minerals

Abstract

The present invention discloses a real-time abundance detection system and method for deep-sea polymetallic nodule minerals. The system comprises a forward-scanning sonar (FLS) acoustic backscatter level analysis module, a forward-looking sonar (FLS) processing saliency detection and analysis module, an image processing module, a side-looking sonar (SLS) processing module, a multi-dimensional nodule counting and sizing module, a machine learning module, an abundance mapping module, and a machine sensor. The system detects and maps local polymetallic nodule abundance trends in deep-sea polymetallic nodule mining areas. A work-class ROV or AUV performs pre-mapping scans, and a deep-sea polymetallic nodule mining vehicle performs real-time scanning and mapping of nodule ore while mining. Real-time abundance detection is used to automatically control the vehicle's path, speed, and collection operations, significantly enhancing the yield and recovery rate of the deep-sea polymetallic nodule mining system.

Description

Technical Field

[0001] The present invention relates to the technical field of mining equipment, and in particular to a real-time abundance detection system and method for deep-sea polymetallic nodule minerals. Background Art

[0002] Mineral resources, also known as mineral resources, refer to a collection of minerals or useful elements that are formed through geological mineralization, naturally occur in the earth's crust or on the surface, buried underground or exposed on the surface, in solid, liquid or gaseous form, and have development and utilization value.

[0003] In recent years, with economic development, the demand for mineral resources has also increased. Extensive resource development has led to the depletion of Earth's terrestrial mineral resources. Terrestrial mineral resources can no longer meet humanity's demand for mineral resources, so humanity needs to explore new ways to supply mineral resources. The seabed is rich in mineral resources. The rational exploitation of marine mineral resources can help solve the problem of insufficient terrestrial mineral resource supply. Polymetallic nodules are rich in dozens of elements such as manganese, iron, and nickel. They are a mineral resource of great economic value and are mostly found in deep sea areas of 4,000 to 6,000 meters. On the one hand, their transportation method is completely different and difficult. On the other hand, production efficiency and economic feasibility are also important considerations. At the same time, due to the high technical difficulty and complexity of deep-sea mining, this poses a great challenge to the technical requirements and solutions.

[0004] The existing technology requires a set of intermediate chamber equipment. Generally speaking, the slurry collected by the mining vehicle is connected to the intermediate chamber at the bottom of the riser through a hose. The intermediate chamber is an important equipment used to buffer the impact of changes in slurry concentration on the slurry lifting pump. The existence of this equipment will make the retraction and deployment of the offshore riser system more complicated and more expensive. There is also the possibility that the output and recovery rate of the deep-sea polymetallic nodule mining system cannot be guaranteed, resulting in low output and recovery rate. Therefore, how to simply and energy-savingly improve the output and recovery rate of the deep-sea polymetallic nodule mining system is an urgent problem that needs to be solved.

[0005] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0006] The main purpose of the present invention is to provide a real-time abundance detection system and method for deep-sea polymetallic nodule minerals, aiming to solve the problems of high complexity of existing deep-sea polymetallic nodule mining systems, great difficulty in retracting and releasing mining buffer bins, and unstable production, which lead to low comprehensive recovery rates.

[0007] To achieve the above-mentioned object, the present invention provides a real-time abundance detection system for deep-sea polymetallic nodule minerals, the real-time abundance detection system for deep-sea polymetallic nodule minerals comprising:

[0008] Forward scanning sonar FLS acoustic backscatter level analysis module, forward looking sonar FLS processing saliency detection and analysis module, image processing module, side looking sonar SLS processing module, multi-dimensional nodule counting and size measurement module, abundance mapping module, machine learning module and machine sensor;

[0009] The machine learning module is respectively connected to the forward scanning sonar FLS acoustic backscatter level analysis module, the forward looking sonar FLS processing prominence detection and analysis module, the image processing module, the side looking sonar SLS processing module, the machine sensor and abundance mapping module, and the multi-dimensional nodule counting and size measurement module is respectively connected to the forward scanning sonar FLS acoustic backscatter level analysis module, the forward looking sonar FLS processing prominence detection and analysis module, the image processing module, the side looking sonar SLS processing module and the abundance mapping module;

[0010] The forward scanning sonar FLS acoustic backscatter level analysis module is used to comprehensively analyze the overall intensity of the signal returned by the FLS sonar, and input the analysis result information into the machine learning module and the multi-dimensional nodule counting and size measurement module;

[0011] The forward-looking sonar FLS processing saliency detection and analysis module is used to input the detected and analyzed target signal into the machine learning module and the multi-dimensional nodule counting and size measurement module;

[0012] The image processing module is used to process the image to obtain a processing result, and input the processing result into the machine learning module and the multi-dimensional tuberculosis counting and size measurement module;

[0013] The side-looking sonar SLS processing module is used to input the data of polymetallic nodules in each scanning sector into the machine learning module and the multi-dimensional nodule counting and size measurement module;

[0014] The multi-dimensional nodule counting and size measurement module is used to calculate the density and average size of polymetallic nodules in each sector through probability estimation and data fusion methods;

[0015] The abundance mapping module is used to receive target data to determine an abundance detection value for each region and store data associating sensor fusion data with corresponding ratios.

