An intelligent investigation system and method for mesopelagic fish resources

Abstract

The present invention discloses an intelligent investigation system and method for deep - sea mesopelagic fish resources. The method obtains topographic detection signal data of the target water area and fish resource detection signal data of the target water layer through a multibeam sonar device, and determines the distribution of fish schools in the target water layer accordingly. A three - dimensional topographic model of the target water area is constructed using the topographic detection signal data, and an underwater robot cruising strategy is determined in combination with the fish school distribution. The underwater robot obtains fish school image data according to this strategy, identifies the types and quantities of fish to obtain fish resource data, and finally transmits the data to the surface control center through underwater acoustic communication technology. The present invention realizes the intelligent and efficient investigation of deep - sea mesopelagic fish resources, provides strong support for the research and development of deep - sea fishery resources, and has broad application prospects and important practical significance.

Description

Technical Field

[0001] The present invention relates to the technical field of resource investigation, and particularly relates to an intelligent investigation system and method for deep-sea mesopelagic fish resources. Background Art

[0002] In the field of deep-sea resource research and development, the investigation of deep-sea mesopelagic fish resources is of great significance. The deep-sea mesopelagic zone is below the ocean's photic layer, with high water pressure, weak light, and a complex environment, containing rich fish resources. These fish play a key role in the material cycle and energy flow of the marine ecosystem and also have potential economic value.

[0003] Traditional methods for investigating deep-sea mesopelagic fish resources have many limitations. On the one hand, the equipment and technologies used are relatively backward, making it difficult to obtain high-precision and comprehensive terrain and fish resource detection signal data. For example, early sonar equipment had low resolution and could not accurately detect the distribution of fish schools, resulting in large errors in judging the number and species of fish schools. On the other hand, when determining the cruising strategy of underwater robots, there is a lack of comprehensive consideration of the terrain and fish school distribution, making the working efficiency of underwater robots low and unable to obtain fish school image data comprehensively and efficiently. In addition, in terms of identifying fish species and numbers, relying on manual identification or simple image analysis methods is not only time-consuming and laborious but also difficult to guarantee accuracy. In terms of data transmission, traditional communication methods have problems such as severe signal attenuation and low transmission rate, resulting in the inability to transmit the obtained fish resource data to the surface control center in a timely and accurate manner.

[0004] With the in-depth development of marine scientific research and the growing demand for marine resource development, there is an urgent need for an efficient, accurate, and intelligent method and system for investigating deep-sea mesopelagic fish resources to overcome the deficiencies of traditional methods, achieve a comprehensive and in-depth understanding of deep-sea mesopelagic fish resources, and provide strong support for marine ecological protection and the sustainable utilization of fishery resources. Based on this background, the present invention aims to solve the above problems in the prior art and provides a new intelligent investigation method and system for deep-sea mesopelagic fish resources. Summary of the Invention

[0005] To solve at least one of the above technical problems, the present invention proposes an intelligent investigation system and method for deep-sea mesopelagic fish resources.

[0006] In the first aspect of the present invention, an intelligent investigation method for deep-sea mesopelagic fish resources is provided, including:

[0007] Obtaining terrain detection signal data of a target water area and fish resource detection signal data of a target water layer based on a multi-beam sonar device, and determining the fish school distribution in the target water layer according to the fish resource detection signal data;

[0008] Construct a three-dimensional terrain model of the target water area based on the terrain detection signal data, and determine the cruising strategy of the underwater robot according to the three-dimensional terrain model and the fish school distribution;

[0009] The underwater robot obtains the fish school image data of the target water layer according to the cruising strategy, and identifies the types and quantities of fish in the target water layer based on the image data to obtain fish resource data;

[0010] Transmit the fish resource data to the surface control center according to the underwater acoustic communication technology.

[0011] In this solution, the terrain detection signal data of the target water area and the fish resource detection signal data of the target water layer are obtained based on the multi-beam sonar device. The fish school distribution in the target water layer is determined according to the fish resource detection signal data. Specifically:

[0012] Obtain the terrain detection data of the target water area and the fish resource detection signal data of the target water layer in the target water area based on the multi-beam sonar device;

[0013] Perform an adaptive filtering operation on the fish resource detection signal data based on the dynamic threshold segmentation algorithm to obtain the fish resource detection signal data with noise removed. Introduce the variational mode decomposition algorithm and set the initial values of the parameters of the variational mode decomposition algorithm;

[0014] Decompose the noise-removed fish resource detection signal data according to the variational mode decomposition algorithm to obtain k intrinsic mode functions;

[0015] Perform a Hilbert transform on the intrinsic mode functions, construct the analytic signal of each intrinsic mode function, calculate the instantaneous amplitude and instantaneous frequency of each intrinsic mode function according to the analytic signal, and construct a time-frequency distribution matrix with the instantaneous amplitude and instantaneous frequency;

[0016] Calculate the energy distribution, amplitude mean, total energy, and amplitude peak of each frequency band emitted by the multi-beam sonar device according to the time-frequency distribution matrix to obtain the fish school detection characteristic spectrum;

[0017] Obtain the measurement point position information of the fish school detection characteristic spectrum according to the fish resource detection signal data, map the fish school detection characteristic spectrum of each measurement point according to the measurement point position information, and construct a three-dimensional fish school detection characteristic spectrum;

[0018] Determine the fish school density of each measurement point according to the three-dimensional fish school detection characteristic spectrum, construct a fish school distribution heat map and a fish school density gradient field of the target water layer according to the fish school density of each measurement point, and determine the fish school distribution in the target water layer according to the fish school distribution heat map and the fish school density gradient field.

