A method for automatically retrieving fish traps from unmanned vessels

Abstract

This invention provides a method for automatically retrieving fish traps from an unmanned vessel. The method utilizes a positioning module on the fish trap to obtain its location information, and controls the unmanned vessel to navigate to the location of the fish trap based on this information. The method then controls a retrieval device on the unmanned vessel to retrieve the fish trap. Specifically, this includes: controlling the retrieval device to extend a hook to prepare to grab the fish trap; capturing images of the fish trap using a video monitoring device mounted on the hook; analyzing the positional relationship between the hook and the fish trap using the captured images; adjusting the hook's position based on this relationship; controlling the hook to grab the fish trap; pulling the grabbed fish trap up to a position higher than the unmanned vessel's railing; controlling the retrieval device to rotate the fish trap to a predetermined position on the unmanned vessel; lowering the grabbed fish trap to the predetermined position; and finally, controlling the hook to release the fish trap. The advantage of this invention is that it enables the automatic retrieval of deployed fish traps.

Description

[Technical Field]

[0001] This invention relates to the field of unmanned vessel technology, and in particular to a method for automatically retrieving fish traps from an unmanned vessel. [Background Technology]

[0002] Currently, one method for fishing in inland rivers and seas is to use fish traps. The retrieval of these traps typically involves manual retrieval by personnel using boats after the tide recedes or the weather improves. However, this manual retrieval method has several drawbacks: for safety reasons, retrieval can only be carried out after the tide recedes or the weather improves, significantly limiting retrieval time and hindering fishing efficiency; furthermore, manual retrieval is physically demanding, increasing worker fatigue and negatively impacting their health. In light of these problems, the inventors of this invention conducted in-depth research, leading to this invention. [Summary of the Invention]

[0003] The technical problem to be solved by the present invention is to provide a method for automatically retrieving fish traps from unmanned vessels, which can realize the automatic retrieval of deployed fish traps, thereby solving the problems of limited retrieval time, which is not conducive to improving fishing efficiency, and which increases the fatigue and health of personnel that exist in manual retrieval.

[0004] This invention is implemented as follows: a method for automatically retrieving fish traps from an unmanned vessel, the method comprising the following steps:

[0005] The location information of the fish trap is obtained by the positioning module on the fish trap, and the unmanned vessel is controlled to travel to the location of the fish trap based on the location information of the fish trap.

[0006] Controlling the recovery device on the unmanned vessel to retrieve the fish traps specifically includes:

[0007] The unmanned vessel's recovery device releases a grappling hook to prepare to grab the fish trap. A video monitoring device mounted on the grappling hook captures images of the fish trap. The positional relationship between the grappling hook and the fish trap is analyzed using these images. Based on this relationship, the grappling hook is adjusted to reach the fish trap's location. The grappling hook then grabs the fish trap and pulls it up above the unmanned vessel's railing. The recovery device then rotates the fish trap to the unmanned vessel's designated position, lowers it to the designated position, and releases the fish trap from the grappling hook.

[0008] Furthermore, the step of controlling the unmanned vessel to travel to the location of the fish trap based on the fish trap's location information specifically includes:

[0009] When the unmanned vessel travels to the fish trap deployment area, it obtains information about the coastline of the fish trap deployment area and the location information of each fish trap.

[0010] Based on the coastline information and the location information of each fish trap, the Dijkstra algorithm is used to calculate the shortest recovery path. Specifically, the current position of the unmanned vessel is recorded as the initial node, and the positions of each fish trap are recorded as other nodes. The weights between each other node and the initial node are calculated, and the other node with the lowest weight is selected as the access node. Starting from this access node, the algorithm iteratively calculates the next other node with the lowest weight as the access node, until all other nodes are visited, thus obtaining the shortest recovery path. When calculating the weight between two nodes, if there is a coastline obstructing the two nodes, the length of the coastline is used to replace the obstructed distance.

[0011] The ship's route is generated based on the shortest recovery path, and the unmanned ship is controlled to travel to the location of each fish trap in sequence to recover the fish traps.

