A highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion

Abstract

The invention discloses a high-current-resistance multi-point observation fish school monitoring underwater robot based on multi-source data fusion, comprising a shell, an adaptive ocean current resistance system and an energy system; the adaptive ocean current resistance system comprises an ocean current information detection module and a PID control algorithm real-time control thruster module; the ocean current information detection module is used to collect ocean current information in real time, including water flow speed, direction and turbulence intensity; the control algorithm real-time control thruster module uses the PID control algorithm to control the power and direction of the thruster in real time according to the changes in the ocean current, so as to ensure that the underwater robot maintains stability in complex ocean currents; the telescopic multi-sensor fusion system collects multi-dimensional data of fish schools by combining multiple sensors, and uses a deep learning algorithm to fuse and analyze these data, thereby realizing high-precision fish school identification and tracking; the integrated ocean current-solar energy system enables the underwater robot to have a long-term endurance capability.

Description

Technical Field

[0001] The invention belongs to the technical field of underwater robots, and in particular relates to a high-current-resistance multi-point observation fish school monitoring underwater robot based on multi-source data fusion. Background Art

[0002] With the rapid development of aquaculture and marine ecological protection, the monitoring and tracking of fish schools has become a key task. However, traditional monitoring methods rely on manual operation or static equipment, which is difficult to meet the needs of efficient monitoring in complex offshore environments. Especially in the case of complex offshore currents, rapid movement of fish schools and overlapping fish individuals, traditional underwater monitoring equipment is prone to lose stability, insufficient endurance, and low image recognition accuracy, and cannot provide real-time high-quality data.

[0003] Deficiencies and reasons of existing technologies: Traditional underwater robots are easily affected by water currents in complex ocean current environments in the open sea, and the equipment is not stable enough, resulting in problems such as data collection distortion. This poses a threat to the continuity of fish monitoring and research and the accuracy of data. The image recognition function of fish schools is insufficient, image recognition is difficult, and the rapid movement of fish schools and the overlap between individuals pose a huge challenge to traditional image recognition algorithms, resulting in low recognition accuracy, large errors, and difficulty in effectively tracking the individual behavior of fish schools. Short battery life. Existing underwater robots rely on limited battery power and have short battery life, which cannot meet long-term monitoring needs, especially when conducting continuous fish status tracking in remote waters. This problem is particularly prominent. Summary of the invention

[0004] The present invention provides a highly current-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion. The robot can remain stable under complex ocean current conditions, has efficient image recognition capabilities, and can accurately track fast-swimming fish schools.

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

[0006] An embodiment of the present invention provides a highly current-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion, comprising a shell (1), an adaptive ocean current resistance system (2) being arranged at the front end position of the top of the shell (1), a telescopic multi-sensor fusion system (3) being arranged at the front end position of the bottom of the shell (1), an energy system (4) being arranged on the shell (1), and four vector thrusters (201) being arranged around the shell (1);

[0007] The adaptive ocean current resistance system (2) comprises an ocean current information detection module and a PID control algorithm real-time control thruster module; the ocean current information detection module is used to collect ocean current information in real time, including water flow speed, direction and turbulence intensity; the control algorithm real-time control thruster module uses the PID control algorithm to control the power and direction of the vector thruster (201) in real time according to the changes in the ocean current, so as to ensure that the underwater robot maintains stability in complex ocean currents;

[0008] The telescopic multi-sensor fusion system (3) comprises an environment monitoring and data acquisition module, a multi-source data fusion and overlap processing module, a fish school identification and trajectory prediction module, an information output module and an energy management module; the telescopic multi-sensor fusion system (3) collects multi-dimensional data of fish schools by combining multiple sensors, and uses a deep learning algorithm to perform fusion analysis on these data, thereby achieving efficient fish school identification and tracking.

[0009] In an optional embodiment of the present invention, the environment monitoring and data acquisition module includes sensor data acquisition and data preprocessing; the data acquisition sensor includes a high-resolution camera (301), an underwater sonar (302), an underwater lighting system (303) and an ocean current sensor (304); the data preprocessing is used for denoising, resolution adjustment and data enhancement processing.

[0010] In an optional embodiment of the present invention, the multi-source data fusion and overlap processing module includes feature extraction, sensor data fusion and overlap elimination technology; the feature extraction extracts features of the fish school image, such as edge and shape features, through a convolutional neural network CNN; the sensor data fusion fuses multi-sensor data of optics and sonar to obtain complete fish school information; the overlap elimination technology adopts deep learning processing, including convolutional layers, pooling layers, fully connected layers and classifier Softmax, and uses deep learning algorithms to classify and track individual fish schools.

