Aquatic product oxygenation unmanned ship intelligent control method and system based on multi-sensor fusion

Abstract

The invention discloses an aquatic product oxygenation unmanned ship intelligent control method and system based on multi-sensor fusion, and relates to the technical field of unmanned ship control, and the method comprises the following steps: synchronously collecting a variety of data, and generating high-precision pose estimation through extended Kalman filtering fusion; discretizing the fish pond into grids, and dynamically calculating the weight of each grid according to the real-time dissolved oxygen value and a preset threshold value; generating an initial bow-shaped path based on the starting point and the ending point, optimizing the path in real time, and adaptively adjusting the scanning line direction and spacing; when grid dissolved oxygen is lower than a threshold value, an oxygenation pump is started, oxygenation flow is dynamically adjusted according to dissolved oxygen deviation, and the larger the deviation is, the higher the flow is; a double-paddle differential model is adopted to track the course and the speed, and three-level response is triggered when an obstacle is encountered; through 5G data and video streams, when the electric quantity is lower than a threshold value, a Dijkstra algorithm is used to plan a shortest obstacle avoidance path for homeward voyage. According to the method, the dynamic probability distribution diagram is constructed by fusing real-time dissolved oxygen data, and oxygenation targeting control is realized in combination with a path cost function.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of unmanned ship control, in particular to an intelligent control method and system for an aquaculture oxygenation unmanned ship based on multi-sensor fusion. BACKGROUND

[0002] Traditional aquaculture oxygenation equipment generally adopts a fixed position deployment mode (such as a propeller type or waterwheel type oxygenation machine), and the core defect is that the operation mode does not match the spatial demand for dynamic oxygen solubility. Due to the fixed position of the equipment and the lack of global oxygen solubility monitoring capability, the oxygenation operation is seriously disconnected from the actual oxygen deficiency area of the fishpond: on the one hand, the water area that has reached the standard is repeatedly oxygenated, causing energy waste; on the other hand, the deep water area or remote area is continuously oxygen-deficient due to insufficient coverage.

[0003] This extensive oxygen supply mode cannot respond to the spatial heterogeneity of oxygen solubility, resulting in a resource mismatch problem of "local oversaturation and overall imbalance", which is the main reason for restricting the increase of aquaculture density and the reduction of energy consumption. Therefore, the following solution is proposed to solve the above problems. SUMMARY

[0004] The purpose of the present application is to provide an intelligent control method and system for an aquaculture oxygenation unmanned ship based on multi-sensor fusion, which can realize spatial adaptive precise configuration of oxygenation resources through a cooperative control mechanism of global oxygen solubility real-time monitoring, dynamic arch-shaped path planning and nonlinear flow regulation, and solve the problem of spatial mismatch between oxygen supply area and oxygen solubility demand caused by the fixed operation mode of the existing fixed oxygenation equipment.

[0005] To solve the above technical problems, the present application is realized by the following technical scheme: The present application is an intelligent control method for an aquaculture oxygenation unmanned ship based on multi-sensor fusion, which comprises the following steps: Step S1, data acquisition and preprocessing: synchronously acquire GPS, IMU, geomagnetic, oxygen solubility, obstacle distance and power data, and generate high-precision pose estimation through extended Kalman filter fusion; Step S2, constructing an oxygen solubility demand probability distribution map: discretize the fishpond into grids, and dynamically calculate the weight of each grid according to the real-time oxygen solubility value and the preset threshold value; Step S3, path generation: generate an initial arch-shaped path based on the starting point and the ending point, and optimize the path in real time to adaptively adjust the scanning line direction and spacing; Step S4, execution control: start the oxygenation pump when the grid oxygen solubility is lower than the threshold value, and dynamically adjust the oxygenation flow according to the oxygen solubility deviation, the greater the deviation, the higher the flow; Step S5, motion control: track the heading and speed using a double-paddle differential model, and trigger a three-level response when encountering an obstacle; Step S6, task management and communication: Data and video streams are transmitted back through 5G. When the battery power is lower than the threshold, the Dijkstra algorithm is used to plan the shortest obstacle avoidance path for return.