[0016] Optionally, the real-time abundance detection system for deep-sea polymetallic nodule minerals, wherein the result information includes the overall acoustic backscatter level, the backscatter intensity histogram of each sector of the seabed cover, the sum of backscatter of highly reflective objects, the seabed background data of deep-sea polymetallic nodules and the average and standard values of nodule intensity and the average water column backscatter level related to turbidity and sonar parameters.

[0017] Optionally, the real-time abundance detection system for deep-sea polymetallic nodule minerals, wherein the target signal includes each regional sector, peak number and geometric area descriptor; wherein the geometric area descriptor includes average area and geometric moment.

[0018] Optionally, the real-time abundance detection system for deep-sea polymetallic nodule minerals, wherein the processing results include: each sector of the mining vehicle's field of view FOV, the count of polymetallic nodules detected in the sector, and the average area of the geometric descriptors.

[0019] Optionally, the real-time abundance detection system for deep-sea polymetallic nodule minerals, wherein the signals collected by the machine sensors include the speed of the mining vehicle, the speed of the conveyor belt, the currently measured slurry concentration and an initial estimate of the abundance of each geometric area.

[0020] Optionally, in the real-time abundance detection system for deep-sea polymetallic nodule minerals, the target data includes: visual analysis data for each area and the ratio between the visual data and the actual ore concentration.

[0021] Optionally, in the real-time abundance detection system for deep-sea polymetallic nodule minerals, the machine learning module is specifically used to:

[0022] obtaining a collected amount of polymetallic nodules and parameters measured by the slurry, and comparing the collected amount of polymetallic nodules and the parameters measured by the slurry to obtain a ratio;

[0023] The visual and sonar detection signals are correlated with the actual mined ore output based on the ratio.

[0024] Optionally, in the real-time abundance detection system for deep-sea polymetallic nodule minerals, the machine learning module is further configured to:

[0025] Store a sequence of sensor data and obtain past measurements related to slurry data based on the sequence of sensor data

[0026] In addition, the present invention also provides a real-time abundance detection method for deep-sea polymetallic nodule minerals, wherein the real-time abundance detection method for deep-sea polymetallic nodule minerals comprises: the forward scanning sonar FLS acoustic backscatter level analysis module and the forward looking sonar FLS processing saliency detection and analysis module analyze the forward scanning sonar data to obtain analysis results, the image processing module processes the camera image to obtain processing results, and the side looking sonar SLS processing module obtains polymetallic nodule data from the structured light system point cloud;

[0027] The multi-dimensional nodule counting and size measurement module calculates the polymetallic nodule density and average size of each sector based on the analysis results, the processing results, and the polymetallic nodule data; and the machine learning module processes the analysis results, the processing results, the polymetallic nodule data, and additional information to obtain nodule collection parameters.

[0028] The abundance mapping module obtains a nodule abundance map based on the density and average size of the polymetallic nodules, the nodule collection parameters, and the location of the mining vehicle.

[0029] Optionally, the real-time abundance detection system method for deep-sea polymetallic nodule minerals, wherein the additional information includes the mining vehicle travel speed, the hoist belt speed, the crusher feed height, the slurry concentration and the initial abundance estimate.

[0030] The present invention discloses a real-time abundance detection system and method for deep-sea polymetallic nodule minerals. The real-time abundance detection system for deep-sea polymetallic nodule minerals comprises: a forward scanning sonar (FLS) acoustic backscatter level analysis module, a forward looking sonar (FLS) processing saliency detection and analysis module, an image processing module, a side looking sonar (SLS) processing module, a multi-dimensional nodule counting and size measurement module, an abundance mapping module, a machine learning module, and a machine sensor. The present invention is for detecting and mapping the abundance change trend of local polymetallic nodule minerals in deep-sea polymetallic nodule mining operation areas. A work-level ROV or AUV performs a pre-mapping scan, and a deep-sea polymetallic nodule mining vehicle performs mining operations while scanning and mapping nodule ores in real time. Real-time abundance detection is used to automatically control the walking path, speed and collection operations of the deep-sea mining vehicle, which plays a significant role in ensuring the output and recovery rate of the deep-sea polymetallic nodule mining system. By accurately detecting the real-time abundance, the walking speed of the mining vehicle is automatically adjusted. When the ore abundance is low, the mining vehicle speeds up to avoid wasting energy. When the ore abundance is high, the mining vehicle slows down, so that the ore flow rate collected by the mining vehicle is always maintained at a constant set value, so that the slurry lifting pump operates in an optimal efficiency range, saving energy consumption while eliminating a set of intermediate warehouse equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a schematic diagram of the principle of the real-time abundance detection system for deep-sea polymetallic nodule minerals in the present invention;