[0011] In an optional embodiment of the present invention, the fish school identification and trajectory prediction module includes fish school individual identification and fish school trajectory prediction and tracking; the fish school individual identification utilizes a deep learning algorithm to classify and track fish school individuals, and the fish school trajectory prediction and tracking predicts the movement trajectory of the fish school based on historical data and a deep learning model.

[0012] In an optional embodiment of the present invention, the energy management module utilizes solar energy, water flow power generation and intelligent energy scheduling to manage the intelligent allocation of different energy sources to ensure long-term operation.

[0013] In an optional embodiment of the present invention, the adaptive ocean current resistance system (2) comprises a water flow sensor, a posture sensor and an acceleration sensor, which are used to monitor the water flow speed and direction in real time.

[0014] In an optional embodiment of the present invention, the PID control algorithm control equation is:

[0015] ;

[0016] Wherein, e(t) represents the error between the actual position and the expected position. The adaptive current resistance system (2) adjusts the power of the propeller through proportion, integration and differentiation, and adjusts the posture of the underwater robot in real time; parameter adjustment range: proportional gain K p The value range is 0.1~1.0, which is used to control the response speed; integral gain K i The value range is 0.01~0.1, which is used to eliminate steady-state error; differential gain K d The value range is 0.01~0.1, which is used to improve the anti-disturbance ability; Applicable conditions: The PID control algorithm is suitable for complex ocean current environments with water flow velocity ranging from 0.5 to 3.0 m / s, and adjusts the thrust output of the thruster in real time to ensure that the underwater robot maintains posture stability and navigation accuracy under sudden changes in ocean currents.

[0017] In an optional embodiment of the present invention, the energy system (4) comprises a solar panel (401) and a water flow power generation device (402), wherein the solar panel (401) is arranged at a middle position on the top of the housing (1) for capturing solar energy; and the water flow power generation device (402) is arranged at a rear end position on the bottom of the housing (1), and the water flow power generation device (402) uses water flow to drive a turbine to generate electricity, thereby providing additional power support.

[0018] In an optional embodiment of the present invention, the water flow power generation device (402) uses a water flow power generation efficiency formula to evaluate the power generation performance of the water flow turbine: ; Where ρ is the density of water, A is the windward area of ​​the turbine, and V is the water flow velocity.

[0019] Compared with the prior art, the embodiment of the present invention provides a highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion, which has the following beneficial effects:

[0020] (1) Fish School Monitoring The underwater robot is specially used for fish school monitoring and tracking tasks. Through the vector thruster adaptive ocean current resistance system and multi-sensor fusion technology, it can maintain stability in complex water currents. At the same time, it uses a cluster underwater robot based on a deep learning algorithm for fish school identification, which greatly improves the accuracy and efficiency of fish tracking. In addition, the integrated ocean current-solar energy system enables the underwater robot to have a long endurance, which is suitable for long-distance and long-term operations.

[0021] (2) Stability optimization and resistance to complex ocean currents. The combination of vector thrusters and ocean current detection sensors enables the underwater robot to sense ocean current changes in real time, and adjust the propulsion force and direction through the PID control algorithm to maintain the stability of the underwater robot. This stable control system ensures that efficient fish tracking and data collection can still be carried out in an environment with drastic changes in ocean currents. Direct effect: By introducing an adaptive ocean current resistance system, the underwater robot can adjust the output and direction of the vector thruster thruster in real time, so that it can maintain a high degree of stability under complex offshore current conditions, significantly reduce the loss of control caused by water current fluctuations, ensure continuous monitoring, and make it possible to track fish schools in strong ocean currents and deep sea environments, significantly improving the availability of ROVs, adapting to a wider range of ocean conditions, and expanding its scope of application.

[0022] (3) Efficient fish identification and tracking. The cluster multi-angle shooting and deep learning model used are trained with a large number of fish pictures, and can extract fish body features for accurate identification when fish overlap and swim quickly. At the same time, the fusion of acoustic sensor and optical sensor data further improves the accuracy of identification. Direct effect: The fish identification technology based on deep learning and multi-sensor fusion greatly improves the accuracy of identifying and tracking individual fish. Even in the case of overlapping and high-speed swimming of fish, it can effectively distinguish individual fish and provide accurate data. Technical effect: This improvement reduces the recognition errors caused by overlap or rapid movement in traditional image recognition, enhances the operating efficiency of underwater robots in complex fish state, and improves the scientificity and accuracy of fish monitoring and research.