[0006] Furthermore, the step S1 specifically includes the following steps: Step S11, data acquisition: synchronously obtain the following sensor data: Positioning data : From the dual-antenna GPS module, output latitude and longitude coordinates and heading angle; Posture data : quaternion from IMU; geomagnetic data : Calibrate electronic compass output; Dissolved oxygen concentration : Real-time measurement value of dissolved oxygen sensor; Obstacle distance : Ultrasonic radar ranging value; Hull power : Battery management system output; Step S12: Data fusion: using extended Kalman filter fusion Generate high-precision pose estimates:

[0007] Where, is the estimated value of the state vector at time k, is the nonlinear state transfer function, is the state vector at time k-1, is the control input vector at time k, is the Kalman gain matrix, is the sensor observation vector, is the observation model function, x is the x-coordinate of the global coordinate system, y is the y-coordinate of the global coordinate system, is the heading angle, is the linear velocity, is the angular velocity; This step synchronously collects GPS positioning, IMU attitude, geomagnetic direction, dissolved oxygen concentration, obstacle distance and battery power data, and generates high-precision hull posture estimation through extended Kalman filter fusion to provide real-time status input for path planning.

[0008] Furthermore, the step S2 of constructing the probability distribution map of dissolved oxygen demand specifically includes the following steps: Step S21, spatial discretization: Divide the fish pond water area into M×N grids, each grid Associated dissolved oxygen demand weight ; Step S22, dynamic weight update: according to real-time dissolved oxygen data With preset threshold , calculate the demand weight:

[0009] Where, is the dissolved oxygen demand weight of grid (i, j), is the sensitivity coefficient, The dissolved oxygen threshold set by the user, is the real-time dissolved oxygen value of grid (i, j), are grid row and column indexes, is the natural exponential function, is the maximum value function; This step discretizes the fish pond into a grid map, dynamically calculates the dissolved oxygen demand weight of each grid based on the deviation between the real-time dissolved oxygen measurement value and the preset threshold, and generates a spatial probability distribution map with low oxygen areas as high priority.

[0010] Furthermore, the step S3, path generation specifically includes the following steps: Step S31, initial path generation: user sets the starting point and end point Using the direction of the line connecting the two points as the reference axis, generate parallel and equidistant scanning lines; Step S32, adaptive steering strategy: define the path cost function J to integrate coverage efficiency and dissolved oxygen demand:

[0011] Where, path The total cost, is the set of candidate paths, is the weight and weight of the path coverage grid, is the total path length, is the path length weight coefficient, is the sum of the reciprocals of the obstacle distances at each point on the path, is the obstacle avoidance risk weight coefficient, is the obstacle distance; Step S33, real-time path optimization: At each waypoint k, based on the updated , and solve it using the gradient descent method:

[0012] Where, The next path to be optimized.

[0013] Adjust the direction and spacing of the next scan line to avoid obstacles and prioritize coverage of high-weight areas; The step generates an initial arcuate scanning path based on the start and end points, and optimizes the path spacing and turning angle in real time by fusing the cost function of the dissolved oxygen demand weight, path length and obstacle avoidance risk, to achieve the collaborative balance of coverage efficiency and oxygen demand.

[0014] Further, the step S4 specifically includes the following steps: Step S41, oxygenation trigger condition: when the ship is located in the grid (i, j) and satisfies , start the oxygenation pump; Step S42, flow self-adaptive adjustment: calculate the oxygenation flow according to the dissolved oxygen deviation

[0015] wherein, is the output flow of the oxygenation pump, is the maximum flow, is the dissolved oxygen deviation, is the flow adjustment parameter, is a natural constant; The oxygenation pump is triggered when the ship enters the low-oxygen grid, and the oxygenation flow is dynamically adjusted according to the dissolved oxygen deviation value, and a nonlinear function is used to realize the precise oxygen supply strategy of slow oxygenation for small deviation and full power for large deviation.

