Marine monitoring system and method based on big data mining

Abstract

The invention relates to the technical field of monitoring, in particular to an ocean monitoring system and method based on big data mining, and the method comprises the steps: obtaining the traveling track and moving time data of a fishing boat generated by a positioning system arranged on the fishing boat; if it is judged that the staying time of the fishing boat in a certain area exceeds a preset threshold value based on the traveling track and the moving time data, marking the fishing boat as a suspicious boat, and collecting marine environment data corresponding to the area; when the suspicious ship is subsequently confirmed to be subjected to the fishing activity, generating a fishing data set according to the marine environment data collected in the corresponding period; a machine learning algorithm is adopted to train a marine monitoring point prediction model based on the fishing data set; and based on the marine environment data obtained through real-time monitoring, adopting a marine monitoring point prediction model to predict an excessive fishing risk area and generating early warning information. Early warning can be performed on the fishing position in the ocean according to the fishing data of the ship, so that a key area can be identified in advance, and the supervision efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of monitoring technologies, and particularly to an ocean monitoring system and method based on big data mining. Background Art

[0002] Monitoring of overfishing by fishing boats in ocean fishery resources aims to ensure that fishing activities are carried out within a sustainable range to prevent the depletion of fishery resources and the destruction of the ecosystem caused by overfishing. Through monitoring, key information such as fishing intensity, catch volume, and species composition can be timely grasped, thereby providing a basis for formulating scientific fishery management policies.

[0003] Existing monitoring methods have certain limitations. Traditional methods based on fishing boat logs and on-board inspections are not only time-consuming and laborious, but also rely on the fishermen's self-awareness and accuracy, which may lead to inaccurate data or delayed reporting, thus making it impossible to effectively supervise fishing. Summary of the Invention

[0004] The purpose of the present invention is to provide an ocean monitoring system and method based on big data mining, aiming to be able to give early warnings about fishing locations in the ocean based on the fishing data of boats, so as to be able to mark key areas in advance and improve the supervision efficiency.

[0005] To achieve the above purpose, in the first aspect, the present invention provides an ocean monitoring method based on big data mining, including obtaining the travel trajectory and movement time data of a fishing boat generated by a positioning system installed on the fishing boat;

[0006] Based on the travel trajectory and movement time data, if the stay time of the fishing boat in a certain area exceeds a predetermined threshold, it is marked as a suspicious boat, and the corresponding ocean environmental data of this area is collected;

[0007] When it is subsequently confirmed that the suspicious boat conducts fishing activities, the ocean environmental data collected during the corresponding period is generated into a fishing data set;

[0008] Based on the fishing data set, a machine learning algorithm is used to train an ocean monitoring point prediction model;

[0009] Based on the ocean environmental data obtained from real-time monitoring, the ocean monitoring point prediction model is used to predict overfishing risk areas and generate warning information.

[0010] Among them, the specific steps of obtaining the travel trajectory and movement time data of the fishing boat generated by the positioning system installed on the fishing boat include:

[0011] Obtaining the travel trajectory and movement time data generated by the GPS module;

[0012] When the GPS signal is lost, a monitoring network formed by multiple buoy-type acoustic monitoring units set at the monitoring location is used to assist in positioning the fishing boat.

[0013] Among them, the specific steps of using the monitoring network formed by multiple buoy-type acoustic monitoring units set at the monitoring location to assist in positioning the fishing boat when the GPS signal is lost include:

[0014] Set multiple buoy-type acoustic monitoring units in the monitoring water area;

[0015] Synchronize the working times of multiple buoy-type acoustic monitoring units;

[0016] The buoy-type acoustic monitoring unit sends an interrogation signal to the fishing boat. After each buoy-type acoustic monitoring unit receives the acoustic signal, it records the feedback information of the signal arrival. The feedback information includes time information and intensity information;

[0017] Use the feedback information received from at least three buoy-type acoustic monitoring units in different positions to apply the triangulation algorithm to calculate the position of the fishing boat.