[0023] (4) Energy recovery and endurance are improved. The water flow power generation device uses the kinetic energy of the water to drive the turbine to generate electricity. At the same time, the solar panel absorbs light energy when operating on the water surface. The combination of these two energy recovery methods ensures that the equipment maintains power supply during long-term operation, greatly extending the endurance time. Direct effect: By integrating water flow power generation and solar energy technology, the underwater robot of the present invention greatly extends the endurance time. It does not need to frequently replace batteries or charge during long-term operation, and is suitable for continuous monitoring needs. Technical effect: Compared with traditional battery-powered underwater robots, the sustainable operation capability of the equipment is improved, making long-term monitoring of remote areas such as the open sea and deep sea possible. This design not only improves environmental friendliness, but also reduces operation and maintenance costs. It has broad market prospects and competitive advantages in the fields of marine ecological protection and resource exploration. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0025] Figure 1 A schematic diagram of the structure of a highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion provided in an embodiment of the present application.

[0026] Figure 2 A schematic diagram of the bottom structure of a highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion provided in an embodiment of the present application.

[0027] Figure 3 A front view of a highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion provided in an embodiment of the present application.

[0028] Figure 4 A schematic diagram of an adaptive ocean current resistance system and a telescopic multi-sensor fusion system of a highly current-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0030] like Figure 1 As shown, an embodiment of the present invention provides a highly current-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion, comprising a shell 1, an adaptive ocean current resistance system 2 is arranged at the front end position of the top of the shell 1, a telescopic multi-sensor fusion system 3 is arranged at the front end position of the bottom of the shell 1, an energy system 4 is arranged on the shell 1, and four vector thrusters 201 are arranged around the shell 1.

[0031] like Figure 2 The adaptive ocean current resistance system 2 includes a water flow sensor, a posture sensor and an acceleration sensor, which are used to monitor the water flow speed and direction in real time. Figure 3The sensors for collecting data of the telescopic multi-sensor fusion system 3 include a high-resolution camera 301, an underwater sonar 302, an underwater lighting system 303 and an ocean current sensor 304. The model trained by the deep learning algorithm is used to identify fish in real time, and the sensor data is fused with the visual information to improve the recognition accuracy.

[0032] like Figure 4 As shown, the adaptive ocean current resistance system 2 includes an ocean current information detection module and a PID control algorithm real-time control thruster module; the ocean current information detection module is used to collect ocean current information in real time, including water flow speed, direction and turbulence intensity. The control algorithm real-time control thruster module uses the PID control algorithm to control the power and direction of the vector thruster 201 in real time according to the changes in the ocean current, ensuring that the underwater robot maintains stability in complex ocean currents.

[0033] The control equation of the PID control algorithm is:

[0034] ;

[0035] Where e(t) represents the error between the actual position and the expected position. The adaptive current resistance system (2) adjusts the propeller power through proportional, integral and differential, and adjusts the posture of the underwater robot in real time. Parameter adjustment range: proportional gain K p The value range is 0.1~1.0, which is used to control the response speed; integral gain K i The value range is 0.01~0.1, which is used to eliminate steady-state error; differential gain K d The value range is 0.01~0.1, which is used to improve the anti-disturbance ability; Applicable conditions: The PID control algorithm is suitable for complex ocean current environments with water flow velocity ranging from 0.5 to 3.0 m / s, and adjusts the thrust output of the thruster in real time to ensure that the underwater robot maintains posture stability and navigation accuracy under sudden changes in ocean currents.

[0036] like Figure 4 As shown. The telescopic multi-sensor fusion system 3 includes an environment monitoring and data acquisition module, a multi-source data fusion and overlap processing module, a fish school identification and trajectory prediction module, an information output module, and an energy management module. The environment monitoring and data acquisition module includes sensor data acquisition and data preprocessing. The data acquisition sensor includes a high-resolution camera 301, an underwater sonar 302, an underwater lighting system 303, and an ocean current sensor 304. Data preprocessing is used for denoising, resolution adjustment, and data enhancement processing.

[0037] The multi-source data fusion and overlap processing module includes feature extraction, sensor data fusion, and overlap elimination technology; feature extraction uses convolutional neural network CNN to extract features of fish school images, such as edge and shape features. Sensor data fusion fuses multi-sensor data such as optics and sonar to obtain complete fish school information; overlap elimination technology uses deep learning processing, including convolution layer, pooling layer, fully connected layer and classifier Softmax, and uses deep learning algorithms to classify and track individual fish schools.

[0038] The fish identification and trajectory prediction module includes individual fish identification and fish trajectory prediction and tracking; individual fish identification uses deep learning algorithms to classify and track individual fish, and fish trajectory prediction and tracking predicts the movement trajectory of fish based on historical data and deep learning models. The energy system uses solar energy, water flow power generation and intelligent energy scheduling to manage the intelligent allocation of different energy sources to ensure long-term operation.