[0032] Figure 2 This is a schematic diagram of the Norbit WBMS-FLS forward-scan sonar device;

[0033] Figure 3 This is a schematic diagram of a forward-scan sonar image of the real-time abundance detection system for deep-sea polymetallic nodule minerals of the present invention;

[0034] Figure 4 This is a schematic diagram of the image sector of the forward scanning sonar FLS of the real-time abundance detection system for deep-sea polymetallic nodule minerals in the present invention;

[0035] Figure 5 This is a schematic diagram of nodule image processing in the real-time abundance detection system for deep-sea polymetallic nodule minerals of the present invention;

[0036] Figure 6 This is a schematic diagram of camera positions in the real-time abundance detection system for deep-sea polymetallic nodule minerals of the present invention;

[0037] Figure 7 Schematic diagram of nodule abundance assessment based on rolling window vision in the present invention;

[0038] Figure 8 This is a schematic diagram of a mining vehicle camera simulation view in the present invention;

[0039] Figure 9 is a schematic diagram of the structured light system and camera;

[0040] Figure 10 This is a schematic diagram of a sensor used for nodule abundance mapping in the real-time abundance detection system for deep-sea polymetallic nodule minerals of the present invention;

[0041] Figure 11 This is a schematic diagram of contour line analysis in the real-time abundance detection system for deep-sea polymetallic nodule minerals of the present invention;

[0042] Figure 12 It is a schematic diagram of the system architecture of the abundance map mapping software of the present invention;

[0043] Figure 13 It is a flow chart of a preferred embodiment of the real-time abundance detection method for deep-sea polymetallic nodule minerals in the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present invention more clear and distinct, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0045] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0046] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features specified as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that ordinary technicians in this field can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0047] The real-time abundance detection system for deep-sea polymetallic nodule minerals according to the preferred embodiment of the present invention is as follows: Figure 1 As shown, the real-time abundance detection system for deep-sea polymetallic nodule minerals includes: a forward scanning sonar FLS acoustic backscatter level analysis module 1, a forward looking sonar FLS processing prominence detection and analysis module 2, an image processing module 3, a side looking sonar SLS processing module 4, a multi-dimensional nodule counting and size measurement module 5, a machine learning module 6 and an abundance mapping module 7, except Figure 1 The system further includes a machine sensor 8; the machine learning module 6 is respectively connected to the forward-scanning sonar FLS acoustic backscattering level analysis module 1, the forward-looking sonar FLS processing prominence detection and analysis module 2, the image processing module 3, the side-looking sonar SLS processing module 4, the machine sensor 8 and the abundance mapping module 7; the multi-dimensional nodule counting and size measurement module 5 is respectively connected to the forward-scanning sonar FLS acoustic backscattering level analysis module 1, the forward-looking sonar FLS processing prominence detection and analysis module 2, the image processing module 3, the side-looking sonar SLS processing module 4 and the abundance mapping module 7.

[0048] The forward scanning sonar FLS acoustic backscatter level analysis module 1 is used to comprehensively analyze the overall intensity of the signal returned by the FLS sonar, and input the analysis result information into the machine learning module 6 and the multi-dimensional nodule counting and size measurement module 5.

[0049] Specifically, the main information source of the nodule abundance map is the forward scanning sonar installed at the front of the mining vehicle, which includes the forward scanning sonar FLS acoustic backscatter level analysis module 1 and the forward looking sonar FLS processing saliency detection and analysis module 2; the acoustic reflectivity of nodules is different from that of seabed sediments. The use of forward scanning sonar for abundance mapping has been widely studied, but has not been applied in deep-sea polymetallic nodule mining systems. When conditions are difficult, side scanning sonar can also be used to measure nodule abundance; in the current situation, close to the bottom of the mining vehicle, close-range (4 to 10 meters) multi-beam acoustic backscatter data can provide more detailed images. At closer distances, higher frequencies can be used, such as in the 400kHz–700kHz range ( Figure 2 A typical forward-scan sonar product model (Norbit WBMS-FLS) is given, and an example of a forward-scan sonar image is shown in the figure. Figure 3 As shown, the average backscattered acoustic echo over a selected area depends on the nodule distribution, but the exact relationship between nodule abundance, size, and acoustic level is unclear, as it also depends on the sonar configuration parameters and the specific acoustic reflectivity and morphology of the seafloor mud. However, a preliminary estimate of the expected nodule abundance, combined with the mud concentration and crusher hopper feed height, provides an indicative measure of the actual nodule collection. Therefore, coupling this information with visual imagery from the forward camera can be used in a machine learning algorithm to correlate the measured acoustic level with a parameter for nodule density. Furthermore, for forward-scan sonar images, where the image resolution is sufficient to represent individual nodules larger than 3 cm in diameter, the sonar image can also be processed to detect individual acoustic echoes for each nodule.