[0019] In this solution, the three-dimensional terrain model of the target water area is constructed based on the terrain detection signal data, and the cruising strategy of the underwater robot is determined according to the three-dimensional terrain model and the fish population distribution. Specifically:

[0020] Perform multi-scale curvature analysis on the terrain detection signal data, construct a terrain feature tensor through grid processing, and identify the bottom object type of the echo signal of the terrain detection signal data based on a support vector machine classifier to generate a three-dimensional terrain model of the target water area including terrain elevation, curvature features, and geological type labels;

[0021] Perform a position mapping operation on the fish population distribution and the three-dimensional terrain model to construct a three-dimensional space distribution model of fish population resources, and divide the three-dimensional space distribution model of fish population resources according to a preset grid size to construct N sub-space regions;

[0022] Obtain the fish population density information of each sub-space region according to the fish population distribution, and perform a clustering operation on the sub-space regions with similar fish population densities in the three-dimensional space distribution model of fish population resources based on the k-means clustering algorithm to obtain a clustering result;

[0023] Determine the distribution probability of the fish population in each sub-space region according to the clustering result, obtain the corresponding position information of each sub-space region in the target water area, and evaluate the importance of fish population detection for each position in the target water layer according to the distribution probability and the corresponding position information to obtain the detection importance score of each position in the target water layer;

[0024] Determine the detection requirement information of each position in the target water layer according to the detection importance score of each position. The detection requirement information includes whether to perform detection and detection accuracy requirement information;

[0025] Obtain the image acquisition clarity data of the underwater robot for fish populations at different distances, determine the image acquisition distance information of the underwater robot for each position in the target water layer according to the image acquisition clarity data and the detection requirement information, and determine the image acquisition point set of the underwater robot according to the image acquisition distance information;

[0026] Determine the cruising strategy of the underwater robot according to the image acquisition point set.

[0027] In this solution, the cruising strategy of the underwater robot is determined according to the image acquisition point set. Specifically:

[0028] Obtain the three-dimensional coordinate information of each image acquisition point and the initial position information of the underwater robot, and determine the terrain obstacles in the target water area according to the three-dimensional terrain model;

[0029] Perform path planning on the initial position information and the three-dimensional coordinate information of each image acquisition point based on the A* algorithm, and use the terrain obstacles as the path planning restricted area to output the shortest initial cruise path of the underwater robot;

[0030] Obtain the attitude change data of the underwater robot during the cruise according to the shortest initial cruise path in real time, and determine the water flow direction and water flow speed of the real-time cruise position of the underwater robot according to the attitude change data;

[0031] Obtain the driving stability data of the underwater robot, and the driving stability data includes the resistance ability data of the underwater robot to water flows in different directions and at different speeds;

[0032] Obtain the recognition reaction time data of the underwater robot to the water flow direction and water flow speed, and determine the deviation lag amount of the cruise path of the underwater robot according to the recognition reaction time data;

[0033] Analyze the water flow direction and water flow speed of the real-time cruise position of the underwater robot according to the resistance ability data and the deviation lag amount of the cruise path, and judge the cumulative deviation amount of the cruise path of the underwater robot within the recognition reaction time;

[0034] If the cumulative deviation amount of the cruise path is less than the preset value, determine the cruise deviation direction and cruise deviation distance of the underwater robot cruising according to the shortest initial cruise path according to the cumulative deviation amount of the cruise path, and determine the deviation correction direction and deviation correction distance of the underwater robot according to the cruise deviation direction and cruise deviation distance to obtain deviation correction data;

[0035] Perform a deviation correction operation on the shortest initial cruise path of the underwater robot during the real-time driving process according to the deviation correction data to obtain the first cruise strategy;

[0036] If the cumulative deviation amount of the cruise path is greater than the preset value, obtain the water flow speed and water flow direction data of the real-time driving path of the underwater robot, construct a water flow change map according to the water flow speed and water flow direction data of the real-time driving path, and determine the water flow change trend of the target water layer in the target water area according to the water flow change map;

[0037] Perform interpolation operation on the water flow change trend based on radial basis function interpolation, determine the water flow information within the preset range of the shortest initial cruise path, determine the driving stability of the underwater robot within the preset range of the shortest initial cruise path according to the water flow information and the resistance ability data, and adjust the paragraph of the shortest initial cruise path with the cumulative deviation amount of the cruise path greater than the preset value according to the driving stability to obtain an updated cruise path, and the underwater robot performs a cruise operation according to the updated cruise path to obtain the second cruise strategy.

[0038] In this solution, the underwater robot obtains the fish school image data of the target water layer according to the cruising strategy, and identifies the types and quantities of fish in the target water layer based on the image data to obtain fish resource data. Specifically:

[0039] Obtain the standard image data of different types of fish in the target water area, and label the names of the fish for the standard image data to obtain labeled image data;

[0040] Construct a fish recognition model based on a convolutional neural network, and import the labeled image data into the fish recognition model for training;

[0041] The underwater robot obtains the fish school image data of the target water layer according to the cruising strategy, imports the fish school image data into the trained fish recognition model for fish species recognition, and counts the quantity of each fish species to obtain fish resource data.

[0042] In this solution, the fish resource data is transmitted to the surface control center according to the underwater acoustic communication technology. Specifically:

[0043] Construct a hierarchical underwater acoustic communication protocol stack based on the underwater acoustic communication technology, modulate the fish resource data, and construct a fish resource acoustic transmission signal;

[0044] Send the fish resource acoustic transmission signal to the surface control center according to the hierarchical underwater acoustic communication protocol stack, and perform a decoding operation on the fish resource acoustic transmission signal to obtain the fish resource survey data of the target water layer.