[0012] Furthermore, during the process of controlling the unmanned vessel to sequentially navigate to the locations of each fish trap for fish trap retrieval, at least one of the following measures is also included:

[0013] The system detects whether the unmanned vessel deviates from the generated vessel route during its operation. If a deviation from the generated vessel route is detected and reaches a preset distance, the system controls the unmanned vessel to initiate route correction; otherwise, the system does not control the unmanned vessel to initiate route correction.

[0014] The system controls the unmanned vessel to slowly approach the fish trap at a set speed. At the same time, the radar on the unmanned vessel is used to detect obstacles in the surrounding area. When obstacles are detected around the unmanned vessel, the distance between the unmanned vessel and the obstacle is calculated by the radar reflection wave. Based on the distance between the unmanned vessel and the obstacle, the system controls the unmanned vessel to automatically adjust its course.

[0015] Real-time monitoring of water flow and wind direction information, and adjustment of the course and speed of unmanned vessels based on the water flow and wind direction information;

[0016] Preset safety thresholds for various parameters. When a parameter is detected to exceed the set safety threshold, an alarm message will be automatically reported.

[0017] Furthermore, after the grappling hook grabs the fish cage, it is pulled upwards a preset distance underwater, and the tension of the grappling hook's rope is obtained using a tension sensor. If the tension of the rope is greater than the weight of the hook, it means that the grappling hook has successfully grabbed the fish cage; if the tension of the rope is less than or equal to the weight of the hook, it means that the grappling hook has failed to grab the fish cage. At this time, the grappling hook is controlled to adjust its position and continue to grab the fish cage.

[0018] Furthermore, the step of pulling the captured fish trap up to a position higher than the railing of the unmanned vessel specifically includes:

[0019] Once the grappling hook successfully grabs the fish trap, the hook's pull rope is controlled to pull the trap upwards. The video monitoring device on the grappling hook captures images of the trap, and the OpenCV vision library is used to analyze its position. When the analysis indicates the fish trap is about to surface, the pull rope is controlled to gradually increase the tension, continuing to pull the trap upwards. When the analysis indicates the trap is completely out of the water, the hook's pull rope continues to pull it upwards. When the analysis indicates the trap is above the railing of the unmanned vessel, the hook's pull rope stops pulling the trap upwards.

[0020] Furthermore, the process of controlling the retrieval device to release the hook and grab the fish trap also includes:

[0021] Based on the collected water area data, the current water area information of the unmanned vessel is obtained; a mathematical model of the grappling hook under the current water area information is established, and multiple simulation tests are conducted to simulate the grappling hook's grasping situation under the current water area information. In each simulation, the hook's lowering direction is adjusted according to the grasping situation of the previous test, so as to obtain a more stable lowering direction.

[0022] Furthermore, the process of controlling the recovery device on the unmanned vessel to retrieve the fish trap also includes:

[0023] The use of neural network model algorithms to control the recovery device on an unmanned vessel to recover fish traps includes:

[0024] Collect various data related to fish trap retrieval and process the collected data to obtain processed data for model training.

[0025] A neural network model is constructed using Gaussian radial basis functions, and the neural network model is trained using the obtained processed data.

[0026] The trained neural network model is integrated into the control system. The control system controls the recovery device on the unmanned vessel to recover the fish trap based on the sensor data collected in real time by the unmanned vessel and the prediction results output by the neural network model.

[0027] During the recycling process, the performance of the neural network model is evaluated, and the neural network model is continuously improved and optimized based on the evaluation results.

[0028] Furthermore, the positioning module adopts the BeiDou positioning module.

[0029] By adopting the technical solution of the present invention, at least the following beneficial effects are achieved: it enables the automatic retrieval of fish traps by unmanned vessels; compared with the manual retrieval method, on the one hand, it does not need to wait for the tide to recede or the weather to improve before retrieval, which can effectively reduce the limitation on retrieval time and thus help improve fishing efficiency; on the other hand, staff do not need to go to the site for manual retrieval, which can reduce staff fatigue, benefit the health of staff, and is also safer. [Attached Image Description]

[0030] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0031] Figure 1 This is a flowchart illustrating the execution process of an unmanned vessel automatic fish trap recovery method according to the present invention.