[0039] The energy system 4 includes a solar panel 401 and a water flow power generation device 402. The solar panel 401 is arranged at the middle of the top of the housing 1 to capture solar energy. The water flow power generation device 402 is arranged at the rear end of the bottom of the housing 1. The water flow power generation device 402 uses water flow to drive a turbine to generate electricity and provide additional power support.

[0040] The hydroelectric power generation device 402 uses the hydroelectric power generation efficiency formula to evaluate the power generation performance of the hydroelectric turbine:

[0041] ; Where ρ is the density of water, A is the windward area of ​​the turbine, and V is the water flow velocity.

[0042] As described above, the fish school monitoring underwater robot of the present invention has a stability optimization system. By adopting advanced hydrodynamic design and intelligent control algorithms, the underwater robot of the present invention can adapt to changes in the external environment in real time and maintain its stability. Specifically, the underwater robot ROV is equipped with multiple vector thrusters, which can automatically adjust the propulsion direction and speed according to the water flow conditions monitored in real time. This technology allows the underwater robot ROV to maintain a high degree of stability in a complex ocean current environment, effectively avoiding the problem that the existing equipment is easily disturbed by the water flow and loses control.

[0043] Specific measures: Real-time collection of water flow information (speed, direction, turbulence intensity, etc.) through multiple ocean current sensors. The adaptive ocean current resistance system uses a PID control algorithm to automatically adjust the vector direction and thrust of the thruster to ensure the stability of the underwater robot ROV in the water. The entire adjustment process has a short response time, ensuring that the underwater robot ROV can respond quickly in a complex ocean current environment. Compared with traditional thruster systems, vector thrusters can accurately adjust thrust according to real-time hydrodynamic characteristics, greatly improving the stability and adaptability of the underwater robot ROV in complex water currents.

[0044] The fish school monitoring underwater robot of the present invention has an efficient image recognition function. The telescopic multi-sensor fusion system 3 technology (acoustic, optical) is adopted, and the individual fish school is accurately tracked by cluster collaborative shooting from multiple angles of the underwater robot ROV, combined with magnetic and acoustic signals to assist in identification. The robot integrates deep learning algorithms, especially convolutional neural networks CNN, and combines data information collected by multiple sensors to enable it to accurately identify targets among fast-swimming fish. Through the training model, the system can identify different types of fish and their characteristics, and ensure the accuracy and speed of identification even when the fish schools overlap. Specific measures: Multiple underwater robots capture multi-angle images of fish schools in real time through optical cameras, and use sonar to obtain the distance and spatial distribution information of fish schools. Multiple images are spliced ​​through deep learning models (such as CNN), and the characteristics of fish are extracted in combination with sensor data to solve the problem of fish overlap. Multi-sensor fusion data analysis uses acoustic and magnetic signals to provide additional physical information to further improve the accuracy of fish school identification. Innovation: Through the combination of multi-angle collaborative shooting and multi-source physical signals, the accuracy of fish school identification is significantly improved, solving the identification problem caused by fish overlap.

[0045] The fish monitoring underwater robot of the present invention has enhanced endurance. The present invention designs an efficient energy system that combines solar energy and water flow power generation technology to enhance the endurance of the underwater robot. Solar panels can collect energy on the water surface, while water flow generators use the kinetic energy of water to convert electricity, ensuring that the robot is not easy to run out of power during long-term operation. This innovation significantly extends the endurance of the equipment, enabling it to conduct long-term and continuous monitoring in offshore areas, overcoming the shortcomings of short endurance of traditional robots. Specific measures: Install a micro water flow generator under the underwater robot ROV to charge the battery by capturing the kinetic energy of water. Cover the surface of the underwater robot ROV with solar panels, and use solar energy to charge when floating to the surface. The energy system intelligently allocates the use of the two energy sources to ensure the long-term endurance of the underwater robot ROV. Innovation: The system extends the endurance of the underwater robot ROV by recycling natural energy, which is suitable for long-term monitoring tasks in the open sea.