[0045] The second aspect of the present invention also provides an intelligent investigation system for deep-sea mesopelagic fish resources. The system includes: a memory and a processor. The memory includes an intelligent investigation method program for deep-sea mesopelagic fish resources. When the intelligent investigation method program for deep-sea mesopelagic fish resources is executed by the processor, the following steps are implemented:

[0046] Obtain the terrain detection signal data of the target water area and the fish resource detection signal data of the target water layer based on a multi-beam sonar device, and determine the fish school distribution in the target water layer according to the fish resource detection signal data;

[0047] Construct a three-dimensional terrain model of the target water area according to the terrain detection signal data, and determine the cruising strategy of the underwater robot according to the three-dimensional terrain model and the fish school distribution;

[0048] The underwater robot obtains the fish school image data of the target water layer according to the cruising strategy, and identifies the types and quantities of fish in the target water layer based on the image data to obtain fish resource data;

[0049] Transmit the fish resource data to the surface control center according to the underwater acoustic wave communication technology.

[0050] The present invention discloses an intelligent investigation system and method for deep - sea mesopelagic fish resources. The method obtains the terrain detection signal data of the target water area and the fish resource detection signal data of the target water layer through a multi - beam sonar device, and accordingly determines the distribution of fish schools in the target water layer. A three - dimensional terrain model of the target water area is constructed using the terrain detection signal data, and the underwater robot cruising strategy is determined in combination with the fish school distribution. The underwater robot obtains fish school image data according to this strategy, identifies the types and quantities of fish to obtain fish resource data, and finally transmits the data to the surface control center through the underwater acoustic wave communication technology. The present invention realizes the intelligent and efficient investigation of deep - sea mesopelagic fish resources, provides strong support for the research and development of deep - sea fishery resources, and has broad application prospects and important practical significance. Brief Description of the Drawings

[0051] Figure 1 Shows the flowchart of an intelligent investigation method for deep - sea mesopelagic fish resources of the present invention;

[0052] Figure 2 Shows the flowchart of obtaining fish resource data of the present invention;

[0053] Figure 3 Shows the flowchart of transmitting fish resource data to the surface control center of the present invention;

[0054] Figure 4 Shows the block diagram of an intelligent investigation system for deep - sea mesopelagic fish resources of the present invention. Detailed Embodiments

[0055] In order to more clearly understand the above - mentioned objects, features and advantages of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0056] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0057] Figure 1 Shows the flowchart of an intelligent investigation method for deep - sea mesopelagic fish resources of the present invention.

[0058] As Figure 1 shown, the first aspect of the present invention provides an intelligent investigation method for deep - sea mesopelagic fish resources, including:

[0059] S102. Based on a multi-beam sonar device, obtain the topographic detection signal data of the target water area and the fish resource detection signal data of the target water layer, and determine the fish school distribution in the target water layer according to the fish resource detection signal data;

[0060] S104. Construct a three-dimensional topographic model of the target water area according to the topographic detection signal data, and determine the cruising strategy of the underwater robot according to the three-dimensional topographic model and the fish school distribution;

[0061] S106. The underwater robot obtains the fish school image data of the target water layer according to the cruising strategy, and identifies the fish species and quantity in the target water layer according to the image data to obtain the fish resource data;

[0062] S108. Transmit the fish resource data to the surface control center according to the underwater acoustic communication technology.

[0063] It should be noted that the multi-beam sonar device can simultaneously transmit and receive multiple beams, so as to quickly and comprehensively obtain the topographic detection signal data of the target water area and the fish resource detection signal data of the target water layer, improving the efficiency and coverage of data collection; determining the fish school distribution in the middle layer of the deep sea through the fish resource detection signal data can pre-determine the main distribution areas of the fish schools, providing targets for subsequent resource identification by the underwater robot, reducing resource waste, and improving the investigation efficiency; determining the cruising strategy of the underwater robot by combining the three-dimensional topographic model and the fish school distribution can enable the underwater robot to avoid topographic obstacles and at the same time detect the fish school dense areas more pertinently, improving the investigation efficiency and accuracy; the underwater robot can obtain high-quality fish school image data of the target water layer at a suitable position and angle according to the optimized cruising strategy, providing a clear image basis for the identification of fish species and quantity; using advanced image recognition technology to analyze the fish school image data can accurately identify the fish species and quantity in the target water layer to obtain detailed fish resource data; the underwater acoustic communication technology can realize the reliable transmission of fish resource data in the underwater environment, and transmit the underwater collected data to the surface control center in time to avoid data loss or delay.

[0064] According to an embodiment of the present invention, the obtaining the topographic detection signal data of the target water area and the fish resource detection signal data of the target water layer based on a multi-beam sonar device, and determining the fish school distribution in the target water layer according to the fish resource detection signal data is specifically:

[0065] Based on a multi-beam sonar device, obtain the topographic detection data of the target water area and the fish resource detection signal data of the target water layer in the target water area;

[0066] Perform adaptive filtering operation on the fish resource detection signal data based on the dynamic threshold segmentation algorithm to obtain the fish resource detection signal data with noise removed, introduce the variational mode decomposition algorithm, and set the initial values of the parameters of the variational mode decomposition algorithm;

[0067] Decompose the fish resource detection signal data with noise removed according to the variational mode decomposition algorithm to obtain k intrinsic mode functions;

[0068] Perform Hilbert transform on the intrinsic mode functions to construct the analytic signals of each intrinsic mode function, calculate the instantaneous amplitude and instantaneous frequency of each intrinsic mode function according to the analytic signals, and construct a time-frequency distribution matrix with the instantaneous amplitude and instantaneous frequency;

[0069] Calculate the energy distribution, amplitude mean, total energy, and amplitude peak of each frequency band emitted by the multibeam sonar device according to the time-frequency distribution matrix to obtain the fish school detection characteristic spectrum;

[0070] Obtain the measurement point position information of the fish school detection characteristic spectrum according to the fish resource detection signal data, map the fish school detection characteristic spectrum of each measurement point according to the measurement point position information, and construct a three-dimensional fish school detection characteristic spectrum;

[0071] Determine the fish school density of each measurement point according to the three-dimensional fish school detection characteristic spectrum, construct a fish school distribution heat map and a fish school density gradient field of the target water layer according to the fish school density of each measurement point, and determine the fish school distribution situation of the target water layer according to the fish school distribution heat map and the fish school density gradient field.