[0032] Figure 2 This is a schematic diagram of the recovery device on the unmanned vessel in this invention;

[0033] Figure 3 This is a schematic diagram of how the present invention calculates the course adjustment of unmanned vessels using wind direction as an example.

[0034] Explanation of reference numerals in the attached figures:

[0035] 100 unmanned vessels, 101 railings;

[0036] Recovery device 200, grab hook 201, video monitoring equipment 202, pull rope 203.

Detailed Implementation Methods

[0037] To better understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Please see Figures 1 to 3 As shown in the figure, a preferred embodiment of the method for automatically retrieving fish traps from an unmanned vessel 100 according to the present invention includes the following steps:

[0039] Step S1: Use the positioning module on the fish trap to obtain the location information of the fish trap, and control the unmanned vessel 100 to travel to the location of the fish trap based on the location information of the fish trap.

[0040] Step S2: Control the recovery device 200 on the unmanned vessel 100 to recover the fish traps, that is, use the recovery device 200 to automatically recover the fish traps instead of manually, specifically including:

[0041] The unmanned vessel 100's recovery device 200 releases a grappling hook 201 to prepare to grab a fish trap. A video monitoring device 202 mounted on the grappling hook 201 captures images of the fish trap. The captured images are used to analyze the positional relationship between the grappling hook 201 and the fish trap. Based on this relationship, the position of the grappling hook 201 is adjusted so that it reaches the location of the fish trap. In a specific implementation, when the video monitoring device 202 captures images of the fish trap and transmits them to the control system, the control system can use the OpenCV computing interface to obtain the image information and analyze the current distance and positional relationship between the grappling hook 201 and the fish trap. This allows for convenient adjustment of the grappling hook 201's position, enabling it to continuously approach the fish trap and thus grab it.

[0042] The control hook 201 grabs the fish cage and pulls it up to a position higher than the railing 101 of the unmanned vessel 100, so that the fish cage can be retrieved onto the unmanned vessel 100; the control retrieval device 200 drives the fish cage to rotate to the set position of the unmanned vessel 100, lowers the fish cage to the set position, and controls the control hook 201 to release the fish cage, thereby completing the automatic retrieval of the fish cage.

[0043] By adopting the above-mentioned technical solution of the present invention, it is possible to automatically retrieve fish traps deployed by unmanned vessels 100. Compared with manual retrieval, on the one hand, it is not necessary to wait for low tide or good weather before retrieval, which can effectively reduce the limitation on retrieval time and thus improve fishing efficiency; on the other hand, staff do not need to go to the site for manual retrieval, which can reduce staff fatigue, benefit the health of staff, and is also safer.

[0044] In some embodiments of the present invention, controlling the unmanned vessel 100 to travel to the location of the fish trap based on the fish trap's location information specifically includes:

[0045] When the unmanned vessel 100 reaches the fish trap deployment area, it acquires the coastline information and the location information of each fish trap. Based on the coastline information and the location information of each fish trap, it uses the Dijkstra algorithm to calculate the shortest retrieval path, specifically including:

[0046] The current position of the unmanned vessel 100 is recorded as the initial node (i.e., node 1), and the positions of each fish trap are recorded as other nodes (i.e., nodes 2 to n, where n represents the total number of fish traps in the fish trap placement area). The weight between each other node and the initial node is calculated. Specifically, the distance between two nodes is represented by weight. The other node with the lowest weight (i.e., the shortest edge) is selected as the access node. Starting from the access node, the next other node with the lowest weight is iteratively calculated as the access node until all other nodes are accessed, thus obtaining the shortest recycling path.

[0047] In practice, if there is a coastline blocking the way between two nodes, the unmanned vessel 100 can only travel along the coastline. Therefore, when calculating the weight between two nodes, if there is a coastline blocking the way between two nodes, the length of the coastline is used to replace the blocked distance, and the weight is calculated based on the distance that is replaced by the coastline.

[0048] The ship's route is generated based on the shortest recovery path, and the unmanned ship 100 is controlled to travel to the location of each fish trap in sequence to recover the fish traps.