[0018] Among them, the specific steps of determining that if the fishing boat stays in a certain area for more than a predetermined threshold based on the driving trajectory and moving time data, marking it as a suspicious vessel, and collecting the corresponding marine environmental data of the area include:

[0019] Generate a monitoring grid based on the positions of the buoy-type acoustic monitoring units;

[0020] Continuously obtain the driving trajectory and moving time data of the fishing boat, and record the time intervals of its entry into and departure from each monitoring grid;

[0021] If the fishing boat stays in a certain monitoring grid for more than the preset threshold, mark the vessel as a suspicious vessel;

[0022] Obtain the position information of the suspicious vessel and collect the marine environmental data of the monitoring grid. The marine environmental data includes water temperature, salinity, dissolved oxygen, pH value, and sea current velocity.

[0023] Among them, the specific steps of generating a fishing dataset from the marine environmental data collected during the corresponding period when the suspicious vessel is subsequently confirmed to be engaged in fishing activities include:

[0024] Obtain the image data of the vessel when leaving the port and the image data when returning to the port;

[0025] Use the edge detection algorithm to identify and locate the draft lines on both sides of the vessel;

[0026] After identifying the draft lines, calculate the draft depth before leaving the port and the draft depth after arriving at the port according to the positions of the standard draft lines marked on the hull;

[0027] Calculate the difference between the draft before departure and the draft after arrival. If it exceeds the preset value, it is considered that fishing activities have been carried out;

[0028] Collect the marine environmental data of the monitoring grid where the corresponding vessel stays for more than the predetermined threshold and generate a fishing dataset.

[0029] Among them, the specific steps of training the marine monitoring point prediction model using the machine learning algorithm based on the fishing dataset include:

[0030] Process the missing values and outliers in the fishing dataset and unify the data format;

[0031] Divide the dataset into a training set and a test set;

[0032] Use the training set to train the random forest model to obtain the marine monitoring point prediction model.

[0033] Among them, the specific steps of predicting the overfishing risk area and generating warning information using the marine monitoring point prediction model based on the real-time monitored marine environmental data include:

[0034] Clean the collected real-time data to remove outliers and incorrect readings;

[0035] Input the preprocessed real-time marine environmental data and fishing vessel activity data into the pre-trained marine monitoring point prediction model to obtain the fishing risk value of the monitoring grid in a future period;

[0036] When the fishing risk value exceeds the warning value, corresponding warning information is generated.

[0037] In a second aspect, the present invention also provides a marine monitoring system based on big data mining, including a fishing vessel position acquisition module, a suspicious vessel judgment module, a marine data collection module, a model training module, and a warning module;

[0038] The fishing vessel position acquisition module is used to acquire the travel trajectory and movement time data of the fishing vessel generated by the positioning system set on the fishing vessel;

[0039] The suspicious vessel judgment module is used to judge that if the fishing vessel stays in a certain area for more than the predetermined threshold based on the travel trajectory and movement time data, mark it as a suspicious vessel, and collect the corresponding marine environmental data of the area;

[0040] The marine data collection module is used to generate a fishing dataset for the marine environmental data collected during the corresponding period when the suspicious vessel is subsequently confirmed to carry out fishing activities;

[0041] The model training module is used to train a marine monitoring point prediction model based on a fishing dataset using a machine learning algorithm;

[0042] The early warning module is used to predict overfishing risk areas based on the marine environment data obtained from real-time monitoring using the marine monitoring point prediction model and generate early warning information.