[0046] Adaptive ocean current resistance system: In a complex offshore environment, ocean currents will produce strong water flow interference, making it difficult for underwater robots (ROVs) to maintain stable operation. Most existing underwater robots rely on fixed propulsion systems. When facing changing water currents, it is difficult to adapt quickly and is prone to loss of control or deviation. Solution: The present invention adopts an adaptive ocean current resistance system, which integrates multi-directional thrusters, accelerometers, ocean current sensors and other equipment on the underwater robot (ROV) to monitor the external water flow speed, direction and intensity in real time. When the system detects changes in water flow, it automatically starts the control algorithm, and by adjusting the power and direction of each thruster, the underwater robot (ROV) automatically adjusts its posture to ensure that it remains stable in the water. Specific steps: 1. Collect water flow information in real time through multiple ocean current sensors, including water flow speed, direction, turbulence intensity, etc. 2. The system's built-in control algorithm (such as an adaptive algorithm based on PID control) calculates the required working parameters of each thruster in real time based on sensor data. 3. Each thruster adjusts its working state according to the calculation results, adjusts the thrust direction and size, thereby resisting the influence of external water flow and ensuring the stability of the underwater robot (ROV). 4. The entire adjustment process has a short response time (in milliseconds), which can respond to sudden changes in water flow in real time, ensuring that the underwater robot ROV can operate safely in complex ocean current environments. Effect: The system ensures that the underwater robot ROV can maintain a stable posture in a complex water flow environment, avoiding equipment loss of control or deviation from the set trajectory due to changes in ocean currents, allowing the underwater robot ROV to continue to track and shoot fish schools.

[0047] Deep learning fish school recognition system with multi-sensor fusion: When fish schools are dense and swimming fast, traditional visual recognition systems are often unable to accurately distinguish individual fish due to the overlap and rapid movement of fish, resulting in reduced recognition accuracy. Solution: The present invention designs a deep learning recognition system with multi-sensor fusion, which collects multi-dimensional data of fish schools by combining multiple sensors (such as optical cameras and sonar equipment), and uses deep learning algorithms to fuse and analyze these data, thereby achieving efficient fish school recognition and tracking. The present invention proposes a deep learning fish school recognition system with multi-sensor fusion, which combines the multi-dimensional data of fish schools collected by underwater robot ROV individual cluster operation and multiple sensors (such as optical cameras and sonar equipment), and uses deep learning algorithms to perform data fusion analysis to achieve efficient recognition and tracking of fish schools. Specific steps: (1) Multi-angle capture: Use a high-resolution optical camera to capture fish school images in real time, and combine with sonar equipment to obtain the distance and distribution information of fish schools. (2) Feature extraction and fusion: Through the convolutional neural network (CNN), the collected (3) Multi-sensor fusion: After being trained with a large amount of fish image data, the deep learning algorithm can automatically learn fish characteristics and adapt to fish tracking scenarios of different species and forms. (4) Spatial collaboration: Multiple underwater robots collaborate to shoot, combined with multi-image stitching, to achieve efficient identification of individual fish from multiple angles, and at the same time use magnetic or acoustic signals to assist in identification, further improve the expected effect of fish identification accuracy. This system solves the recognition problem of traditional visual recognition systems in the case of overlapping fish and fast-moving fish, significantly improves the accuracy and efficiency of fish tracking, and ensures that ROV can complete monitoring tasks in complex fish environments.

[0048] Energy recovery endurance system: The battery capacity of existing underwater robots ROV is limited. During long-term monitoring tasks, the battery needs to be frequently replaced or returned for charging, which affects the efficiency of continuous operation, especially in remote areas such as the open sea. Solution: The present invention introduces an energy recovery endurance system, which combines water flow power generation and solar energy technology. By capturing external natural energy (such as ocean water flow energy and solar energy), it provides a continuous energy supply for the underwater robot ROV and reduces dependence on batteries. Specific steps: Flow energy generation: A micro-turbine generator is installed under or at the tail of the underwater robot ROV. When the underwater robot ROV moves in the water or encounters a water flow, the water flow drives the turbine to rotate to generate electricity, which directly charges the battery or provides power for the propulsion system. Solar charging: The surface of the underwater robot ROV is covered with high-efficiency solar panels. When the underwater robot ROV floats to the surface, it uses sunlight for charging. The system works in conjunction with the water flow power generation system to ensure that additional power support can be obtained in different scenarios. Intelligent energy management The energy system keeps the power reserve at an optimal level by intelligently allocating these two energy sources. Effect: The energy recovery endurance system significantly extends the endurance of underwater robots (ROVs), which is particularly suitable for offshore or long-term monitoring missions, and solves the bottleneck problem of insufficient endurance of existing underwater robots (ROVs).

[0049] The present invention uses the following classic algorithms to achieve the key function of the underwater robot ROV for fish school monitoring:

[0050] (1) Convolutional Neural Network (CNN) for Fish School Recognition: In the image recognition process, we extract features from fish school images through convolution operations. The convolutional neural network performs a convolution operation on the input image f and the convolution kernel using the following formula: ; This formula is used to extract image features such as edges and shapes to help identify fast-moving and overlapping schools of fish. (2) PID control algorithm for thruster adjustment: The thruster adjustment is based on the PID control algorithm to maintain the stability of the ROV underwater. The control equation is: ; Where e(t) represents the error between the actual position and the expected position. The system adjusts the power of the thruster through proportion, integration and differentiation to adjust the attitude of the ROV in real time. (3) Energy recovery efficiency of the water flow power generation system: In the energy recovery system, we used the water flow power generation efficiency formula to evaluate the power generation performance of the water flow turbine: . Among them, ρ is the density of water, A is the windward area of ​​the turbine, and V is the water flow velocity. This formula is used to analyze the efficiency of the power generation system under different water flow conditions to ensure that the endurance can meet the long-term monitoring needs.