[0072] It should be noted that the variational mode decomposition algorithm (VMD) is an adaptive signal processing method that can decompose complex signals into multiple intrinsic mode functions (IMFs) with different center frequencies and bandwidths. In the deep-sea environment, the fish resource detection signal data received by the multibeam sonar device contains various noises and interferences. VMD can adaptively decompose the signal according to the inherent characteristics of the signal, thereby effectively extracting the feature information related to the fish school. By setting different decomposition levels k, IMF components of different scales can be obtained. Each IMF component contains the feature information of the signal in different frequency bands, so that the fish school detection signal can be analyzed from multiple angles, improving the accuracy and comprehensiveness of feature extraction. For example, the IMF component in the low-frequency band may reflect the overall distribution trend of the fish school, while the IMF component in the high-frequency band may contain the detailed features of the fish school. The three-dimensional fish school detection feature spectrum is constructed by performing position mapping on parameters such as the energy distribution, amplitude mean, total energy, and amplitude peak of each frequency band emitted by the multibeam sonar device. These parameters are closely related to the density of the fish school. For example, areas with larger energy distribution and amplitude peak usually indicate a denser fish school. Since the presence of the fish school will cause scattering and reflection of the sonar signal, resulting in changes in the energy and amplitude of the signal, the density of the fish school can be inferred by analyzing these changes. Using the dynamic threshold segmentation algorithm for adaptive filtering, combined with the variational mode decomposition algorithm, can effectively remove the complex noises and interferences in the deep-sea environment, deeply mine and analyze the fish resource detection signal data, extract an accurate and comprehensive fish school detection feature spectrum. Based on the three-dimensional fish school detection feature spectrum, using the correlation between features and density, spatial information, as well as statistical analysis and model verification, the fish school density at each measurement point can be accurately determined, and then a heat map of the fish school distribution and a fish school density gradient field of the target water layer can be constructed, clearly and intuitively showing the distribution of the fish school in the target water layer. The fish school distribution includes the fish school distribution boundary and the approximate distribution density of the fish school at each position. The parameters of the variational mode decomposition algorithm include the number of modal components, the maximum number of iterations, the convergence accuracy, the modal component value, the center frequency value, and the Lagrange multiplier value.

[0073] According to an embodiment of the present invention, constructing a three-dimensional terrain model of the target water area based on the terrain detection signal data, and determining the cruising strategy of the underwater robot according to the three-dimensional terrain model and the fish school distribution, specifically:

[0074] Perform multi-scale curvature analysis on the terrain detection signal data, construct a terrain feature tensor through grid processing, and identify the bottom object type of the echo signal of the terrain detection signal data based on a support vector machine classifier to generate a three-dimensional terrain model of the target water area including terrain elevation, curvature features, and geological type labels;

[0075] Perform a position mapping operation on the fish school distribution and the three-dimensional terrain model to construct a three-dimensional spatial distribution model of fish school resources. Divide the three-dimensional spatial distribution model of fish school resources according to a preset grid size to construct N sub-space regions;

[0076] Obtain the fish school density information of each sub-space region according to the fish school distribution. Based on the k-means clustering algorithm, perform a clustering operation on the sub-space regions with similar fish school densities in the three-dimensional spatial distribution model of fish school resources to obtain a clustering result;

[0077] Determine the distribution probability of the fish school in each sub-space region according to the clustering result. Obtain the corresponding position information of each sub-space region in the target water area. Evaluate the importance of fish school detection for each position in the target water layer according to the distribution probability and the corresponding position information to obtain the detection importance score of each position in the target water layer;

[0078] Determine the detection requirement information of each position in the target water layer according to the detection importance score of each position. The detection requirement information includes whether to perform detection and the detection accuracy requirement information;

[0079] Obtain the image acquisition clarity data of the underwater robot for fish schools at different distances. Determine the image acquisition distance information of the underwater robot for each position in the target water layer according to the image acquisition clarity data and the detection requirement information. Determine the image acquisition point set of the underwater robot according to the image acquisition distance information;

[0080] Determine the cruising strategy of the underwater robot according to the image acquisition point set.

[0081] It should be noted that by using the k-means clustering algorithm, it is possible to effectively cluster the subspace regions in the three-dimensional space distribution model of fish stock resources based on fish stock density. By determining the distribution probability of fish stocks in each subspace region through the clustering results, the dense and sparse regions of fish stocks can be clearly distinguished. This enables the investigation work to focus on the regions with high fish stock distribution probability, avoiding wasting excessive resources and time in regions with sparse fish stocks, and significantly improving the pertinence and efficiency of resource detection. Based on the distribution probability of fish stocks in the subspace region and the corresponding position information, the importance of fish stock detection is evaluated for each position in the target water layer, and the detection importance score is obtained. This process can quantify the detection value of different positions, determine the detection requirement information according to the score, and clarify which regions need to be detected with key focus, and which regions can appropriately reduce the detection accuracy or be ignored. In this way, the detection resources of the underwater robot can be reasonably allocated to ensure obtaining the most valuable fish stock information under limited energy and time conditions. By combining the detection importance score with the image acquisition clarity data of the underwater robot for fish stocks at different distances, the image acquisition distance information and the set of image acquisition points are determined. This ensures that during the image acquisition process, the underwater robot can take pictures of the fish stock regions of key concern at the most appropriate distance and position. On the one hand, it avoids blurry images caused by too far a distance, making it impossible to accurately identify the species and quantity of fish; on the other hand, it prevents missing other important regions due to too close a distance. Through precise planning of the image acquisition points, the quality of the acquired image data is improved, providing more reliable data support for subsequent fish stock resource identification and analysis. The higher the detection importance score, the higher the requirement for detection accuracy.