[0049] This invention acquires the coastline information and the location information of each fish trap in the fish trap deployment area when the unmanned vessel 100 travels to the fish trap deployment area, and calculates the shortest retrieval path using the Dijkstra algorithm. This enables faster retrieval of fish traps in the deployment area, thereby further improving retrieval efficiency. At the same time, it can also reduce the distance traveled by the unmanned vessel 100, which is beneficial to reducing costs.

[0050] In some embodiments of the present invention, since the unmanned vessel 100 is affected by water flow direction, wind direction, etc. during its navigation, in order to ensure that the unmanned vessel 100 can stably and safely approach the fish traps, the process of controlling the unmanned vessel 100 to navigate to the locations of each fish trap in sequence for fish trap retrieval also includes taking at least one of the following measures:

[0051] (1) The system detects whether the unmanned vessel 100 deviates from the generated vessel route during its operation. If a deviation from the generated vessel route is detected and reaches a preset distance (which can be set according to actual needs, for example, 2 meters), the system controls the unmanned vessel to initiate route correction. Otherwise, the system does not control the unmanned vessel to initiate route correction. In a specific implementation of this invention, the unmanned vessel 100 is equipped with a Beidou navigation system. By using a high-precision Beidou navigation system, the system ensures that the unmanned vessel 100 can accurately locate its own navigation position. Thus, during actual use, when the unmanned vessel 100 is detected to have deviated from the generated vessel route and reached a preset distance, the system can control the unmanned vessel 100 to automatically correct its route, enabling the unmanned vessel 100 to accurately and quickly reach the required location.

[0052] (2) Control the unmanned vessel 100 to approach the fish cage slowly at a set speed (the specific speed can be set according to actual needs); In order to avoid large fluctuations and collisions caused by the rapid movement of the unmanned vessel 100, the present invention adopts the method of controlling the unmanned vessel 100 to approach the fish cage slowly at a set speed when the unmanned vessel 100 approaches the fish cage, which helps the unmanned vessel 100 to approach the fish cage better and achieve stable recovery.

[0053] Simultaneously, the radar on the unmanned vessel 100 detects surrounding obstacles. When an obstacle is detected, the distance between the unmanned vessel 100 and the obstacle is calculated using the radar's reflected waves. Based on this distance, the unmanned vessel 100 automatically adjusts its course (including turning angle and speed) to avoid collisions. It should be noted that automatic obstacle avoidance is a function required by existing vessels; therefore, the automatic obstacle avoidance function of the unmanned vessel 100 will not be described in detail here.

[0054] (3) Real-time monitoring of water flow and wind direction information, and adjustment of the unmanned vessel's course and speed based on the water flow and wind direction information; In specific implementation, this invention can calculate the component speed in the same direction based on the wind direction, water flow direction, and the unmanned vessel's sailing angle, and calculate the speed and course to be adjusted based on the sailing angle and speed magnitude, such as Figure 3 As shown, taking wind direction as an example, the following condition must be met: V 风 cosB=V 船 cosD;V 风 cosA=V 船 cosC±V 水 The method for calculating water flow is the same as described above.

[0055] In addition, in specific implementations of the present invention, stabilizers or anti-roll fins or other automatic adjustment devices can be installed on the unmanned vessel 100 to reduce the swaying of the unmanned vessel 100 during navigation.

[0056] (4) Preset safety thresholds for various parameters, such as distance thresholds and speed thresholds. When a parameter is detected to exceed the set safety threshold, an alarm message is automatically reported. The unmanned vessel 100 can take measures automatically based on the alarm message, such as controlling the unmanned vessel 100 to automatically decelerate.

[0057] In some embodiments of the present invention, after the hook 201 grabs the fish trap, the hook 201 is pulled upwards a preset distance underwater, i.e., the output power of the recovery device 200 is increased. The preset upward pulling distance can be set according to actual needs. The tension of the pull rope 203 of the hook 201 is obtained using a tension sensor. If the tension of the pull rope 203 is greater than the weight of the hook 201, it indicates that the hook 201 has successfully grabbed the fish trap; if the tension of the pull rope 203 is less than or equal to the weight of the hook 201, i.e., the tension F of the pull rope 203 is not equal to the weight of the hook 201. 拉 =G 钩 -F 钩 Among them, G 钩 F represents the weight of the grappling hook 201. 钩The buoyancy of the grab hook 201 indicates that the grab hook 201 has failed to grab the fish cage. At this time, the grab hook 201 is controlled to adjust its position and continue to grab the fish cage. Since the grab hook 201 may fail to grab the fish cage during the process of controlling the grab hook 201, the present invention sets a tension sensor on the pull rope 203 of the grab hook 201, and after the grab hook 201 grabs the fish cage, it is pulled upwards underwater a preset distance. Thus, the tension detected by the tension sensor can be used to determine whether the grab hook 201 has successfully grabbed the fish cage.