[0051] The high-current-resistance multi-point observation fish school monitoring underwater robot based on multi-source data fusion of the present invention has the following innovative features:

[0052] Innovation 1: Intelligent perception of underwater environment and prediction of underwater target motion based on multi-source data fusion. The present invention proposes an intelligent perception technology for underwater environment based on multi-source data fusion and deep learning model, which combines the data input of multiple sensors such as optics and acoustics to solve the recognition challenges that are difficult to overcome by traditional technologies, such as individual overlap, rapid swimming and complex ocean current interference of fish schools. This technology can accurately identify and track the dynamic position of individual fish schools in real time in complex water environments, and has high robustness and adaptability. Technical Solution 1: Multi-source data fusion and high-precision recognition: The system realizes collaborative work through multi-angle synchronous shooting, and uses high-resolution cameras and acoustic sensors equipped with multiple underwater robots to shoot and collect data of fish schools in real time from different directions, and splices the image data of different sensors into a complete three-dimensional image of fish schools, solving the recognition difficulties caused by fish overlap. In the process of image splicing, acoustic and visual data are used to further separate overlapping individuals, making the recognition and tracking of individual fish schools more accurate. On the basis of providing visual information, the acoustic sensor further provides spatial depth and distance information. Through data fusion algorithms, data from multiple sensors are integrated in real time to form rich three-dimensional images and location information, thereby maintaining high recognition accuracy and efficiency even when individual fish overlap.

[0053] Real-time prediction and motion prediction: The system combines visual and acoustic information and uses a deep learning model to predict the swimming direction and speed changes of fish schools in real time. The PID control algorithm is used to infer the movement trajectory in the short term in the future based on the current position and speed data of the fish school. This multi-parameter optimization process monitors the speed, distance and direction changes of fish in real time to form a reliable motion prediction mechanism. Optimization of the target tracking prediction model: By combining the hydrodynamic motion characteristics of the underwater robot (such as turning, acceleration, deceleration characteristics, etc.), a mapping relationship between the distance between the underwater robot and the target fish school, the speed of the target object and its speed change is constructed, thereby refining the accuracy of the target tracking prediction model. The model integrates the six-degree-of-freedom motion parameters of the underwater robot to achieve continuous tracking and shooting of fish school targets, ensure complete data collection of fish school behavior, and maintain stable tracking effects even in the case of sudden changes in ocean currents.

[0054] Innovation 2: High-maneuverability underwater travel technology based on full-turn vector propulsion. The present invention proposes an underwater travel technology based on full-turn vector thrusters and real-time ocean current detection. Through the coordinated work of multi-directional vector propulsion and ocean current sensors, the underwater robot can adaptively adjust the propulsion direction and thrust size in a complex water flow environment to maintain a highly stable travel effect. The system combines hydrodynamic optimization and PID control algorithms to accurately respond to water flow changes, achieve high maneuverability and precise positioning in turbulent environments, and provide technical support for long-term ocean monitoring and fish tracking tasks. Technical solution 2: Propulsion arrangement and working mechanism: The underwater robot is equipped with four full-turn vector thrusters, which are installed at the four corners of the robot and are symmetrically distributed to ensure balanced thrust and maneuverability. Each thruster can rotate 360 ​​degrees on the vertical axis, and its thrust size can be adjusted independently. Through the coordinated work of the thruster and the ocean current sensor, when the sensor detects a lateral water flow, the thruster on the side will automatically increase the reverse thrust to balance the impact of the external water flow, while the thrusters in other directions are adjusted as needed to maintain the overall propulsion effect. For example, when the underwater robot encounters a lateral current while moving forward, the system will cause the side thrusters to apply thrust in the opposite direction, while the rear thrusters will increase thrust according to the current path, thereby effectively offsetting the impact of the lateral current. The blades and motors of the thrusters are optimized according to the hydrodynamic characteristics, with stable thrust and small drag, ensuring that the underwater robot is not easily deviated from its direction during propulsion.