[0050] Furthermore, the forward-scan sonar FLS acoustic backscatter level analysis module 1 is based on a comprehensive analysis of the overall intensity of the FLS sonar return signal. The forward-scan sonar FLS acoustic backscatter level analysis module 1 outputs comprehensive acoustic backscatter level data as the result of sonar gain and terrain comparison. The result can be used to determine the threshold value for peak detection in the sonar echo (the threshold is used to reflect the reflection result of polymetallic nodules); the backscatter intensity histogram obtained by the threshold shows the distribution of the dynamic characterization of the backscatter intensity and displays the average backscatter level of the seabed (background) and seabed nodules and other highly reflective objects; the results of these front-end signal processing and analysis can be used as input information for the machine learning module 6, which will include: comprehensive acoustic backscatter level, that is, the normalization factor of the backscatter intensity, the backscatter intensity histogram provided by FLS for each sector of the seabed cover, the sum of the backscatter of highly reflective objects (by sector), the seabed background data of deep-sea polymetallic nodules and the average and standard value of the nodule intensity, and the average water column backscatter level related to turbidity and sonar parameters.

[0051] Among them, sectors are spatial partitions divided by distance and angle within the FLS scanning area, so as to facilitate measurement of the scanning area separately, such as Figure 4 shown.

[0052] The forward-looking sonar FLS processing prominence detection and analysis module 2 is used to input the target signal after detection and analysis into the machine learning module 6 and the multi-dimensional nodule counting and size measurement module 5.

[0053] Specifically, the forward-looking sonar FLS processing saliency detection and analysis module 2 outputs signals to the machine learning module 6 including: each regional sector, number of peaks (detected saliency) and geometric area descriptors (average area, geometric moments); these correspond to detected sonar echoes exceeding an adaptive threshold, reflecting "bright" spots caused by reflections from polymetallic nodules (because their reflection signals are greater than those from soft sandy seabeds); the average geometric description provides a rough size measurement and is therefore related to the size of the polymetallic nodules, which is indirectly related to the abundance of the original ore material; the sector divisions in the FLS image (and the scanned seabed area) can have different FLS peak counts and have spatially discrete sonar information detection.

[0054] The image processing module 3 is used to process the image to obtain a processing result, and input the processing result into the machine learning module 6 and the multi-dimensional tuberculosis counting and size measurement module 5.

[0055] In particular, the use of camera images for short-range nodule detection and characterization is a very valuable tool; images taken from relatively short distances to the bottom (between 4 and 8 meters) have been widely used for nodule characterization and ground truth analysis in surface acoustic surveys; because nodules have different grayscale values from the seabed, image processing can be used to provide an overall estimate of nodule abundance by analyzing color / grayscale and spot areas. Individual visible nodules can also be detected in these images by edge detection, mathematical morphology methods, and area fitting (see [1]). Figure 5 ); these results can not only provide an estimate of relative nodule abundance, but can also be used to determine the mean nodule diameter or size distribution and the nodule count; the actual relationship between the actual nodule diameter and the vision-based estimate (because many nodules are partially buried) can be adjusted by observing and measuring the nodules on the grid and corrected using the machine learning module 6; the inputs from the image processing module 3 to the machine learning module 6 are each sector of the mining vehicle's field of view FOV, the count of the number of blocks detected in the sector (the number of polymetallic nodules detected in the sector) and the average area of the geometric descriptors (from the area descriptors and geometric moments).