[0016] Further, the step S5 specifically includes the following steps: Step S51, double-propeller differential drive model: the left / right propeller rotation speed is determined by the heading angle deviation and the linear speed v:

[0017] wherein, is the left propeller rotation speed, is the right propeller rotation speed, is the target linear speed, is the turning gain coefficient, is the heading angle deviation, is the propeller radius; Step S52, emergency obstacle avoidance strategy: when is less than the threshold value, trigger a three-level response: the speed is reduced to and the ship deviates along the tangent direction of the obstacle ; when it is still approaching, the ship is stopped urgently and an audible and visual alarm is sounded; This step calculates the double-propeller differential rotation speed command according to the path planning result, realizes the track following through the heading angle deviation control, and when encountering close-range obstacles, executes the deceleration, deviation avoidance and emergency stop in sequence. ​

[0018] Further, the step S6, task management and communication specifically includes the following steps: Step S61, state monitoring: real-time feedback of sensor data, video stream and path trajectory to APP through 5G image transmission module; Step S62, low power self-return: when , terminate the current task, and generate the shortest path to return to the charging point:

[0019] use Dijkstra algorithm to calculate the shortest obstacle avoidance path from the current position to the starting point S wherein, is the return path, is the shortest path algorithm, is the current state vector, is the charging point coordinates; This step returns the hull state and video monitoring data to the user terminal through the 5G network; when low power is detected, the shortest obstacle avoidance path is automatically planned to return to the charging, ensuring the continuity of the task.

[0020] The intelligent control system of the aquatic oxygenation unmanned ship based on multi-sensor fusion includes a sensor acquisition module, a data fusion processing module, an environment perception module, an intelligent path planning module, an oxygenation decision module, a motion control module, a task management module, a communication and interaction module, a user terminal and a propeller / oxygenation pump. The sensor acquisition module, data fusion processing module, environment perception module, intelligent path planning module are connected in turn, the output end of the environment perception module is unidirectionally connected with the oxygenation decision module, the output end of the intelligent path planning module is unidirectionally connected with the motion control module and the task management module respectively, the output end of the task management module is unidirectionally connected with the communication and interaction module, the motion control module and the oxygenation decision module respectively, the output end of the communication and interaction module is unidirectionally connected with the user terminal, the output end of the user terminal is unidirectionally connected with the task management module, the output end of the motion control module is unidirectionally connected with the oxygenation decision module, the output end of the oxygenation decision module is unidirectionally connected with the propeller / oxygenation pump, and the output end of the propeller / oxygenation pump is unidirectionally connected with the sensor acquisition module.

[0021] Further, the sensor acquisition module is used for real-time acquisition of environment and hull state data such as positioning, attitude, dissolved oxygen and obstacles; The data fusion processing module is used for fusion of multi-source sensor data to generate high-precision pose estimation through EKF algorithm; The environment perception module is used for constructing a dissolved oxygen demand probability distribution map and dynamically calculating grid weights; The intelligent path planning module is used for planning The function dynamically generates and optimizes the arch-shaped cruise path; The oxygen-increasing decision module is used for adaptively adjusting the oxygen-increasing pump flow Q according to the dissolved oxygen threshold value; The motion control module is used for realizing path tracking and emergency obstacle avoidance through double-propeller differential control; The task management module is used for coordinating cruise mode switching, low-power return and fault handling processes; The communication and interaction module is used for providing 5G image transmission, remote control signal receiving and ship-shore data bidirectional interaction; The user terminal module is used for realizing monitoring, parameter setting and historical data playback functions on a mobile terminal APP.

[0022] The present application has the following beneficial effects: 1. The present application constructs a dynamic probability distribution graph by fusing real-time dissolved oxygen data, and realizes targeted oxygen-increasing control in combination with a path cost function; the function integrates dissolved oxygen demand, path length and obstacle avoidance risk into an optimization framework, drives the unmanned ship to preferentially cover low-oxygen areas and dynamically adjusts the coverage density; the oxygen-increasing pump is based on a nonlinear flow adjustment mechanism of dissolved oxygen deviation, ensuring that the oxygen supply intensity and the degree of hypoxia are accurately matched; such design can reduce invalid oxygen-increasing energy consumption, and avoid the problems of repeated oxygen supply in local areas or insufficient coverage in key areas.

[0023] 2. The present application realizes real-time path adjustment through a three-level obstacle avoidance response mechanism combined with gradient optimization, so that the unmanned ship can still maintain continuous operation capability in obstacle-dense areas; the obstacle avoidance risk term in the path cost function is used to actively avoid navigation threats, and a double-propeller differential driving model is used to realize rapid heading correction; such collaborative control design can improve the robustness and task completion rate of the equipment under complex working conditions such as wind and waves, and water entanglement.