[0043] A marine monitoring system and method based on big data mining according to the present invention. In modern marine monitoring systems, each fishing boat is equipped with positioning devices (such as GPS), which can record the precise position and navigation route of the fishing boat in real time. The system regularly receives the data generated by these positioning systems through wireless communication technology, including the driving trajectory of the fishing boat and its staying time at different geographical locations. Based on the obtained driving trajectory and moving time data, the system will automatically analyze to determine whether there are abnormal behaviors. Specifically, if a fishing boat stays in a certain specific area for a time exceeding a preset time threshold, the fishing boat will be marked as a suspicious vessel. At this time, the system will further start the marine environment monitoring program for this area to collect key parameters such as water temperature, salinity, and water flow velocity. When it is confirmed that the suspicious vessel is engaged in fishing operations, the system will record the relevant marine environment data. Using the previously generated fishing dataset and combining machine learning algorithms, the system can train a prediction model specifically for marine monitoring points. This process involves a large amount of complex data processing and pattern recognition work, aiming to find patterns from historical data to improve the prediction accuracy of future situations. Such a prediction model is crucial for understanding the long-term impact of fishing activities on the marine ecosystem. Using the already trained marine monitoring point prediction model and combining real-time updated marine environment data, dynamically evaluate the risk levels of overfishing in each sea area. Once a high-risk area is found, the system will automatically generate early warning information and timely notify relevant departments to take necessary protection measures to prevent overexploitation and damage of marine resources. This way not only improves the supervision efficiency but also provides strong support for realizing sustainable marine management. Description of the Drawings

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0045] Figure 1 It is a flowchart of a marine monitoring method based on big data mining according to the present invention.

[0046] Figure 2It is a flowchart of obtaining the travel trajectory and movement time data of a fishing boat generated by the positioning system set on the fishing boat according to the present invention.

[0047] Figure 3 It is a flowchart of assisting in positioning a fishing boat by using a monitoring network formed by a plurality of buoy-type acoustic monitoring units set at a monitoring location when the GPS signal is lost according to the present invention.

[0048] Figure 4 It is a flowchart of the present invention for determining that if the stay time of a fishing boat in a certain area exceeds a predetermined threshold based on the travel trajectory and movement time data, marking it as a suspicious vessel, and collecting the corresponding marine environment data of the area.

[0049] Figure 5 It is a flowchart of the present invention for generating a fishing dataset from the marine environment data collected during the corresponding period when a suspicious vessel is subsequently confirmed to be engaged in fishing activities.

[0050] Figure 6 It is a flowchart of training a marine monitoring point prediction model based on the fishing dataset by using a machine learning algorithm according to the present invention.

[0051] Figure 7 It is a flowchart of predicting an overfishing risk area and generating a warning message based on the marine environment data obtained from real-time monitoring by using a marine monitoring point prediction model according to the present invention. Specific Embodiment

[0052] The embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present invention and should not be construed as limiting the present invention.

[0053] First Embodiment

[0054] Please refer to Figures 1 to 7 , the present invention provides a marine monitoring method based on big data mining, including:

[0055] S101 Obtain the travel trajectory and movement time data of a fishing boat generated by the positioning system set on the fishing boat;

[0056] The specific steps include:

[0057] S201 Obtain the travel trajectory and movement time data generated by the GPS module;

[0058] First, install one or more GPS receivers on the fishing boat, and these devices are responsible for receiving signals from the global positioning system satellites.

[0059] The GPS module works continuously and records the position coordinates (longitude, latitude) and accurate timestamps of the fishing boat at regular intervals (e.g., every second).

[0060] The captured data can be either stored locally on the storage medium of the fishing boat or sent in real-time to a remote server via wireless communication technology for further processing and analysis.

[0061] When the GPS signal is lost, S202 uses a monitoring network formed by multiple buoy-type acoustic monitoring units set at the monitoring location to assist in positioning the fishing boat.

[0062] When the fishing boat enters an area with poor or no GPS signal reception, a backup solution is needed to determine the position of the fishing boat. The specific steps include:

[0063] S301 Set up multiple buoy-type acoustic monitoring units in the monitoring waters;

[0064] According to the monitoring requirements, deploy a certain number of buoy-type acoustic monitoring units within the predetermined waters. These buoys should be evenly distributed to cover as large an area as possible and ensure that at least three buoys can monitor the fishing boat simultaneously.