[0056] Furthermore, the mining vehicle has two front cameras that can observe the collection grid, spiral rake and the area near the seabed in front of the grid; these cameras are placed at a height of about 3 meters and can observe the seabed 6 meters in front of the seabed according to the field of view and longitudinal positioning; Figure 6 As shown, for example, a camera with a 60-degree vertical field of view installed at a height of 3 meters on a mining vehicle can observe approximately 6 meters ahead. Considering that the mining vehicle travels at a speed of 1.5 meters per second and captures images at a rate of 20 frames per second, each frame covers a ground strip of approximately 30 centimeters. Therefore, multiple strips in front of the mining head can be analyzed and the number of nodules in each strip determined, including an updated estimate of the number of nodules in the strip closest to the mining head. Figure 7 This is a top view of the mining machine moving to the left. Figure 7 Figure 2 shows the coverage of the terrain ahead of the mining vehicle by the vision cameras. The two cameras shown can observe the collection path ahead as well as a short section of the adjacent track, with two strips of terrain highlighted. These correspond to the expected area traversal completed by the mining vehicle at nominal speed during one frame capture. Therefore, in this case, for the duration of the image frame (for example, 50 milliseconds for a 20fps camera), a strip of approximately 30 to 50 cm is traversed. Since the camera can see several meters ahead, these multiple strips in a frame can be analyzed and the polymetallic nodule information of each strip estimated by continuous analysis of each frame. As the machine moves forward, a new strip enters the image from the top and the bottom strip exits. Therefore, multiple observations contribute to the data corresponding to the closer strips (just before entering the collection grid). Considering that typical 3-megapixel cameras are readily available, a density of more than 9 pixels per square centimeter can be obtained at a distance of up to 2 meters, a resolution sufficient for the detection and accurate counting of individual nodules.

[0057] Figure 8 An example of a simulated view is shown, which is captured by a 3m high camera with a horizontal field of view of 90 degrees, observing nodules with equal spacing of 0.5m and a 1m square grid with nodule diameters of 10cm (upper part of the figure) and 5cm (lower part of the figure), and 50% (right part of the figure) or 30% (left part of the figure) of the nodules exposed; it can be seen from the figure that in this configuration, each camera is 5m wide in front of the machine and 8m wide 3m away in front of the machine. The two cameras on both sides of the front of the machine can cover the entire mining tunnel and a few meters wide range of adjacent mining tunnels; considering the requirements of the camera and structured light system, the front camera needs to use a structured light flash; in addition, the relatively fast movement of the mining vehicle (about 1.5m / s) and pulsating light will require the selected camera to have a global shutter to eliminate the rolling band effect of the rolling shutter image sensor.

[0058] Furthermore, the mining vehicle is equipped with a structured light system that is used in conjunction with a camera (e.g. Figure 9 As shown in the figure, 3D point clouds of objects at close range (up to 2-3 meters) can be obtained. The operability and range of these structured light systems are severely restricted by turbidity and overall water visibility. The camera is located directly below the structured light system so that it can scan the seabed grid in front and the nearby area next to it. When visibility conditions permit, the structured light system provides a moving 3D point line near the mining vehicle, as shown in the figure. Figure 10 As shown, this line, which is very dependent on the actual seabed mud conditions, is projected in front of the spiral collection head. This line can also be used to provide more information on the nodule collection that is about to be collected, which helps the abundance measurement process; by analyzing this contour line, its irregularities and small contour deviations related to the seabed contour can be evaluated, thus providing complementary information on the number and size of nodules.

[0059] The side-looking sonar SLS processing module 4 is used to input the data of polymetallic nodules in each scanning sector into the machine learning module 6 and the multi-dimensional nodule counting and size measurement module 5.

[0060] Specifically, the signal output by the side-looking sonar SLS processing module 4 to the machine learning module 6 is the data of polymetallic nodules in each scanning sector (similar to the sector of FLS, with range / angle subdivision) (i.e., the number, average size, average volume or average projected area of nodules detected on the seabed); since the laser-based structured light system can provide precise size measurement, the cross-section of polymetallic nodules can still be accurately detected when the turbidity of seawater is low.

[0061] Furthermore, the signals collected by the machine sensors 8 include: mining vehicle speed, conveyor belt speed, currently measured slurry concentration, and an initial estimate of abundance in each geometric area (considered as a rectangular partition of the seabed area); these signal inputs are important for determining the delay between the collection area and the slurry measurement; the main sensors on the mining vehicle for nodule abundance detection are as follows: Figure 11 As shown, the circled ones are the main sensors.

[0062] The multi-dimensional nodule counting and size measurement module 5 is used to calculate the density and average size of polymetallic nodules in each sector through probability estimation and data fusion methods.

[0063] Specifically, the multidimensional nodule counting and size measurement module 5 introduces the input sensor signal of the machine learning module 6 into this module as the input signal, and takes the number and average size of polymetallic nodules in each geometric area as the output (the same area) to evaluate the output of the machine learning module 6; the multidimensional nodule counting and size measurement module 5 uses classical probability estimation and data fusion methods to combine multiple sensor data into a global measurement to combine multiple redundant and complementary measurements to smoothly estimate the density and average size of polymetallic nodules in each sector / geometric area.