[0082] According to an embodiment of the present invention, determining the cruising strategy of the underwater robot according to the set of image acquisition points specifically includes:

[0083] Obtain the three-dimensional coordinate information of each image acquisition point and the initial position information of the underwater robot, and determine the terrain obstacles in the target water area according to the three-dimensional terrain model;

[0084] Based on the A* algorithm, perform path planning on the initial position information and the three-dimensional coordinate information of each image acquisition point, and use the terrain obstacles as the path planning restricted area to output the shortest initial cruising path of the underwater robot;

[0085] It should be noted that after determining the distribution of fish stocks, it is necessary to use an underwater robot equipped with a camera device to obtain fish stock images in the fish stock distribution area and then conduct an investigation on the species and quantity of fish stocks. By analyzing each image acquisition point through the A* algorithm and planning the cruising path with the shortest driving path, the energy consumption of the underwater robot can be greatly reduced and the investigation efficiency can be improved.

[0086] Obtain the attitude change data of the underwater robot during the cruise according to the shortest initial cruise path in real time, and determine the water flow direction and water flow speed of the real-time cruise position of the underwater robot according to the attitude change data;

[0087] Obtain the driving stability data of the underwater robot, where the driving stability data includes the resistance ability data of the underwater robot to water flows in different directions and at different speeds;

[0088] Obtain the recognition reaction time data of the underwater robot to the water flow direction and water flow speed, and determine the deviation lag amount of the cruise path of the underwater robot according to the recognition reaction time data;

[0089] Analyze the water flow direction and water flow speed of the real-time cruise position of the underwater robot according to the resistance ability data and the deviation lag amount of the cruise path, and judge the cumulative deviation amount of the cruise path of the underwater robot within the recognition reaction time;

[0090] If the cumulative deviation amount of the cruise path is less than the preset value, determine the cruise deviation direction and cruise deviation distance of the underwater robot cruising according to the shortest initial cruise path according to the cumulative deviation amount of the cruise path, and determine the deviation correction direction and deviation correction distance of the underwater robot according to the cruise deviation direction and cruise deviation distance to obtain deviation correction data;

[0091] Perform a deviation correction operation on the shortest initial cruise path of the underwater robot during the real-time driving process according to the deviation correction data to obtain the first cruise strategy;

[0092] It should be noted that in order to efficiently complete the task of collecting images of school of fish, the underwater robot needs to quickly reach each image collection point according to the planned path. However, in the actual navigation process, due to the influence of factors such as water flow, if the path cannot be adjusted in time, it may not be able to reach the point on time, or the quality of the image collected when arriving at the point will be affected, which cannot meet the needs of fish resource identification and analysis. When determining the amount of correction of the cruise path, the underwater robot needs to obtain its own posture change data, water flow direction and speed data in real time. The collection, transmission and processing of these data all take time. In a complex underwater environment, data transmission may be affected by signal interference, attenuation, etc., resulting in data transmission delays. In the data processing process, such as the solution of posture data, the analysis of water flow data, and the calculation of correction amount based on these data, a certain amount of calculation time is required. This series of data-related operations together cause correction lag. When the accumulated amount of cruise path deviation is less than the preset value, the water flow direction and speed are determined by obtaining the underwater robot posture change data, and the cruise deviation direction and distance can be accurately calculated by combining the driving stability data and the cruise path correction lag, and then the correction direction and distance can be determined. This allows the underwater robot to make timely fine-tuning of the shortest initial cruising path with small deviations, and always keep on a trajectory close to the predetermined path.

[0093] If the accumulated amount of the cruise path deviation is greater than a preset value, the water flow velocity and water flow direction data of the real-time driving path of the underwater robot are obtained, a water flow change diagram is constructed according to the water flow velocity and water flow direction data of the real-time driving path, and the water flow change trend of the target water layer in the target water area is determined according to the water flow change diagram;

[0094] An interpolation operation is performed on the water flow change trend based on radial basis function interpolation to determine the water flow information within the preset range of the shortest initial cruising path, and the driving stability of the underwater robot within the preset range of the shortest initial cruising path is determined according to the water flow information and resistance capacity data. According to the driving stability, the shortest initial cruising path section with a cumulative amount of cruise path deviation greater than a preset value is adjusted to obtain an updated cruise path, and the underwater robot performs a cruise operation according to the updated cruise path to obtain a second cruise strategy.

[0095] It should be noted that when the cumulative offset of the cruise path is greater than the preset value, it indicates that the underwater robot encounters strong water flow interference during navigation. The complex and variable water flow direction and high-speed water flow cause continuous external forces on the robot, making it unable to maintain the predetermined cruise path. Therefore, by obtaining the water flow speed and water flow direction data of the real-time driving path of the underwater robot and constructing a water flow change map, the water flow change situation of the target water layer in the target water area can be presented intuitively and comprehensively. By using radial basis function interpolation to interpolate the water flow change trend, more detailed and accurate water flow information can be obtained within the shortest initial cruise path preset range. This refined processing can fill the data gap and optimize the continuity and integrity of the water flow information, so as to more accurately evaluate the impact of the water flow on the driving stability of the underwater robot. By adjusting the shortest initial cruise path segment with the cumulative offset of the cruise path greater than the preset value according to the driving stability, the unreasonable path planning can be corrected in time, so that the updated cruise path can better adapt to the complex and variable water flow environment. The water flow information includes water flow speed and water flow direction.