[0058] In some embodiments of the present invention, in order to better achieve the goal of pulling the fish trap up to a position higher than the railing 101 of the unmanned vessel 100, the specific steps of pulling the captured fish trap up to a position higher than the railing 101 of the unmanned vessel 100 include:

[0059] After the hook 201 successfully grabs the fish trap, the pull rope 203 controls the hook 201 to pull the fish trap upwards; in this invention, when the hook 201 successfully grabs the fish trap and is in a balanced state, the tension F of the pull rope 203 is... 拉 =G 钩 +G 笼 -F 钩 -F 笼 Among them, G 钩 G represents the gravity of the grappling hook 201. 笼 F represents the weight of the fish trap. 钩 F represents the buoyancy of the grab hook 201. 笼 This indicates the buoyancy of the fish trap; in this case, simply increase the tension F of the pull rope 203. 拉 This allows the fish trap to be pulled upwards in the water;

[0060] The video monitoring device 202 on the grab hook 201 captures images of the fish trap, and the OpenCV vision library is used to analyze the position of the fish trap. When the analysis shows that the fish trap is about to float to the surface, the pull rope 203 of the grab hook 201 is controlled to gradually increase the pulling force, that is, to gradually increase the output power of the retrieval device 200, and to continue to pull the fish trap upward. Since the buoyancy of the fish trap and the grab hook 201 will gradually decrease until it disappears during the process of floating to the surface, it is necessary to control the pull rope 203 of the grab hook 201 to gradually increase the pulling force to pull the fish trap above the water surface. When the analysis shows that the fish trap has completely left the water surface, the pull rope 203 of the grab hook 201 is controlled to continue to pull the fish trap upward. When the analysis shows that the fish trap is higher than the railing 101 of the unmanned vessel 100, the pull rope 203 of the grab hook 201 is controlled to stop pulling the fish trap upward, so as to facilitate the control of the retrieval device 200 to rotate the fish trap to the set position of the unmanned vessel 100.

[0061] In some embodiments of the present invention, during the process of controlling the retrieval device 200 to release the hook 201 to grab the fish trap, the following is also included:

[0062] Based on the collected water area data, including water flow information, riverbed information, and coastline information, the current water area information of the unmanned vessel 100 is obtained. A mathematical model of the grappling hook 201 under the current water area information is established. Multiple simulation tests are conducted to simulate the grappling hook 201's grasping behavior under the current water area information. For each simulation, the position of the grappling hook 201 relative to the target fish trap due to water flow is recorded. Furthermore, the hooking direction of the grappling hook 201 is adjusted based on the grasping behavior of the previous test during each simulation to obtain a more stable hooking direction. Since the process of controlling the retrieval device 200 to release the grappling hook 201 to grasp the fish trap may be affected by water flow, this invention uses a mathematical model of the grappling hook 201 and simulates the grappling hook 201's grasping behavior under the current water area information of the unmanned vessel 100 to obtain more stable model output data. This model output data can help the unmanned vessel 100 grasp the fish trap with fewer adjustments.

[0063] In some embodiments of the present invention, since the fish traps are affected by various changes in environmental conditions during the actual recovery process, in order to better cope with various changes in environmental conditions and improve the stability and reliability of the recovery process, the recovery device 200 on the unmanned vessel 100 is further included when controlling the fish traps to be recovered:

[0064] The recovery device 200 on the unmanned vessel 100 is controlled by a neural network model algorithm to recover the fish traps, specifically including:

[0065] First, various data related to fish trap retrieval are collected and processed to obtain processed data for model training; in specific implementation, this invention includes:

[0066] Data Acquisition: Various data related to fish trap retrieval are collected through sensors and other related equipment, such as GPS location, depth, and current velocity information, and the collected data are transmitted to the data processing system; the various sensors and related equipment required can be installed on the unmanned vessel 100.