[0064] The abundance mapping module 7 is used to receive target data to determine the abundance detection value of each area, and store data associated with the sensor fusion data and the corresponding ratio.

[0065] Specifically, the goal of the abundance mapping module 7 is to obtain the abundance (Kg / m2) of each area in front of the mining vehicle's travel path; wherein, the front-end module provides two types of data as inputs to the abundance mapping module 7: one is used for visual analysis of each area (counting detection and morphological information), and the other is the ratio between visual data and actual ore concentration (i.e., machine learning); these two inputs are analyzed and integrated in the abundance mapping module 7 to obtain the count and average nodule size of polymetallic nodules in each area, and the final abundance detection value of each area is determined with the actual ratio (for the previous paragraph), and is expressed in Kg / m2; and the abundance mapping module 7 has a storage function, which will be used to store data associated with sensor fusion data and corresponding ratios.

[0066] Furthermore, the process output of the machine learning module 6 is a comparison of the polymetallic nodule collection volume with the slurry measurement parameters, that is, the relationship between the number and size of polymetallic nodules that can be detected per unit area and the measured slurry content concentration, which can systematically correlate the visual and sonar detection signals with the actual mined ore output; wherein the nodule abundance assessment window is a rectangular partition located in front of the mining vehicle on the seabed, corresponding to a rolling window of a certain period in the past (including the expected signal delay), such as Figure 7 shown.

[0067] The machine learning module 6 will perform the corresponding measurements, and it is initially equipped with a delay mechanism, which reflects the relative time difference between the sensing data and the slurry concentration measurement, and reflects that the current slurry reading is the sensor data obtained a short time ago (delay). The duration of this delay is the estimated time of the estimator, and the time delay setting value can be estimated based on the mining vehicle speed, the conveyor belt speed and the estimated slurry transportation delay; in addition, the machine learning module 6 is also provided with a storage function, in which the sensing data sequence is stored in order to obtain past measurement values related to the slurry data; finally, within a determined time, the signal input data of the machine sensor 8, the forward scanning sonar FLS acoustic backscatter level analysis module 1, the forward looking sonar FLS processing saliency detection and analysis module 2, the image processing module 3 and the side looking sonar SLS processing module 4 are compared with the output results of the machine learning module 6; therefore, the output will include a set of ratios for comparing the relationship between the sensor measurement data of each unit area and the actual mineral measurement.

[0068] Furthermore, the machine learning module 6 can also: (1) estimate and detect polymetallic nodules from available heterogeneous sensor data; (2) calculate nodule abundance based on the polymetallic nodule collection volume and slurry measurements (ideally, it should be possible to do this for a small part of the terrain to obtain "weak" ground truth information); (3) use a multivariate regression model to relate the abundance to the acquired data; (4) estimate the mineral abundance based on a learning model that can be continuously updated in real time based on the slurry measurement information using a reinforcement learning scheme; among others, in order to solve the multivariate regression problem, different algorithms can be tested and implemented, namely: random forest regression, which is usually used when only a few training samples are available, support vector regression, using kernel-based SVMS or ANN in the case of highly nonlinear models.

[0069] Furthermore, the expected abundance estimate output (Kg / m2) is the result of data fusion from multiple sources (sonar, vision and potentially 3D laser scanning). For this estimate, other information can also be obtained on the mining vehicle, such as the pile height of the feed to the crusher and the pulp concentration (this is mainly nodules, because the silt has been largely removed); this information fusion will integrate machine learning algorithms such as artificial neural networks or vector machines, which can adjust multiple parameters. Therefore, in actual operation, the system can achieve continuous abundance mapping calibration by learning the relationship between observable parameters such as acoustic backscatter levels or image features and actual abundance values; when using a work-class ROV for preliminary surveys, the initial nodule estimate of the area can be obtained by analyzing the acoustic and image information obtained in the preliminary survey. During actual mining operations, acoustic images acquired by sensors on mining vehicles and work-class ROVs, as well as video images, can further improve the accuracy of abundance mapping through machine learning and data fusion; in addition, after slurry dewatering and separation treatment on the mining support vessel, the measured values of actual ore output can be used to input the machine learning process, thereby further improving the accuracy of abundance mapping; for each geometric area, the measured counts are compared with the measured slurry concentration, which is done at a certain time (data detected in the past); the internal estimator parameters are adjusted (for example, in the mining vehicle, the weights are adjusted by back propagation) to match the sensor counts / sizes with the measured broken minerals (because from a certain number of polymetallic nodules, the expected ratio of slurry to count can be derived).