[0096] Figure 2 The flowchart of obtaining fish resource data according to the present invention is shown.

[0097] According to an embodiment of the present invention, the underwater robot obtains fish school image data of the target water layer according to the cruise strategy, and identifies the fish species and quantity of the target water layer according to the image data to obtain fish resource data. Specifically:

[0098] S202, obtain standard image data of different species of fish in the target water area, and label the fish names of the standard image data to obtain labeled image data;

[0099] S204, construct a fish recognition model based on a convolutional neural network, and import the labeled image data into the fish recognition model for training;

[0100] S206, the underwater robot obtains fish school image data of the target water layer according to the cruise strategy, imports the fish school image data into the trained fish recognition model for fish species recognition, and counts the quantity of each fish species to obtain fish resource data.

[0101] It should be noted that the underwater robot works according to the established cruising strategy, and can comprehensively scan the target water layer in a targeted manner to accurately obtain fish school image data. This avoids blind collection, greatly improves the data collection efficiency, and at the same time ensures that the collected data is representative. The fish recognition model constructed based on the convolutional neural network can accurately identify different species of fish through learning and training on a large number of labeled image data. The powerful feature extraction ability of the convolutional neural network enables it to capture the subtle features of fish, thereby effectively distinguishing similar species of fish, reducing recognition errors, and improving the accuracy of fish species recognition. By counting the number of each fish species, the distribution of fish resources in the target water layer can be comprehensively grasped. It can not only understand the population numbers of different fish, but also analyze the proportional relationship between them, providing key data for fishery resource assessment.

[0102] Figure 3 The flowchart showing the transmission of fish resource data of the present invention to the surface control center is shown.

[0103] According to an embodiment of the present invention, the transmission of the fish resource data to the surface control center according to the underwater acoustic communication technology is specifically as follows:

[0104] S302, construct a hierarchical underwater acoustic communication protocol stack based on the underwater acoustic communication technology, modulate the fish resource data, and construct a fish resource acoustic transmission signal;

[0105] S304, send the fish resource acoustic transmission signal to the surface control center according to the hierarchical underwater acoustic communication protocol stack, and perform a decoding operation on the fish resource acoustic transmission signal to obtain the fish resource survey data of the target water layer.

[0106] It should be noted that constructing a hierarchical underwater acoustic communication protocol stack based on the underwater acoustic communication technology can adapt to complex underwater environments, effectively resist interference factors such as seawater absorption, scattering, and multipath effects, ensure the stability of the fish resource acoustic transmission signal during the transmission process, reduce data loss and errors, and ensure that the fish resource data is accurately transmitted from underwater to the surface control center. Modulating the fish resource data to construct an acoustic transmission signal and using the hierarchical underwater acoustic communication protocol stack for sending and decoding optimize the data transmission process and improve the transmission efficiency. The modulation process can convert the data into a form suitable for underwater acoustic transmission, and the protocol stack standardizes the data transmission rules, enabling the data to quickly reach the surface control center and be quickly decoded and processed, saving time costs and meeting real-time requirements.

[0107] Figure 4 The block diagram showing an intelligent investigation system for deep-sea mesopelagic fish resources of the present invention is shown.

[0108] In a second aspect of the present invention, an intelligent investigation system 4 for deep-sea mesopelagic fish resources is further provided. The system includes: a memory 41 and a processor 42. The memory includes an intelligent investigation method program for deep-sea mesopelagic fish resources. When the intelligent investigation method program for deep-sea mesopelagic fish resources is executed by the processor, the following steps are implemented:

[0109] Based on a multi-beam sonar device, obtain topographic detection signal data of a target water area and fish resource detection signal data of a target water layer, and determine the fish school distribution in the target water layer according to the fish resource detection signal data;

[0110] Construct a three-dimensional topographic model of the target water area according to the topographic detection signal data, and determine the cruising strategy of the underwater robot according to the three-dimensional topographic model and the fish school distribution;

[0111] The underwater robot obtains fish school image data of the target water layer according to the cruising strategy, and identifies the fish species and quantity in the target water layer according to the image data to obtain fish resource data;

[0112] Transmit the fish resource data to the surface control center according to underwater acoustic communication technology.

[0113] The present invention discloses an intelligent investigation system and method for deep-sea mesopelagic fish resources. The method obtains topographic detection signal data of a target water area and fish resource detection signal data of a target water layer through a multi-beam sonar device, and accordingly determines the fish school distribution in the target water layer. A three-dimensional topographic model of the target water area is constructed by using the topographic detection signal data, and the cruising strategy of the underwater robot is determined in combination with the fish school distribution. The underwater robot obtains fish school image data according to this strategy, identifies the fish species and quantity to obtain fish resource data, and finally transmits the data to the surface control center through underwater acoustic communication technology. The present invention realizes the intelligent and efficient investigation of deep-sea mesopelagic fish resources, provides strong support for the research and development of deep-sea fishery resources, and has broad application prospects and important practical significance.

[0114] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed with each other can be through some interfaces. The indirect coupling or communication connection of the devices or units can be electrical, mechanical, or other forms.

[0115] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0116] In addition, each functional unit in the embodiments of the present invention may all be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0117] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including the above method embodiments; and the foregoing storage medium includes: removable storage devices, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks and other various media that can store program codes.