[0065] S302 Synchronize the working times of multiple buoy-type acoustic monitoring units;

[0066] To ensure that the information received from different buoys can be accurately used for position calculation, high-precision time synchronization must be maintained among all buoy-type acoustic monitoring units. Usually, the GPS time when the signal can be received is used as the reference standard.

[0067] S303 The buoy-type acoustic monitoring unit sends an interrogation signal to the fishing boat. After each buoy-type acoustic monitoring unit receives the acoustic signal, it records the feedback information on the signal arrival, and the feedback information includes time information and intensity information;

[0068] Each buoy will periodically send an interrogation signal. After the receiver on the fishing boat receives these signals, it will make a response. Subsequently, each buoy records information such as the specific time and signal intensity when the response signal is received.

[0069] S304 Use the feedback information received from at least three buoy-type acoustic monitoring units at different positions and apply the triangulation algorithm to calculate the position of the fishing boat.

[0070] Based on the signal arrival time and intensity information provided by at least three buoys, use the triangulation method or other appropriate algorithms to estimate the exact position of the fishing boat. This method relies on the propagation speed of sound waves in water and the time difference between different buoys for positioning.

[0071] S102 Based on the driving trajectory and moving time data, if the staying time of a fishing boat in a certain area exceeds a predetermined threshold, it is marked as a suspicious vessel, and the corresponding marine environment data of this area is collected;

[0072] The specific steps include:

[0073] S401 Generate a monitoring grid based on the location of the buoy-type acoustic monitoring unit;

[0074] First, based on the location information of the deployed buoy-type acoustic monitoring units, establish a virtual monitoring grid system. These grids cover all waters to be monitored, ensuring no blind spots.

[0075] Each grid represents a specific-sized sea area, and the specific size depends on the requirements of monitoring accuracy and the distance between buoys. For example, it can be set that the side length of each grid is 1 kilometer, so as to accurately track the activities of fishing boats in different grids.

[0076] S402 Continuously obtain the driving trajectory and moving time data of the fishing boat, and record the time intervals when it enters and leaves each monitoring grid;

[0077] Use the GPS module or the buoy-type acoustic monitoring network to continuously track the driving path and speed of the fishing boat, and record the exact time when the fishing boat enters and leaves each monitoring grid. For each fishing boat, the system automatically records the timestamps of its entry and exit from each grid, so as to calculate the specific duration of the fishing boat staying in each grid.

[0078] S403 If the staying time of a fishing boat in a certain monitoring grid exceeds the preset threshold, mark the vessel as a suspicious vessel;

[0079] According to the characteristics of different sea areas and management requirements, set a reasonable upper limit of staying time for each monitoring grid as the warning threshold. For example, in some sensitive areas, it may be stipulated that fishing boats are not allowed to stay continuously for more than 4 hours. Once the system detects that the staying time of a certain fishing boat in a specific grid exceeds the set threshold, it immediately automatically marks it as a suspicious vessel and triggers the subsequent data collection procedure.

[0080] S404 Obtain the location information of the suspicious vessel and collect the marine environment data of this monitoring grid, where the marine environment data includes water temperature, salinity, dissolved oxygen, pH value, and sea current speed.

[0081] Obtain the exact location information of the vessel marked as suspicious to ensure that its location in the monitoring grid can be accurately identified. Initiate a comprehensive monitoring of the marine environment in the target monitoring grid. This includes but is not limited to:

[0082] Water temperature measurement: Use a temperature sensor to record the current water temperature.

[0083] Salinity measurement: Measure the salt concentration of seawater using a conductivity meter or other specialized equipment.

[0084] Dissolved oxygen content: Determine the dissolved oxygen content in water using chemical titration or electrochemical sensors.

[0085] pH value detection: Measure the acidity and alkalinity of seawater using a pH meter.

[0086] Analysis of sea current velocity: Evaluate the velocity and direction of water flow with the help of tools such as an Acoustic Doppler Current Profiler (ADCP).