[0070] Furthermore, if Figure 12The real-time abundance detection system architecture of deep-sea polymetallic nodule minerals is shown in the figure, which includes a forward-scan sonar FLS acoustic backscatter level analysis module, a forward-looking sonar FLS processing prominence detection and analysis module, an image processing module, a side-looking sonar SLS processing module, a multi-dimensional nodule counting and size measurement module, an abundance mapping module and a machine learning module; wherein, the forward-scan sonar FLS acoustic backscatter level analysis module and the forward-looking sonar FLS processing prominence detection and analysis module process the forward-scan sonar data and send the processing results to the machine learning module and the multi-dimensional nodule counting and size measurement module; the image processing module is used to process the image to obtain a processing result, and send the processing result to the machine learning module and the multi-dimensional nodule counting and size measurement module; Input into the machine learning module and the multi-dimensional nodule counting and size measurement module; the side-looking sonar SLS processing module is used to input the data of polymetallic nodules in each scanning sector into the machine learning module and the multi-dimensional nodule counting and size measurement module; the machine learning module processes the received data with the mining vehicle travel speed, hoist belt speed, crusher feed height, slurry concentration and initial abundance estimation, and sends the processing results to the abundance mapping module; the multi-dimensional nodule counting and size measurement module is used to calculate the polymetallic nodule density and average size of each sector through probability estimation and data fusion methods; the abundance mapping module obtains a nodule abundance map based on the received data and the mining vehicle position.

[0071] Further, based on Figure 1 The real-time abundance detection system of deep-sea polymetallic nodule minerals shown in the figure, the real-time abundance detection method of deep-sea polymetallic nodule minerals of the real-time abundance detection system of deep-sea polymetallic nodule minerals according to the preferred embodiment of the present invention, as shown in FIG. Figure 13 As shown, the real-time abundance detection method for deep-sea polymetallic nodule minerals includes the following steps:

[0072] Step S10: the forward-scan sonar FLS acoustic backscatter level analysis module and the forward-looking sonar FLS processing saliency detection and analysis module analyze the forward-scan sonar data to obtain analysis results, the image processing module processes the camera image to obtain processing results, and the side-looking sonar SLS processing module obtains polymetallic nodule data from the structured light system point cloud;

[0073] Step S20: The multi-dimensional nodule counting and size measurement module calculates the density and average size of polymetallic nodules in each sector based on the analysis results, the processing results, and the polymetallic nodule data. The machine learning module processes the analysis results, the processing results, the polymetallic nodule data, and additional information to obtain nodule collection parameters.

[0074] Step S30: The abundance mapping module obtains a nodule abundance map based on the density and average size of the polymetallic nodules, the nodule collection parameters, and the position of the mining vehicle.

[0075] Furthermore, the additional information includes mining vehicle travel speed, hoist belt speed, crusher feed height, slurry concentration and initial abundance estimation.

[0076] In summary, the present invention can bring the following beneficial effects: the present invention is for detecting and mapping the abundance change trend of local polymetallic nodule minerals in the deep-sea polymetallic nodule mining operation area, and a pre-mapping scan is performed by a work-level ROV or AUV, and a deep-sea polymetallic nodule mining vehicle performs mining operations while scanning and mapping nodule ores in real time. Real-time abundance detection is used to automatically control the walking path, speed and collection operations of the deep-sea mining vehicle, which plays a significant role in ensuring the output and stable recovery rate of the deep-sea polymetallic nodule mining system. By accurately detecting the real-time abundance, the walking speed of the mining vehicle is adjusted, so that the ore flow rate collected by the mining vehicle is always maintained at a constant set value, so that the slurry lifting pump works at an optimal efficiency period, saving energy consumption while eliminating a set of intermediate warehouse equipment.

[0077] It should be noted that, in this document, the terms "comprise," "include," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or intelligent terminal comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or intelligent terminal. In the absence of further limitations, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or intelligent terminal comprising the element.

[0078] Of course, those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware (such as a processor, controller, etc.) through a computer program. The program can be stored in a computer-readable storage medium that can be read by a computer. When the program is executed, it can include the processes in the above-described method embodiments. The computer-readable storage medium can be a memory, a magnetic disk, an optical disk, etc.