[0067] Data preprocessing: Cleaning, denoising, and formatting the collected raw data to ensure data accuracy and consistency; at the same time, removing outliers that suddenly appear and have large changes from the previous data, and filling in missing values ​​in blank spaces.

[0068] Feature extraction: Extract features related to fish trap retrieval from the preprocessed data, such as location, speed, and weight. Because the power information output by the system during and after the capture is different, the power output needs to be increased after the capture. At this time, the acquired data will change. The changed values ​​include the fish trap's floating speed, gravity information, etc. The data during this period is also the fish trap feature data.

[0069] Data fusion: Kalman filtering is used to fuse data from different sensors and sources to obtain more accurate and comprehensive information. The filtering algorithm can be implemented using third-party libraries such as JKalmanFilter and kalmanfilter. Specifically, data from sensors located in different positions, such as data collected by sensors at the bow, stern, and sides of the ship, can be fused.

[0070] Data Classification and Recognition: Because each fish trap is located in a different environment, each can be considered an independent entity. Therefore, the Naive Bayes classification algorithm is used to classify and recognize the extracted features to determine the state and location of the fish trap. Specifically, during the classification process, when the retrieval device 200 releases the hook 201 to grab the fish trap, a corresponding mathematical model is established based on the current water area information. When two models are similar, their output data are also similar. Therefore, the Naive Bayes classification algorithm can be used to obtain the probability that the output value of the current water area is the same as the output value of the previous or similar water area, thereby classifying the fish trap.

[0071] State estimation and prediction: Utilizing historical and real-time data, the state of the fish trap and retrieval device 200 is estimated through fused data, and future states are predicted. This helps in predicting and controlling the fish trap retrieval process. During specific evaluation, each retrieval operation is recorded, and the fused data and the number of adjustments are recorded and stored in the database, generating corresponding parameter curves. Data deviating from the error range is adjusted, and the aforementioned data acquisition and processing steps are repeated to generate new data. In specific prediction, the output power data for the next location can be predicted based on historical data and current data trends. This helps reduce the system's computational consumption. When reaching the next location, only data acquisition and comparison are needed. If the data is similar to historical data, that data can be used for the retrieval operation; if the data model for that location is not similar to any historical location, data extraction is then performed, effectively reducing performance loss and improving operational efficiency.

[0072] Anomaly Detection: By comparing actual data with expected or historical data, anomalies (such as sensor malfunctions, actuator malfunctions, etc.) are detected. Anomaly detection can be achieved using pattern recognition technology. Anomaly identification methods include, but are not limited to: comparing currently acquired data with historical data; when the error range is large, calculating the average of multiple acquired data; or excluding individual sensors from the calculation; if a sensor is excluded from the calculation and performs normally, it may indicate a problem with that sensor, triggering an alert; if the position of the hook 201 does not change when it is lowered or retrieved, it indicates a problem with the retrieval device 200, triggering an alarm; if a discrepancy is detected between the fish cage and the positioning information, it may indicate that the positioning module of the fish cage has detached, triggering an alert; and so on.

[0073] Data storage and recording: The processed and analyzed data is stored in the database for subsequent analysis and querying; at the same time, key operations and events such as fetching and failure, and fetching but not moving are recorded in detail to facilitate troubleshooting and accountability.

[0074] Visualization and Report Generation: The results of processing and analysis are visualized in the form of charts, reports, etc., which makes it easier for operators to understand and monitor the recycling process; at the same time, relevant reports and documents can be generated as needed, and key information can be extracted to issue information prompts to relevant personnel.