[0055] Optimization of hydrodynamic characteristics and thruster response: To ensure stability under strong ocean current conditions, the thruster adopts a high-efficiency propeller design and is equipped with a high-precision motor control module. This design can respond quickly in an environment with sudden changes in flow rate, and accurately control the hydrodynamic balance by changing the angle and thrust of the thruster. The system uses a PID control algorithm to input real-time ocean current data into the control module, and optimizes the dynamic response of the thruster through proportional, integral, and differential adjustments, ensuring that the thruster adjusts to water flow changes within milliseconds to maintain the stability of the underwater robot. Specific implementation of the adaptive adjustment system: The system's built-in adaptive adjustment algorithm receives real-time data from the ocean current sensor and compares it with the target path parameters of the underwater robot. If the sensor detects that the change in water flow exceeds the preset threshold (for example, the water flow speed exceeds 1 meter per second or the flow direction deviation exceeds 15 degrees), the control module will immediately adjust the angle and thrust of the corresponding thruster to offset the influence of the external water flow. For example, when a countercurrent is detected in front, the rear thruster will quickly increase the thrust, while the front thruster will relatively reduce the thrust, thereby maintaining the forward speed of the underwater robot. The adaptive adjustment system can stably control the speed and direction of the underwater robot in a strong ocean current environment, ensuring its precise positioning and stable operation in complex waters. Multi-purpose scenarios and performance advantages: This full-turn vector propulsion technology can achieve extremely high flexibility and stability, and is suitable for scenarios such as marine fish monitoring, deep-sea scientific research, and seabed topography exploration. Compared with traditional propulsion systems, this design has the ability to respond quickly to ocean currents and resist interference, which extends the stability and endurance of underwater robots in long-term monitoring missions. In actual operation, the real-time collaborative work of the thruster and ocean current sensor significantly reduces the operational difficulty caused by changes in the external environment, ensures the stable operation of the underwater robot in complex waters, and improves the consistency and accuracy of data collection.

[0056] Innovation point 3: Long-endurance operation technology based on the combination of energy capture technology and chemical battery technology. The present invention proposes a hybrid clean energy supply system that combines flow energy power generation, solar power generation and seawater batteries to provide underwater robots with efficient long-endurance capabilities. The system greatly extends the working time of underwater robots and reduces the need for frequent battery replacement through the combined application of energy capture and energy storage technology. This solution is particularly suitable for long-term monitoring tasks in distant seas and complex waters to ensure the stable operation of underwater robots for a long time. Technical solution 3: Flow energy power generation and thruster collaborative working mechanism: The system designs a set of flow energy power generation devices, which uses water flow to drive the turbine to rotate and generate electricity by installing a micro-water turbine on the bottom or side of the underwater robot. The device can not only provide continuous power for the underwater robot, but also assist the work of the thruster when necessary. When the water flow speed is suitable for power generation, the rotation of the turbine generates electricity, which is charged into the battery or directly powered to the propulsion system; when the propulsion demand is large, the system can switch to the flow energy power generation mode so that the kinetic energy of the turbine directly assists the thruster to work, thereby saving battery energy consumption. This dual-mode collaborative working mechanism not only realizes the capture of flow energy, but also can efficiently utilize external kinetic energy for auxiliary propulsion, effectively improving the system's endurance and efficiency. Solar energy capture system and power management: The surface of the underwater robot is equipped with high-efficiency solar panels. When it floats to the surface, the panels can quickly capture solar energy, convert it into electrical energy and store it in the battery. Through the intelligent allocation of the energy system, the system will reasonably allocate the use of solar energy according to mission requirements and ambient lighting conditions, so as to give priority to powering basic functions such as control systems and sensor modules, reduce battery consumption, and thus extend battery life. In addition, under conditions of sufficient sunlight, solar energy can charge the battery, further extending the independent operation time of the underwater robot.

[0057] Application of seawater battery technology and reverse chemical process: The system also integrates a chemical battery based on seawater electrolyte to provide additional clean energy for underwater robots. Seawater batteries use seawater as an electrolyte to generate electricity through reverse chemical reactions. Specifically, seawater batteries convert chemical energy into electrical energy by reducing and oxidizing ions in seawater, thereby providing a stable power supply. This reverse chemical process can not only directly provide power to the robot, but also provide backup power for the system in emergencies or during long-term operations. Compared with traditional batteries, seawater batteries have a longer service life and good environmental protection, and do not require frequent replacement of electrolytes, which is particularly suitable for long-term underwater operation needs. Optimal allocation of intelligent energy systems: The system's built-in intelligent energy management module can be dynamically allocated according to the current power demand and the status of different energy sources. Power distribution and switching between flow energy, solar energy and seawater batteries ensure that the underwater robot can maintain the best power reserve at any time. The intelligent management module can automatically detect the current battery power and the output status of each energy system. When the power is insufficient, it will give priority to starting the seawater battery or flow energy generation to supplement the required electricity. In an environment with sufficient light, it will give priority to using solar power generation to keep the main battery fully charged, thereby greatly reducing the battery consumption frequency and improving the endurance stability of the equipment.