[0118] Alternatively, if the above-mentioned integrated units of the present invention are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention essentially or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. And the foregoing storage medium includes: removable storage devices, ROM, RAM, magnetic disks or optical disks and other various media that can store program codes.

[0119] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, and all of them should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An intelligent survey method for deep-sea mesopelagic fish resources, characterized in that: The following steps are involved: Acquire terrain detection signal data of the target water area and fish resource detection signal data of the target water layer based on the multi-beam sonar equipment, and determine the distribution of fish schools in the target water layer according to the fish resource detection signal data; Constructing a three-dimensional terrain model of the target waters according to the terrain detection signal data, and determining a cruising strategy of the underwater robot according to the three-dimensional terrain model and the distribution of the fish school; The underwater robot acquires fish school image data of the target water layer according to the cruising strategy, identifies the fish species and quantity of the target water layer according to the image data, and obtains fish resource data; Transmitting the fish resource data to a surface control center according to underwater acoustic wave communication technology; The method of obtaining terrain detection signal data of the target water area and fish resource detection signal data of the target water layer based on the multi-beam sonar device, and determining the fish distribution of the target water layer according to the fish resource detection signal data, is specifically as follows: Acquire terrain detection data of target waters and fish resource detection signal data of target water layers in target waters based on multi-beam sonar equipment; Based on the dynamic threshold segmentation algorithm, an adaptive filtering operation is performed on the fish resource detection signal data to obtain the fish resource detection signal data with noise removed, a variational mode decomposition algorithm is introduced, and initial values ​​of the variational mode decomposition algorithm parameters are set; Performing signal decomposition on the fish resource detection signal data after noise removal according to the variational mode decomposition algorithm to obtain k intrinsic mode functions; Performing Hilbert transform on the intrinsic mode function, constructing an analytical signal of each intrinsic mode function, calculating the instantaneous amplitude and instantaneous frequency of each intrinsic mode function according to the analytical signal, and constructing a time-frequency distribution matrix of the instantaneous amplitude and instantaneous frequency; Calculate the energy distribution, amplitude mean, energy sum, and amplitude peak of each frequency band emitted by the multi-beam sonar device according to the time-frequency distribution matrix to obtain a fish detection characteristic spectrum; Acquire the measurement point position information of the fish detection characteristic spectrum according to the fish resource detection signal data, perform position mapping of the fish detection characteristic spectrum of each measurement point according to the measurement point position information, and construct a three-dimensional fish detection characteristic spectrum; The fish density at each measuring point is determined according to the three-dimensional fish detection characteristic spectrum, a fish distribution thermodynamic map and a fish density gradient field of the target water layer are constructed according to the fish density at each measuring point, and the fish distribution situation of the target water layer is determined according to the fish distribution thermodynamic map and the fish density gradient field.

2. The intelligent survey method for deep-sea mesopelagic fish resources according to claim 1 is characterized in that: The three-dimensional terrain model of the target waters is constructed according to the terrain detection signal data, and the cruising strategy of the underwater robot is determined according to the three-dimensional terrain model and the distribution of the fish school, specifically: Performing multi-scale curvature analysis on the terrain detection signal data, constructing a terrain feature tensor through gridding processing, identifying the water bottom type of the echo signal of the terrain detection signal data based on a support vector machine classifier, and generating a three-dimensional terrain model of the target water area including terrain elevation, curvature characteristics and geological type labels; Performing a position mapping operation on the fish distribution situation and the three-dimensional terrain model to construct a three-dimensional spatial distribution model of fish resources, and dividing the three-dimensional spatial distribution model of fish resources according to a preset grid size to construct N subspace areas; According to the fish distribution, fish density information of each subspace area is obtained, and subspace areas with similar fish density in the three-dimensional spatial distribution model of fish resources are clustered based on the k-means clustering algorithm to obtain a clustering result; Determine the distribution probability of fish schools in each subspace area according to the clustering result, obtain the corresponding position information of each subspace area in the target water area, and evaluate the importance of fish school detection for each position of the target water layer in the target water area according to the distribution probability and the corresponding position information to obtain the detection importance score of each position in the target water layer; Determining detection requirement information of each position in the target water layer according to the detection importance score of each position, wherein the detection requirement information includes whether to perform detection and detection accuracy requirement information; Obtaining image acquisition clarity data of the underwater robot for fish schools at different distances, determining image acquisition distance information of the underwater robot for each position of the target water layer according to the image acquisition clarity data and detection requirement information, and determining an image acquisition point set of the underwater robot according to the image acquisition distance information; The cruising strategy of the underwater robot is determined according to the image acquisition point set.