[0087] S103 When the suspected vessel is subsequently confirmed to be engaged in fishing activities, generate a fishing dataset from the marine environmental data collected during the corresponding period;

[0088] The specific steps include:

[0089] S501 Obtain the image data of the vessel when leaving the port and when returning to the port;

[0090] Use equipment installed at the port or drones, etc., to take high-definition images of the fishing vessel when leaving the port and when returning to the port respectively. These images need to cover the side of the hull, especially the area near the waterline. Ensure that the obtained images are accurately synchronized with the time stamps of the GPS positioning system so that the position information of the vessel can be accurately associated during subsequent analysis.

[0091] S502 Use edge detection algorithms to identify and locate the waterlines on both sides of the vessel;

[0092] Preprocess the obtained images (such as denoising, contrast enhancement, etc.) to improve the accuracy of edge detection. Use the Canny edge detection algorithm to identify the contour features on the hull, especially the position of the waterline.

[0093] Due to the complex and changeable marine environment, the images of fishing vessels taken may be affected by noise interference, such as factors like wave reflection and light changes. To reduce the impact of these factors on subsequent processing steps, filtering techniques are usually used to remove the noise in the images. Common methods include Gaussian filtering, median filtering, etc. Gaussian filtering can effectively smooth the image while retaining important edge information; while median filtering is particularly suitable for removing salt-and-pepper noise. To better highlight the difference between the hull and the background and thus improve the effect of edge detection, contrast enhancement of the image is required. This step can be completed through histogram equalization or Adaptive Histogram Equalization (AHE).

[0094] Then, the Sobel operator is used to calculate the gradient magnitude and direction of each pixel in the image. This step is to determine where the edges might be. The edge width is refined through non-maximum suppression, that is, only the maximum response value along the gradient direction of each pixel is retained as a potential edge point. Two thresholds, high and low, are applied to distinguish strong edges from weak edges. Only when a weak edge is connected to a strong edge will it be retained, which can effectively reduce the appearance of false edges. The edge points that meet the conditions are tracked and connected to form a complete edge contour. During this process, special attention is paid to the lines under the hull as they represent the position of the waterline.

[0095] After identifying the waterline, S503 calculates the pre-departure draft depth and the post-arrival draft depth according to the position of the standard waterline marked on the hull.

[0096] Detailed registration needs to be carried out for each fishing boat, including the specific position of its designed waterline. This is usually clearly marked on the outside of the hull. This data can be obtained by referring to the technical documents of the ship or directly measuring on the hull and digitally recorded to form a database. This database not only contains data on the designed waterline but also can include basic information about the vessel, historical navigation records, etc.

[0097] Using the aforementioned image preprocessing (denoising, contrast enhancement, etc.) and the Canny edge detection algorithm, the hull contour is accurately extracted from the captured fishing boat image, especially focusing on the part under the hull that might represent the waterline. During this process, it is necessary to ensure that the results of edge detection are accurate enough to avoid misjudgment. Color segmentation technology can be combined because the color of the waterline part may be different from other parts, thereby improving the detection accuracy.

[0098] Once the position of the actual waterline is determined, it is then necessary to compare it with the known designed waterline. This step involves converting the position of the actual waterline into a distance value relative to a fixed reference point on the hull, such as the distance from the bow or a certain feature point. Then, this distance value is compared with the theoretical distance value corresponding to the designed waterline.

[0099] For the convenience of calculation, a local coordinate system can be established on the hull with a certain fixed feature point as the origin, such as the very front of the bow. In this way, both the actual waterline and the designed waterline can be represented by specific numerical values in the coordinate system.

[0100] S504 calculates the difference between the pre-departure draft depth and the post-arrival draft depth. If it exceeds the preset value, it is considered that fishing activities have been carried out.

[0101] A reasonable draft depth change threshold is preset according to factors such as ship type and deadweight. This threshold represents the maximum allowable change range under normal circumstances when fishing operations are not involved.

[0102] If the calculated draft depth difference exceeds this preset threshold, it is initially inferred that the vessel has carried out fishing activities.