[0024] 3. The dynamic and adaptive arch-shaped path generation algorithm of the present application adjusts the scanning line spacing and turning strategy by online solving the minimum value of the path cost function; in the area with sufficient dissolved oxygen, the coverage spacing is automatically expanded to reduce redundant navigation, and in the hypoxic area, the spacing is contracted to realize fine operation, effectively balancing the coverage efficiency and navigation energy consumption; in combination with the shortest path planning of low-power autonomous return, the invalid navigation distance is reduced; such design reduces the overall energy consumption of the system from the dual dimensions of path planning and energy management, prolongs the single operation cycle.

[0025] 4、The application is based on the dynamic mapping mechanism of gridding dissolved oxygen monitoring and weight updating, so that the unmanned ship has real-time sensing ability for dissolved oxygen distribution of fish ponds; the deep coupling of oxygenation decision and path planning ensures that the device can automatically adjust the operation mode according to the spatial heterogeneity of dissolved oxygen; through the multi-parameter monitoring and path setting function integrated by the mobile phone APP, the user can remotely implement fine operations such as partition control and key area oxygenation strengthening, thereby improving the scientific nature and response speed of large water area management.

[0026] Of course, implementing any product of the present application does not necessarily require all the advantages described above. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed for the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0028] Figure 1 The flowchart of the intelligent control method of the aquaculture oxygenation unmanned ship based on multi-sensor fusion of the present application; Figure 2 The structure diagram of the intelligent control system of the aquaculture oxygenation unmanned ship based on multi-sensor fusion of the present application. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0030] Please refer to Figure 1 The intelligent control method of the aquaculture oxygenation unmanned ship based on multi-sensor fusion of the present application includes the following steps: Step S1, data acquisition and preprocessing: synchronously acquiring GPS, IMU, geomagnetic, dissolved oxygen, obstacle distance and power data, and generating high-precision pose estimation through extended Kalman filter fusion; Step S1 specifically includes the following steps: Step S11, data acquisition: synchronously acquiring the following sensor data: Positioning data : from a dual-antenna GPS module, outputting latitude and longitude coordinates and heading angle; Attitude data : quaternion from IMU; magnetic data : calibrated electronic compass output; dissolved oxygen concentration : real-time measurement value of dissolved oxygen sensor; obstacle distance : ultrasonic radar ranging value; hull electric quantity : battery management system output; Step S12, data fusion: extended Kalman filter fusion generate high-precision pose estimation:

[0031] wherein, is the state vector estimation value at time k, is a nonlinear state transition function, is the state vector at time k-1, is the control input vector at time k, is the Kalman gain matrix, is the sensor observation vector, is the observation model function, x is the global coordinate system x coordinate, y is the global coordinate system y coordinate, is the heading angle, is the linear velocity, is the angular velocity.

[0032] Step S2, build dissolved oxygen demand probability distribution map: discretize the fish pond into a grid, and dynamically calculate the weight of each grid according to the real-time dissolved oxygen value and the preset threshold value; Step S2, building a dissolved oxygen demand probability distribution map specifically includes the following steps: Step S21, spatial discretization: divide the fish pond water area into MxN grids, and each grid correlation dissolved oxygen demand weight ; Step S22, dynamic weight update: according to the real-time dissolved oxygen data and the preset threshold value , calculate the demand weight:

[0033] wherein, is the dissolved oxygen demand weight of grid (i, j), is the sensitivity coefficient, is the user-set dissolved oxygen threshold value, is the real-time dissolved oxygen value of grid (i, j), are all grid row and column indexes, is the natural exponential function, is the maximum value function.

[0034] Step S3, path generation: generate initial arch-shaped path based on start and end points, and optimize path in real time, adaptively adjust scanning line direction and interval; Step S3, path generation specifically includes the following steps: Step S31, initial path generation: user sets start point and end point Generate parallel equidistant scanning lines with the direction of the line connecting the two points as the reference axis; Step S32, adaptive steering strategy: define path cost function J to integrate coverage efficiency and dissolved oxygen demand:

[0035] In the formula, the total cost of the path, is the path coverage grid weight sum, is the total length of the path, is the path length weight coefficient, is the sum of the reciprocal of the obstacle distance of each point of the path, is the obstacle avoidance risk weight coefficient, is the obstacle distance; Step S33, real-time path optimization: at each waypoint k, based on the updated , use gradient descent method to solve:

[0036] In the formula, is the next segment of the optimized path.