[0058] In summary, although the present invention has been disclosed as above in terms of preferred embodiments, the above preferred embodiments are not intended to limit the present invention. A person skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be based on the scope defined in the claims.

Claims

1. A highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion, characterized in that: It comprises a shell (1), wherein an adaptive ocean current resistance system (2) is arranged at the front end position of the top of the shell (1), a telescopic multi-sensor fusion system (3) is arranged at the front end position of the bottom of the shell (1), an energy system (4) is arranged on the shell (1), and four vector thrusters (201) are arranged around the shell (1); The adaptive ocean current resistance system (2) comprises an ocean current information detection module and a PID control algorithm real-time control thruster module; the ocean current information detection module is used to collect ocean current information in real time, including water flow speed, direction and turbulence intensity; the control algorithm real-time control thruster module uses the PID control algorithm to control the power and direction of the vector thruster (201) in real time according to the changes in the ocean current, so as to ensure that the underwater robot maintains stability in complex ocean currents; The telescopic multi-sensor fusion system (3) comprises an environment monitoring and data acquisition module, a multi-source data fusion and overlap processing module, a fish school identification and trajectory prediction module, an information output module and an energy management module; the telescopic multi-sensor fusion system (3) collects multi-dimensional data of fish schools by combining multiple sensors, and uses a deep learning algorithm to perform fusion analysis on the data, thereby achieving efficient fish school identification and tracking; The environmental monitoring and data acquisition module includes sensor data acquisition and data preprocessing; the data acquisition sensor includes a high-resolution camera (301), an underwater sonar (302), an underwater lighting system (303) and an ocean current sensor (304); the data preprocessing is used for denoising, resolution adjustment and data enhancement processing; The multi-source data fusion and overlap processing module includes feature extraction, sensor data fusion and overlap elimination technology; the feature extraction extracts the features of the fish school image through the convolutional neural network CNN; the sensor data fusion fuses the multi-sensor data of optics and sonar to obtain complete fish school information; the overlap elimination technology adopts deep learning processing, including convolution layer, pooling layer, fully connected layer and classifier Softmax, and uses deep learning algorithm to classify and track individual fish school; The fish school identification and trajectory prediction module includes fish school individual identification and fish school trajectory prediction and tracking; the fish school individual identification uses a deep learning algorithm to classify and track fish school individuals, and the fish school trajectory prediction and tracking predicts the movement trajectory of the fish school based on historical data and a deep learning model; The control equation of the PID control algorithm is: ； Wherein, e(t) represents the error between the actual position and the expected position. The adaptive current resistance system (2) adjusts the power of the propeller through proportion, integration and differentiation, and adjusts the posture of the underwater robot in real time; parameter adjustment range: proportional gain K p The value range is 0.1~1.0, which is used to control the response speed; integral gain K i The value range is 0.01~0.1, which is used to eliminate steady-state error; differential gain K d The value range is 0.01~0.1, which is used to improve the anti-disturbance ability; Applicable conditions: The PID control algorithm is suitable for complex ocean current environments with water flow velocity ranging from 0.5 to 3.0 m / s, and adjusts the thrust output of the thruster in real time to ensure that the underwater robot maintains posture stability and navigation accuracy under sudden changes in ocean currents.

2. According to claim 1, a highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion, characterized in that: The energy management module utilizes solar energy, water flow power generation and intelligent energy scheduling to manage the intelligent allocation of different energy sources to ensure long-term operation.

3. The highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion according to claim 1 is characterized in that: The adaptive ocean current resistance system (2) comprises a water flow sensor, a posture sensor and an acceleration sensor, which are used to monitor the water flow speed and direction in real time.

4. The highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion according to claim 1 is characterized in that: The energy system (4) comprises a solar panel (401) and a water flow power generation device (402); the solar panel (401) is arranged at the middle position of the top of the housing (1) for capturing solar energy; the water flow power generation device (402) is arranged at the rear end position of the bottom of the housing (1); the water flow power generation device (402) uses water flow to drive a turbine to generate electricity, thereby providing additional power support.

5. The highly flow-resistant multi-point observation fish school monitoring underwater robot based on multi-source data fusion according to claim 4 is characterized in that: The water flow power generation efficiency formula of the water flow power generation device (402) is used to evaluate the power generation performance of the water flow turbine: ;in, ρ is the density of water, A is the frontal area of ​​the turbine, V is the water flow velocity.

Images