[0103] S505 collects the marine environmental data of the monitoring grid where the corresponding vessel stays for more than the predetermined threshold and generates a fishing dataset.

[0104] Once the fishing behavior is confirmed, immediately extract all relevant marine environmental data (such as water temperature, salinity, dissolved oxygen, pH value, sea current speed, etc.) within the monitoring grid where the vessel is located during the suspected fishing period from the previously recorded data.

[0105] Combine the above marine environmental parameters with the specific time and geographical location where the fishing behavior occurs to form a complete fishing dataset. This dataset not only helps to understand the impact of fishing activities on the local ecosystem but also provides a scientific basis for fishery management.

[0106] S104 trains a marine monitoring point prediction model using machine learning algorithms based on the fishing dataset;

[0107] The specific steps include:

[0108] S601 processes the missing values and outliers in the fishing dataset and unifies the data format;

[0109] First, evaluate the degree of missing values in the dataset. For a small amount of missing data, methods such as mean filling, median filling, or nearest neighbor interpolation can be used to complete it; if the missing proportion is too high, corresponding records or features need to be considered for deletion.

[0110] Apply statistical methods (such as the 3σ principle) or distance-based methods (such as DBSCAN) to identify and process outliers. For outliers caused by measurement errors, reasonable methods can be used to correct them; for real but extreme data points, they should be treated with caution to avoid simply removing them to prevent loss of valuable information. Ensure the consistency of all data items, including date and time formats, numerical units, etc. For example, convert all date and time fields to a unified timestamp format and ensure that numerical variables are expressed in the same unit.

[0111] S602 divides the dataset into a training set and a test set;

[0112] Data is usually randomly assigned to the training set and the test set according to a certain ratio (such as 7:3 or 8:2). To ensure the generalization ability of the model, especially in the case of time series features, it is recommended to use the time splitting method, that is, using the data in the earlier time period as the training set and the subsequent data as the test set. To further improve the reliability of model performance evaluation, the K-fold cross-validation technique can be introduced. On the basis of dividing the training set and the test set, the training set is further subdivided for multiple trainings and validations, so as to obtain more robust model evaluation indicators.

[0113] S603 uses the training set to train the random forest model to obtain the ocean monitoring point prediction model.

[0114] Analyze the importance of each feature variable and select the most relevant features according to their contribution degrees for input into the model. For example, environmental parameters such as water temperature, salinity, dissolved oxygen level, etc. and relevant indicators of vessel fishing behavior may need to be considered. Use the random forest algorithm for training. Random forest is an ensemble learning method that forms the final prediction result by constructing multiple decision trees and integrating their results. During the training process, adjust the key hyperparameters (such as the number of trees, maximum depth, etc.) to optimize the model performance. Use the test set to evaluate the trained random forest model, mainly examining indicators such as accuracy, recall rate, F1 score, etc. At the same time, draw the ROC curve and calculate the AUC value to comprehensively understand the classification effect of the model. According to the evaluation results, if it is found that the model performs poorly, it can be optimized by adjusting hyperparameters, increasing the data volume or improving feature engineering, etc. until satisfactory prediction accuracy is achieved.

[0115] S105 uses the ocean monitoring point prediction model to predict the overfishing risk area and generate early warning information based on the real-time monitored ocean environment data.

[0116] The specific steps include:

[0117] S701 cleans the collected real-time data to remove outliers and incorrect readings;

[0118] First, perform preliminary processing on the real-time data obtained from various sensors (including but not limited to water temperature, salinity, dissolved oxygen, pH value, sea current speed, etc.) and the fishing vessel positioning system. This step includes identifying and removing data points that are clearly illogical, such as values outside the reasonable range or incorrect readings due to equipment failures. Use statistical methods (such as the Z-score method), rule-based methods or machine learning techniques to automatically detect outliers. For the detected outliers, they can be interpolated and filled based on the data at adjacent time points, or reasonable default values can be set according to professional knowledge for replacement. Ensure the consistency of all input data formats, such as converting the date and time fields to the standard timestamp format and unifying the units of numerical variables, etc., for subsequent processing.