[0037] Adjust the direction and interval of the next scanning line to ensure avoiding obstacles and preferentially covering high weight areas.

[0038] Step S4, execution control: start the oxygenation pump when the grid dissolved oxygen is lower than the threshold value, dynamically adjust the oxygenation flow according to the dissolved oxygen deviation, the larger the deviation, the higher the flow; Step S4, execution control specifically includes the following steps: Step S41, oxygenation triggering condition: when the ship is located in grid (i,j) and satisfies , start the oxygenation pump; Step S42, flow adaptive adjustment: calculate the oxygenation flow according to the dissolved oxygen deviation

[0039] In the formula, is the output flow of the oxygenation pump, ​​​For maximum flow, For dissolved oxygen deviation, For flow regulation parameters, For natural constants.

[0040] Step S5, motion control: adopt double propeller differential model to track heading and speed, and trigger three-level response when encountering obstacles; Step S5, motion control specifically includes the following steps: Step S51, double propeller differential drive model: left / right propeller rotation speed Determined by the heading angle deviation And linear speed v:

[0041] In the formula, The left propeller rotation speed is The right propeller rotation speed is The target linear speed is The steering gain coefficient is The heading angle deviation is The propeller radius is Step S52, emergency obstacle avoidance strategy: when Less than the threshold value, trigger three-level response: speed deceleration to Yaw along the tangent direction of the obstacle When still in continuous approach, emergency stop and sound-light alarm.

[0042] Step S6, task management and communication: back data and video stream through 5G, use Dijkstra algorithm to plan the shortest obstacle avoidance path when the power is lower than the threshold value.

[0043] Step S6, task management and communication specifically includes the following steps: Step S61, state monitoring: real-time back sensor data, video stream and path trajectory to APP through 5G image transmission module; Step S62, low power autonomous return: when Terminate the current task and generate the shortest path to return to the charging point:

[0044] Use Dijkstra algorithm to calculate the shortest obstacle avoidance path from the current position to the starting point S In the formula, The return path is The shortest path algorithm is The current state vector is The charging point coordinates are

[0045] Please refer to Figure 2As shown, the application is an intelligent control system for an aquaculture oxygenation unmanned ship based on multi-sensor fusion, which comprises a sensor acquisition module, a data fusion processing module, an environment perception module, an intelligent path planning module, an oxygenation decision module, a motion control module, a task management module, a communication and interaction module, a user terminal and a propeller / oxygenation pump. The sensor acquisition module, the data fusion processing module, the environment perception module and the intelligent path planning module are sequentially connected, the output end of the environment perception module is unidirectionally connected with the oxygenation decision module, the output end of the intelligent path planning module is unidirectionally connected with the motion control module and the task management module, the output end of the task management module is unidirectionally connected with the communication and interaction module, the motion control module and the oxygenation decision module, the output end of the communication and interaction module is unidirectionally connected with the user terminal, the output end of the user terminal is unidirectionally connected with the task management module, the output end of the motion control module is unidirectionally connected with the oxygenation decision module, the output end of the oxygenation decision module is unidirectionally connected with the propeller / oxygenation pump, and the output end of the propeller / oxygenation pump is unidirectionally connected with the sensor acquisition module.

[0046] The sensor acquisition module is used for acquiring real-time positioning, attitude, dissolved oxygen, obstacle and other environmental and ship state data; The data fusion processing module is used for generating high-precision pose estimation by fusing multi-source sensor data through an EKF algorithm; The environment perception module is used for constructing a dissolved oxygen demand probability distribution map and dynamically calculating grid weights; The intelligent path planning module is used for dynamically generating and optimizing a zigzag cruise path based on The oxygenation decision module is used for adaptively adjusting the oxygenation pump flow Q according to the dissolved oxygen threshold value; The motion control module is used for realizing path tracking and emergency obstacle avoidance through double-paddle differential control; The task management module is used for coordinating cruise mode switching, low-power return and fault handling processes; The communication and interaction module is used for providing 5G image transmission, remote control signal reception and ship-shore data bidirectional interaction; The user terminal module is used for realizing monitoring, parameter setting and historical data playback functions on a mobile terminal APP.