[0079] It should be understood that the application of the present invention is not limited to the above examples. For those skilled in the art, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A real-time abundance detection system for deep-sea polymetallic nodule minerals, characterized in that: The real-time abundance detection system for deep-sea polymetallic nodule minerals includes: a forward scanning sonar (FLS) acoustic backscatter level analysis module, a forward looking sonar (FLS) processing saliency detection and analysis module, an image processing module, a side looking sonar (SLS) processing module, a multi-dimensional nodule counting and size measurement module, a machine learning module, an abundance mapping module, and a machine sensor. The machine learning module is respectively connected to the forward scanning sonar FLS acoustic backscatter level analysis module, the forward looking sonar FLS processing prominence detection and analysis module, the image processing module, the side looking sonar SLS processing module, the machine sensor and abundance mapping module, and the multi-dimensional nodule counting and size measurement module is respectively connected to the forward scanning sonar FLS acoustic backscatter level analysis module, the forward looking sonar FLS processing prominence detection and analysis module, the image processing module, the side looking sonar SLS processing module and the abundance mapping module; The forward scanning sonar FLS acoustic backscatter level analysis module is used to comprehensively analyze the overall intensity of the signal returned by the FLS sonar, and input the analysis results information into the machine learning module and the multi-dimensional nodule counting and size measurement module. The result information is used to determine the threshold value of the peak detection in the sonar echo, and the corresponding threshold is used to reflect the reflection result of the polymetallic nodule; The forward-looking sonar FLS processing saliency detection and analysis module is used to input the detected and analyzed target signal into the machine learning module and the multi-dimensional nodule counting and size measurement module; The image processing module is used to process the image to obtain a processing result, and input the processing result into the machine learning module and the multi-dimensional tuberculosis counting and size measurement module; The side-looking sonar SLS processing module is used to input the data of polymetallic nodules in each scanning sector into the machine learning module and the multi-dimensional nodule counting and size measurement module; The machine learning module is specifically used to: obtaining a collected amount of polymetallic nodules and parameters measured by the slurry, and comparing the collected amount of polymetallic nodules and the parameters measured by the slurry to obtain a ratio; correlating the visual and sonar detection signals with actual mined ore output based on the ratio; The machine learning module is also used to: storing a sequence of sensed data and obtaining past measurements related to the slurry data based on the sequence of sensed data; The multi-dimensional nodule counting and size measurement module is used to calculate the density and average size of polymetallic nodules in each sector through probability estimation and data fusion methods; The abundance mapping module is configured to receive target data to determine an abundance detection value for each region and store data associating sensor fusion data with corresponding ratios; The target data includes visual analysis data for each area and a ratio between the visual data and the actual ore concentration.

2. The real-time abundance detection system for deep-sea polymetallic nodule minerals according to claim 1 is characterized in that: The resulting information includes global acoustic backscatter levels, backscatter intensity histograms for each sector of the seafloor cover, summed backscatter for highly reflective objects, seafloor background data for deep-sea polymetallic nodules, average and standard values of nodule intensity, and average water column backscatter levels associated with turbidity and sonar parameters.

3. The real-time abundance detection system for deep-sea polymetallic nodule minerals according to claim 1 is characterized in that: The target signal includes each region sector, a peak number and a geometric region descriptor; wherein the geometric region descriptor includes an average area and a geometric moment.

4. The real-time abundance detection system for deep-sea polymetallic nodule minerals according to claim 1, characterized in that: The processing results include: for each sector of the mining vehicle's field of view FOV, the count of polymetallic nodules detected in the sector and the average area of the geometric descriptors.

5. The real-time abundance detection system for deep-sea polymetallic nodule minerals according to claim 1 is characterized in that: The signals collected by the machine sensors include the speed of the mining vehicle, the speed of the conveyor belt, the currently measured slurry concentration, and an initial estimate of the abundance in each geometric region.

6. A method for real-time abundance detection of deep-sea polymetallic nodule minerals based on the real-time abundance detection system of deep-sea polymetallic nodule minerals according to any one of claims 1 to 5, characterized in that: The real-time abundance detection method for deep-sea polymetallic nodule minerals comprises: The forward-scan sonar FLS acoustic backscatter level analysis module and the forward-looking sonar FLS processing saliency detection and analysis module analyze the forward-scan sonar data to obtain analysis results, the image processing module processes the camera image to obtain processing results, and the side-looking sonar SLS processing module obtains polymetallic nodule data from the structured light system point cloud; The multi-dimensional nodule counting and size measurement module calculates the polymetallic nodule density and average size of each sector based on the analysis results, the processing results, and the polymetallic nodule data; and the machine learning module processes the analysis results, the processing results, the polymetallic nodule data, and additional information to obtain nodule collection parameters. The abundance mapping module obtains a nodule abundance map based on the density and average size of the polymetallic nodules, the nodule collection parameters, and the location of the mining vehicle.

7. The real-time abundance detection method for deep-sea polymetallic nodule minerals according to claim 6, characterized in that: The additional information includes mining vehicle travel speed, hoist belt speed, crusher feed height, slurry concentration and initial abundance estimate.

Images