3. The intelligent survey method for deep-sea mesopelagic fish resources according to claim 2 is characterized in that: The cruise strategy of the underwater robot is determined according to the image acquisition point set, specifically: Obtaining the three-dimensional coordinate information of each image acquisition point and the initial position information of the underwater robot, and determining the terrain obstacles in the target waters according to the three-dimensional terrain model; Perform path planning on the initial position information and the three-dimensional coordinate information of each image acquisition point based on the A* algorithm, and use the terrain obstacles as the path planning restriction area to output the shortest initial cruising path of the underwater robot; Acquire in real time the underwater robot's attitude change data during the underwater robot's cruising according to the shortest initial cruising path, and determine the water flow direction and water flow speed at the underwater robot's real-time cruising position according to the attitude change data; Acquiring driving stability data of the underwater robot, wherein the driving stability data includes data on the underwater robot's resistance to water currents of different directions and different speeds; Acquire recognition reaction time data of the underwater robot to the water flow direction and water flow speed, and determine the cruise path deviation correction lag of the underwater robot according to the recognition reaction time data; Analyze the water flow direction and water flow speed at the real-time cruising position of the underwater robot according to the resistance data and the cruising path deviation correction hysteresis, and determine the accumulated amount of the cruising path deviation of the underwater robot within the recognition reaction time; If the accumulated amount of the cruise path deviation is less than a preset value, determining a cruise deviation direction and a cruise deviation distance of the underwater robot according to the shortest initial cruise path according to the accumulated amount of the cruise path deviation, determining a deviation correction direction and a deviation correction distance of the underwater robot according to the cruise deviation direction and the cruise deviation distance, and obtaining deviation correction data; According to the deviation correction data, a deviation correction operation is performed on the shortest initial cruising path of the underwater robot during real-time driving to obtain a first cruising strategy; If the accumulated amount of the cruise path deviation is greater than a preset value, the water flow velocity and water flow direction data of the real-time driving path of the underwater robot are obtained, a water flow change diagram is constructed according to the water flow velocity and water flow direction data of the real-time driving path, and the water flow change trend of the target water layer in the target water area is determined according to the water flow change diagram; An interpolation operation is performed on the water flow change trend based on radial basis function interpolation to determine the water flow information within the preset range of the shortest initial cruising path, and the driving stability of the underwater robot within the preset range of the shortest initial cruising path is determined according to the water flow information and resistance capacity data. According to the driving stability, the shortest initial cruising path section with a cumulative amount of cruise path deviation greater than a preset value is adjusted to obtain an updated cruise path, and the underwater robot performs a cruise operation according to the updated cruise path to obtain a second cruise strategy.

4. The intelligent survey method for deep-sea mesopelagic fish resources according to claim 1 is characterized in that: The underwater robot obtains fish school image data of the target water layer according to the cruising strategy, identifies the fish species and quantity of the target water layer according to the image data, and obtains fish resource data, specifically: Obtain standard image data of different types of fish in the target waters, and label the standard image data with fish names to obtain labeled image data; Constructing a fish recognition model based on a convolutional neural network, and importing the labeled image data into the fish recognition model for training; The underwater robot obtains fish school image data of the target water layer according to the cruising strategy, imports the fish school image data into the trained fish recognition model to identify the fish species, and counts the number of each fish species to obtain fish resource data.

5. The intelligent survey method for deep-sea mesopelagic fish resources according to claim 1 is characterized in that: The fish resource data is transmitted to the surface control center according to the underwater acoustic wave communication technology, specifically: Building a layered underwater acoustic communication protocol stack based on underwater acoustic wave communication technology, modulating the fish resource data, and building a fish resource acoustic wave transmission signal; The fish resource acoustic wave transmission signal is acoustically transmitted to a water surface control center according to the layered underwater acoustic communication protocol stack, and the fish resource acoustic wave transmission signal is decoded to obtain fish resource survey data of the target water layer.

6. An intelligent deep-sea mid-layer fish resource survey system, characterized in that: The deep-sea mesopelagic fish resource intelligent survey system comprises a storage device and a processor, wherein the storage device comprises a deep-sea mesopelagic fish resource intelligent survey method program, and when the deep-sea mesopelagic fish resource intelligent survey method program is executed by the processor, the following steps are implemented: Acquire terrain detection signal data of the target water area and fish resource detection signal data of the target water layer based on the multi-beam sonar equipment, and determine the distribution of fish schools in the target water layer according to the fish resource detection signal data; Constructing a three-dimensional terrain model of the target waters according to the terrain detection signal data, and determining a cruising strategy of the underwater robot according to the three-dimensional terrain model and the distribution of the fish school; The underwater robot acquires fish school image data of the target water layer according to the cruising strategy, identifies the fish species and quantity of the target water layer according to the image data, and obtains fish resource data; Transmitting the fish resource data to a surface control center according to underwater acoustic wave communication technology; The method of obtaining terrain detection signal data of the target water area and fish resource detection signal data of the target water layer based on the multi-beam sonar device, and determining the fish distribution of the target water layer according to the fish resource detection signal data, is specifically as follows: Acquire terrain detection data of target waters and fish resource detection signal data of target water layers in target waters based on multi-beam sonar equipment; Based on the dynamic threshold segmentation algorithm, an adaptive filtering operation is performed on the fish resource detection signal data to obtain the fish resource detection signal data with noise removed, a variational mode decomposition algorithm is introduced, and initial values ​​of the variational mode decomposition algorithm parameters are set; Performing signal decomposition on the fish resource detection signal data after noise removal according to the variational mode decomposition algorithm to obtain k intrinsic mode functions; Performing Hilbert transform on the intrinsic mode function, constructing an analytical signal of each intrinsic mode function, calculating the instantaneous amplitude and instantaneous frequency of each intrinsic mode function according to the analytical signal, and constructing a time-frequency distribution matrix of the instantaneous amplitude and instantaneous frequency; Calculate the energy distribution, amplitude mean, energy sum, and amplitude peak of each frequency band emitted by the multi-beam sonar device according to the time-frequency distribution matrix to obtain a fish detection characteristic spectrum; Acquire the measurement point position information of the fish detection characteristic spectrum according to the fish resource detection signal data, perform position mapping of the fish detection characteristic spectrum of each measurement point according to the measurement point position information, and construct a three-dimensional fish detection characteristic spectrum; The fish density at each measuring point is determined according to the three-dimensional fish detection characteristic spectrum, a fish distribution thermodynamic map and a fish density gradient field of the target water layer are constructed according to the fish density at each measuring point, and the fish distribution situation of the target water layer is determined according to the fish distribution thermodynamic map and the fish density gradient field.

Images