[0047] One specific application of the embodiment is: The implementation scenario is set as follows: fishpond size: 120m*80m rectangular water area; starting point coordinate: S(0,0), ending point coordinate: E(120,80); preset dissolved oxygen threshold value: ; initial cruise speed: ; maximum oxygenation flow: .

[0048] Implementation steps: ​Step S1: Sensor data fusion Acquire real-time data: GPS positioning IMU quaternion ; dissolved oxygen concentration ; obstacle distance (right front); Extended Kalman filter fusion pose:

[0049] Step S2: Dissolved oxygen demand map construction Divide the fish pond into 120x80 grids (Δs=1 m); Update the weight of the current grid (15, 23) (α=0.5):

[0050] Historical data shows the dissolved oxygen of grid (20, 30) , weight:

[0051] Step S3: Dynamic adjustment of arch-shaped path Initial path: generate parallel scan lines with a distance of 40; Optimize at waypoint (15.21, 22.73): Detect right front obstacle (safety threshold >1 m); Calculate the cost of the alternative path : Option A (original straight line): Cover the weight of grid (15, 23)

[0052] Path length 1 m, cost item 0.3x1=0.3; Obstacle avoidance risk 0.2x(1 / 4.3)≈0.046; Total cost ; Option B (east deviation 15°): New grid (16, 24) weight ; Path length 1.04 m, cost 0.3x1.04=0.312; Increased to 6.1 m, risk cost 0.2x(1 / 6.1)≈0.033; Total cost ; Option C (west deviation 10°): New grid (14, 24) weight ; Path length 0.98 m, cost 0.294; Risk cost 0.028; Total cost ; Decision: Choose The smallest westward path; Step S4: Oxygenation control implementation; Current grid (15, 23) dissolved oxygen Trigger oxygenation; Flow calculation ( ):

[0053] ; The oxygenation pump runs at 77.6 L / min (97% of maximum flow); Step S5: Motion control implementation; Target heading angle: (Westward correction); Dual propeller differential control (r=0.05 m, k=0.1):

[0054] Left propeller slows to 5.65 rad / s, right propeller speeds up to 6.35 rad / s to achieve left turn; Step S6: Global task monitoring; Power management: Current , ongoing task; Exception event: Detected (Sudden obstacle); Execute level 3 response: Slow down to ; head due north (Tangent direction to avoid obstacle); resume cruising after obstacle is removed.

[0055] In the description of the present specification, the description of the terms "one embodiment", "example", "specific example" and the like means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0056] The preferred embodiments of the application disclosed above are only to facilitate the elucidation of the application. The preferred embodiments do not describe all the details of the application and limit the application to the specific embodiments described. Obviously, many modifications and variations can be made in light of the teachings above. The description is chosen and described in order to best explain the principles of the application and its practical application to thereby enable others skilled in the art to best utilize the application and get the best results from the application. The application is only limited by the claims and their full scope and equivalents.

Claims

1. An intelligent control method for an unmanned aquaculture oxygenation boat based on multi-sensor fusion, characterized in that: The control method comprises the following steps: Step S1, data acquisition and preprocessing: synchronously collect GPS, IMU, geomagnetic, dissolved oxygen, obstacle distance and power data, and generate high-precision pose estimation through extended Kalman filter fusion; Step S2: Construct a probability distribution map of dissolved oxygen demand: discretize the fish pond into grids, and dynamically calculate the weight of each grid based on the real-time dissolved oxygen value and the preset threshold; Step S3, path generation: generating an initial bow-shaped path based on the starting point and the end point, optimizing the path in real time, and adaptively adjusting the direction and spacing of the scan lines; Step S4, execution control: when the grid dissolved oxygen is lower than the threshold, the oxygen pump is started, and the oxygen flow rate is dynamically adjusted according to the dissolved oxygen deviation. The larger the deviation, the higher the flow rate; Step S5, motion control: using a dual-propeller differential model to track heading and speed, triggering a three-level response when encountering an obstacle; Step S6, task management and communication: Data and video streams are transmitted back through 5G. When the battery power is lower than the threshold, the Dijkstra algorithm is used to plan the shortest obstacle avoidance path for return.