[0119] S702 inputs the pre - processed real - time marine environmental data and fishing vessel activity data into a pre - trained marine monitoring point prediction model to obtain the fishing risk value of the monitoring grid for a certain period in the future;

[0120] Combine the pre - processed marine environmental data with fishing vessel activity data (such as location, speed, operation status, etc.) to form a comprehensive data set for predictive analysis. Input the above - mentioned comprehensive data into a pre - trained random forest marine monitoring point prediction model according to the format required by the model. This model is trained based on historical data and can predict the fishing risk level in each monitoring grid for a certain period in the future according to the current and recent marine environmental conditions and fishing vessel activity. The model outputs the specific fishing risk values for each monitoring grid, and these values reflect the likelihood of overfishing in the corresponding area.

[0121] S703 generates corresponding warning information when the fishing risk value exceeds the warning value.

[0122] According to the characteristics, ecological sensitivity and management requirements of different sea areas, set appropriate fishing risk warning thresholds for each monitoring grid. Once the predicted risk value exceeds this threshold, it indicates that there is a high risk of overfishing in this area. When it is found that the fishing risk value of a certain monitoring grid exceeds the standard, the system automatically generates a warning notice containing detailed information (such as geographical location, expected impact range, recommended countermeasures, etc.). These information can be conveyed to relevant management personnel, fishermen's organizations and other stakeholders through various channels. Establish a rapid response mechanism to ensure that actions can be taken promptly after receiving the warning information, such as strengthening patrol supervision, adjusting fishing quotas or issuing temporary fishing bans, etc., to prevent overfishing and protect the marine ecological environment.

[0123] Second Embodiment

[0124] The present invention also provides an ocean monitoring system based on big data mining, which includes a fishing boat position acquisition module, a suspicious boat judgment module, an ocean data collection module, a model training module, and an early warning module; the fishing boat position acquisition module is used to obtain the travel trajectory and movement time data of the fishing boat generated by the positioning system installed on the fishing boat; the suspicious boat judgment module is used to judge that if the stay time of the fishing boat in a certain area exceeds a predetermined threshold based on the travel trajectory and movement time data, it is marked as a suspicious boat, and the corresponding ocean environment data of the area is collected; the ocean data collection module is used to generate a fishing dataset for the ocean environment data collected during the corresponding period when the suspicious boat is subsequently confirmed to be engaged in fishing activities; the model training module is used to train an ocean monitoring point prediction model based on the fishing dataset using a machine learning algorithm; the early warning module is used to predict an overfishing risk area and generate an early warning message using the ocean monitoring point prediction model based on the ocean environment data obtained by real-time monitoring.

[0125] In this embodiment, the fishing boat position acquisition module plays a crucial role. It uses the positioning device (such as GPS) equipped on the fishing boat to track and record the travel trajectory and movement time data of the fishing boat in real time. This not only helps to understand the operation range of the fishing boat, but also provides basic information for subsequent data analysis. The suspicious boat judgment module conducts in-depth analysis based on the above-obtained position and time data. Once it is found that the stay time of a certain fishing boat in a specific area exceeds the preset safety threshold, the system will automatically mark it as a suspicious boat and immediately start detailed monitoring of the ocean environment in that area. This mechanism is particularly effective for identifying illegal fishing behavior, because excessive stay in a certain area may be an important indicator of overfishing.

[0126] When the suspicious boat is further confirmed to be engaged in fishing activities, the ocean data collection module comes into play. It is responsible for collecting all relevant ocean environment parameters involved in this process, such as water temperature, salinity, dissolved oxygen content, etc., and integrating this information into a detailed fishing dataset. These datasets are crucial for understanding the impact of fishing activities on the marine ecosystem.

[0127] The model training module uses a machine learning algorithm to train a prediction model based on the previously collected fishing dataset. The early warning module uses the real-time updated ocean environment data and combines the already trained ocean monitoring point prediction model to dynamically evaluate the risk levels of overfishing in different sea areas. Once a potential risk area is found, the system will automatically generate an early warning message to remind the relevant departments to take timely actions to protect the marine resources from irreversible damage.