[0075] Secondly, a neural network model is constructed using Gaussian radial basis functions (RBFs), and trained using the obtained processed data. By adjusting the weights and biases of the neural network model, it can accurately predict the movement state of the fish trap. In specific training, different training methods such as supervised learning, unsupervised learning, or reinforcement learning can be employed. For example, water area information and equipment output information can be used, and supervised learning can be employed to simulate and learn how to capture fish in each water area. By inputting water area information and equipment output information, the machine simulates the process and selects "Yes" or "No" based on the results, thereby optimizing the result set. The Gaussian radial basis function has the characteristic that the closer to the target, the larger the output value of the function. Therefore, by setting the fish trap position as the target center position and using the distance between the hook 201 and the fish trap position each time it is lowered as a parameter, the Gaussian radial basis function shows that the output value is maximized at a certain point. At this point, the equipment output parameters correspond to the water area parameters, resulting in the optimal output. This invention uses Gaussian radial basis functions to construct the neural network model, thus achieving faster learning speed.

[0076] Third, the trained neural network model is integrated into the control system. Based on the sensor data collected in real time by the unmanned vessel and the prediction results output by the neural network model, the control system controls the recovery device on the unmanned vessel to recover the fish traps. At the same time, the parameters of the neural network model can be adjusted and optimized online through the information fed back by the sensors to adapt to changes in different environments and conditions. In addition, based on the prediction results output by the neural network model, control strategies can be customized. By adjusting the parameters or weights in the control strategy, precise control of stable fish trap recovery can be achieved.

[0077] Fourth, during the recovery process, the performance of the neural network model is evaluated, and the model is continuously improved and optimized based on the evaluation results to enhance the stability and reliability of the recovery process. Performance indicators may include control accuracy, response time, and stability. Specifically, during performance evaluation, each recovery process is recorded, and information such as empty hooks and stuck hooks is investigated. When an empty hook occurs, the corresponding water area data and equipment output data are recorded, and the neural network model is further trained to improve accuracy and reduce the probability of empty hooks. For situations where the system starts but reacts too slowly (e.g., reaction time exceeds 2 or 3 seconds), log entries are made to facilitate subsequent investigation of the cause; and so on.

[0078] In some embodiments of the present invention, in order to improve positioning accuracy and enable the unmanned vessel 100 to accurately approach the fish trap, the positioning module adopts the Beidou positioning module.

[0079] While specific embodiments of the present invention have been described above, those skilled in the art should understand that the specific embodiments described are merely illustrative and not intended to limit the scope of the present invention. Equivalent modifications and variations made by those skilled in the art in accordance with the spirit of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method of automatically retrieving a fishnet by an unmanned ship, characterized by: The method comprises the following steps: The position information of the fish cage is obtained by using the positioning module on the fish cage, and the unmanned ship is controlled to travel to the position of the fish cage according to the position information of the fish cage; The fish cage is recycled by the recycling device on the unmanned ship, specifically including: The hook is released by the recycling device on the unmanned ship to prepare to grab the fish cage, the image of the fish cage is captured by the video monitoring device arranged on the hook, the positional relationship between the hook and the fish cage is analyzed by using the captured image of the fish cage, the position of the hook is adjusted according to the positional relationship between the hook and the fish cage, so that the hook reaches the position of the fish cage; the fish cage is grabbed by the hook, the grabbed fish cage is pulled up to a position higher than the rail of the unmanned ship, the fish cage is rotated to a set position of the unmanned ship by the recycling device, the grabbed fish cage is lowered to the set position, and the hook is released from the fish cage; In the process of controlling the recycling device to release the hook to grab the fish cage, it also includes: according to the collected water area data, the water area information of the current position of the unmanned ship is obtained; a mathematical model of the hook under the current water area information is established, the current water area information is simulated to test the grabbing situation of the hook multiple times, and the hook direction of the hook is adjusted according to the grabbing situation of the previous test each time, so as to obtain a stable hook direction; In the process of recycling the fish cage by the recycling device on the unmanned ship, it also includes: using a neural network model algorithm to control the recycling device on the unmanned ship to recycle the fish cage, specifically including: Various data related to fish cage recycling are collected, and the collected data are processed to obtain processed data for model training; a neural network model is constructed using Gaussian radial basis function, and the neural network model is trained using the obtained processed data; the trained neural network model is integrated into the control system, the control system controls the recycling device on the unmanned ship to recycle the fish cage according to the real-time sensor data collected by the unmanned ship and the prediction results output by the neural network model; in the process of recycling, the performance of the neural network model is evaluated, and the neural network model is continuously improved and optimized according to the evaluation results; wherein, the processing of the collected data includes: Feature extraction: extracting features related to fish cage recycling from preprocessed data; Data fusion: using Kalman filtering method to fuse data from different sensors and different sources; Data classification and identification: using Naive Bayes classification algorithm to classify and identify the extracted features.