2. The intelligent control method for aquaculture oxygenation unmanned boat based on multi-sensor fusion according to claim 1 is characterized in that: The step S1 specifically includes the following steps: Step S11, data acquisition: synchronously collect multi-source sensor data, obtain positioning data including latitude and longitude coordinates and heading angles through the dual-antenna GPS module, collect quaternion data representing the hull posture through the IMU inertial measurement unit, obtain calibrated three-dimensional geomagnetic data through the electronic compass, and simultaneously read the dissolved oxygen concentration value output by the dissolved oxygen sensor, the obstacle distance value detected by the ultrasonic radar, and the remaining power percentage provided by the battery management system in real time; Step S12, data fusion: The extended Kalman filter technology is used to fuse the above heterogeneous data: the state estimation value at the previous moment is predicted with the current control input using the nonlinear state transfer function, and then the difference between the current sensor observation value and the observation model prediction value is combined, and correction is performed through the dynamically calculated Kalman gain matrix, and finally a high-precision five-dimensional state vector estimation result is output.

3. The intelligent control method for aquaculture oxygenation unmanned boat based on multi-sensor fusion according to claim 1 is characterized in that: The step S2, constructing a probability distribution map of dissolved oxygen demand, specifically comprises the following steps: Step S21, spatial discretization: dividing the fish pond water area into rectangular grid arrays according to preset sizes to form discretized spatial management units; Step S22, dynamic weight update: Each grid unit is bound to the dissolved oxygen concentration data collected by the dissolved oxygen sensor in real time and compared with the dissolved oxygen safety threshold set by the user; when the real-time dissolved oxygen value of a grid is lower than the set threshold, the system automatically assigns a high demand weight to the grid; conversely, when the dissolved oxygen value reaches or exceeds the threshold, it is assigned a low demand weight; The weight change follows a nonlinear response law: the greater the dissolved oxygen shortage, the faster the weight growth rate.

4. The intelligent control method for aquaculture oxygenation unmanned boat based on multi-sensor fusion according to claim 1 is characterized in that: The step S3, path generation specifically includes the following steps: Step S31, generating an initial path: generating an initial bow-shaped cruising path based on the start and end points set by the user, wherein the cruising path is composed of a series of parallel and equidistant scanning lines; Step S32, Adaptive Steering Strategy: Define a path cost function containing three key elements: the first is the weighted sum of dissolved oxygen demand within the path coverage area, where the weight value is dynamically calculated based on real-time dissolved oxygen data; the second is the total path length multiplied by the length weight coefficient; and the third is the sum of the reciprocal distances from each point on the path to the obstacle multiplied by the obstacle avoidance risk coefficient. Step S33, real-time path optimization: when making decisions at each waypoint, based on the latest updated dissolved oxygen distribution weight map and obstacle distance data, a gradient descent method is used to solve the optimized path that minimizes the cost function in real time; This process adaptively adjusts the direction angle and line spacing of the scanning line: it automatically expands the spacing in high dissolved oxygen areas to reduce repeated coverage, and reduces the spacing in low dissolved oxygen areas to enhance oxygenation. At the same time, it dynamically corrects the steering angle according to the location of obstacles to ensure safe passage.

5. The intelligent control method for aquaculture oxygenation unmanned boat based on multi-sensor fusion according to claim 1 is characterized in that: The step S4, executing the control, specifically includes the following steps: Step S41, oxygenation triggering condition: When the unmanned boat sails to a grid area of ​​the fish pond, the system will detect the dissolved oxygen concentration value of the area in real time and compare it with the minimum dissolved oxygen threshold preset by the user; if the current dissolved oxygen concentration is detected to be lower than the threshold, the oxygenation pump will be immediately started; Step S42, adaptive flow adjustment: The oxygen flow is adjusted based on the dynamic calculation of the deviation between the current dissolved oxygen concentration and the threshold value: the larger the deviation, the higher the flow. Specifically, when the initial deviation is small, the flow rises gently, and as the deviation increases, the flow accelerates nonlinearly until it approaches the preset maximum flow limit, ensuring that the low-oxygen area is adequately supplied with oxygen while avoiding energy waste in the high-oxygen area.

6. The intelligent control method for aquaculture oxygenation unmanned boat based on multi-sensor fusion according to claim 1 is characterized in that: The step S5, motion control specifically includes the following steps: Step S51, dual-propeller differential drive model: Based on the deviation between the target heading angle and the actual heading angle and the preset linear velocity, the speed of the left propeller is calculated using the dual-propeller differential drive model. The speed of the left propeller is determined by adding the target linear velocity to the steering gain coefficient multiplied by the heading angle deviation, divided by the blade radius. The speed of the right propeller is determined by subtracting the same steering gain coefficient multiplied by the heading angle deviation from the target linear velocity, divided by the blade radius. Step S52, emergency obstacle avoidance strategy: When the ultrasonic radar detects that the obstacle is less than 1 meter away, the system immediately activates a three-level obstacle avoidance response mechanism. The first level reduces the ship speed to the minimum safe speed; the second level controls the hull to deflect 15 degrees along the tangent direction of the obstacle to avoid it; if it continues to approach the obstacle, the third level emergency stop command is triggered and the sound and light alarm device is activated to forcibly lock the hull; the entire process realizes dynamic track correction by real-time integration of heading angle deviation, obstacle distance and power parameters.

7. The intelligent control method for aquaculture oxygenation unmanned boat based on multi-sensor fusion according to claim 1 is characterized in that: The step S6, task management and communication, specifically includes the following steps: Step S61, status monitoring: The real-time sensor data, monitoring video stream and navigation trajectory of the unmanned boat are synchronously transmitted to the user's mobile phone APP through the 5G wireless image transmission module to realize remote status monitoring; Step S62, autonomous return due to low battery: When it is detected that the battery level of the hull is lower than the threshold, the system immediately terminates the current oxygen enrichment task, and uses the shortest path algorithm to plan a return route that avoids all known obstacles based on the current position and the coordinates of the initial charging point, and controls the unmanned boat to autonomously return to the charging point; during the voyage, the position and battery level data are continuously transmitted back, and after the task is completed, the historical operation records are automatically saved and the boat enters a dormant state.

8. An intelligent control system for an unmanned aquaculture oxygenation vessel based on multi-sensor fusion, used to implement the control method according to any one of claims 1 to 7, characterized in that: The control system includes a sensor acquisition module, a data fusion processing module, an environmental perception module, an intelligent path planning module, an oxygenation decision module, a motion control module, a task management module, a communication and interaction module, a user terminal and a propeller / oxygenation pump; The sensor acquisition module, data fusion processing module, environmental perception module, and intelligent path planning module are connected in sequence; the output end of the environmental perception module is unidirectionally connected to the oxygenation decision module; the output end of the intelligent path planning module is unidirectionally connected to the motion control module and the task management module, respectively; the output end of the task management module is unidirectionally connected to the communication and interaction module, the motion control module, and the oxygenation decision module, respectively; the output end of the communication and interaction module is unidirectionally connected to the user terminal; the output end of the user terminal is unidirectionally connected to the task management module; the output end of the motion control module is unidirectionally connected to the oxygenation decision module; the output end of the oxygenation decision module is unidirectionally connected to the propeller / oxygenation pump; and the output end of the propeller / oxygenation pump is unidirectionally connected to the sensor acquisition module.

9. The intelligent control system for aquaculture oxygenation unmanned boat based on multi-sensor fusion according to claim 8 is characterized in that: The sensor acquisition module is used to obtain real-time positioning, attitude, dissolved oxygen, obstacles and other environmental and hull status data; The data fusion processing module is used to generate high-precision pose estimation by fusing multi-source sensor data through the EKF algorithm; The environmental perception module is used to construct a probability distribution map of dissolved oxygen demand and dynamically calculate grid weights; The intelligent path planning module is used based on The function dynamically generates and optimizes the bow-shaped cruise path; The oxygenation decision module is used to adaptively adjust the oxygenation pump flow rate Q according to the dissolved oxygen threshold; The motion control module is used to achieve path tracking and emergency obstacle avoidance through dual-propeller differential control; The task management module is used to coordinate the cruise mode switching, low battery return and fault handling process; The communication and interaction module is used to provide 5G image transmission, remote control signal reception and two-way interaction of ship-shore data; The user terminal module is used to implement monitoring, parameter setting and historical data playback functions on the mobile APP.

Images