2. A method of automatically retrieving a fishing net from an unmanned vessel as claimed in claim 1, wherein: The control of the unmanned ship to travel to the position of the fish cage according to the position information of the fish cage specifically includes: When the unmanned ship travels to the fish cage throwing area, the coastline information of the fish cage throwing area and the position information of each fish cage are obtained; According to the coastline information and the position information of each fish cage, the shortest recovery path is calculated by using the Dijkstra algorithm, specifically including: recording the current position of the unmanned ship as an initial node, recording the positions of each fish cage as other nodes, calculating the weight between each other node and the initial node, obtaining the other node with the lowest weight as an access node, and continuing to iterate to calculate the next other node with the lowest weight as the access node from the access node until all other nodes are accessed, so that the shortest recovery path is obtained; when calculating the weight between two nodes, if there is a coastline block between the two nodes, the length of the coastline is used to replace the distance of the blocked section; According to the shortest recovery path, a ship driving route is generated, and the unmanned ship is controlled to drive to the positions of each fish cage in turn to recover the fish cages.

3. A method of automatically retrieving a fishing net from an unmanned vessel as claimed in claim 2, wherein: In the process of controlling the unmanned ship to drive to the positions of each fish cage in turn to recover the fish cages, at least one of the following measures is taken: Detecting whether the unmanned ship deviates from the generated ship driving route during driving, and if it is detected that the deviation from the generated ship driving route reaches a preset distance, controlling the unmanned ship to start route correction; otherwise, not controlling the unmanned ship to start route correction; Controlling the unmanned ship to slowly approach the fish cage at a set speed; at the same time, detecting the surrounding obstacles by using the radar on the unmanned ship, and when detecting that there are obstacles around the unmanned ship, calculating the distance between the unmanned ship and the obstacles by the reflection wave of the radar, and controlling the unmanned ship to automatically adjust the heading according to the distance between the unmanned ship and the obstacles; Real-time monitoring of water flow related information and wind direction related information, and adjusting the heading and speed of the unmanned ship according to the water flow related information and the wind direction related information; Presetting safety thresholds of various parameters, and when detecting that a parameter exceeds the set safety threshold, automatically reporting an alarm information.

4. A method of automatically retrieving a fishing net from an unmanned vessel according to claim 1, wherein: After controlling the grappling hook to grab the fish cage, the grappling hook is pulled upward underwater by a preset distance, and the tension of the grappling hook's pull rope is obtained by using a tension sensor, and if the tension of the pull rope is greater than the weight of the grappling hook, it means that the grappling hook successfully grabs the fish cage; if the tension of the pull rope is less than or equal to the weight of the grappling hook, it means that the grappling hook fails to grab the fish cage, at which time the grappling hook is controlled to adjust the position and continue to grab the fish cage.

5. A method of automatically retrieving a fishing net from an unmanned vessel as claimed in claim 1, wherein: The upward pulling of the grabbed fish cage to a position above the rail of the unmanned ship specifically includes: After the grappling hook successfully grabs the fish cage, the pull rope of the grappling hook is controlled to pull the fish cage upward; the fish cage image is captured by the video monitoring device on the grappling hook, and the position state of the fish cage is analyzed by using the OpenCV vision library, and when it is analyzed that the grabbed fish cage is about to float out of the water, the pull rope of the grappling hook is controlled to gradually increase the tension value and continue to pull the grabbed fish cage upward; when it is analyzed that the grabbed fish cage completely separates from the water surface, the pull rope of the grappling hook is controlled to continue to pull the fish cage upward; when it is analyzed that the grabbed fish cage is higher than the rail position of the unmanned ship, the pull rope of the grappling hook is controlled to stop pulling the fish cage upward.

6. A method of automatically retrieving a fishing net from an unmanned vessel as claimed in claim 1, wherein: The positioning module adopts a Beidou positioning module.

Images