[0128] In summary, this ocean monitoring system based on big data mining realizes the full-process automated management from fishing boat tracking to environmental monitoring and then to risk warning by integrating a variety of advanced technical means, providing strong technical support for the realization of a sustainable marine economy.

[0129] The above-disclosed is only a preferred embodiment of the present invention. Of course, it cannot be used to limit the scope of the rights of the present invention. Those of ordinary skill in the art can understand all or part of the processes of the above embodiments, and the equivalent changes made according to the claims of the present invention still fall within the scope covered by the invention.

Claims

1. An ocean monitoring method based on big data mining, characterized in that: It includes: Obtain the travel trajectory and movement time data of fishing boats generated by the positioning system installed on the fishing boats; Based on the travel trajectory and movement time data, if the stay time of a fishing boat in a certain area exceeds a predetermined threshold, mark it as a suspicious boat and collect the corresponding ocean environment data of this area; When it is subsequently confirmed that the suspicious boat is engaged in fishing activities, generate a fishing dataset from the ocean environment data collected during the corresponding period; Based on the fishing dataset, use machine learning algorithms to train an ocean monitoring point prediction model; Based on the ocean environment data obtained through real-time monitoring, use the ocean monitoring point prediction model to predict overfishing risk areas and generate warning information.

2. The ocean monitoring method based on big data mining according to claim 1, characterized in that: The specific steps for obtaining the travel trajectory and movement time data of fishing boats generated by the positioning system installed on the fishing boats include: Obtain the travel trajectory and movement time data generated by the GPS module; When the GPS signal is lost, use the monitoring network formed by multiple buoy-type acoustic monitoring units set at the monitoring location to assist in positioning the fishing boat.

3. The ocean monitoring method based on big data mining according to claim 2, characterized in that: The specific steps for using the monitoring network formed by multiple buoy-type acoustic monitoring units set at the monitoring location to assist in positioning the fishing boat when the GPS signal is lost include: Set multiple buoy-type acoustic monitoring units in the monitoring waters; Synchronize the working times of multiple buoy-type acoustic monitoring units; The buoy-type acoustic monitoring unit sends an interrogation signal to the fishing boat. After each buoy-type acoustic monitoring unit receives the acoustic signal, record the feedback information of the signal arrival. The feedback information includes time information and intensity information; Use the feedback information received from at least three different-position buoy-type acoustic monitoring units to apply the triangulation algorithm to calculate the position of the fishing boat.

4. The ocean monitoring method based on big data mining according to claim 3, characterized in that: The specific steps for judging that if the stay time of a fishing boat in a certain area exceeds a predetermined threshold based on the travel trajectory and movement time data, mark it as a suspicious boat and collect the corresponding ocean environment data of this area include: Generate a monitoring grid based on the location of the buoy-type acoustic monitoring unit; Continuously obtain the travel trajectory and movement time data of the fishing boat, and record the time intervals for it to enter and leave each monitoring grid; If the fishing boat stays in a certain monitoring grid for more than the preset threshold, mark the boat as a suspicious boat; Obtain the location information of the suspicious boat and collect the ocean environment data of this monitoring grid. The ocean environment data includes water temperature, salinity, dissolved oxygen, pH value, and sea current speed.

5. The ocean monitoring method based on big data mining according to claim 4, characterized in that: The specific steps for generating a fishing dataset from the ocean environment data collected during the corresponding period when it is subsequently confirmed that the suspicious boat is engaged in fishing activities include: Obtain the image data of the boat when leaving the port and the image data when returning to the port; Use the edge detection algorithm to identify and locate the draft lines on both sides of the boat; After identifying the waterline, calculate the draft depth before leaving the port and the draft depth after arriving at the port according to the standard waterline position marked on the hull; Calculate the difference between the draft depth before leaving the port and the draft depth after arriving at the port. If it exceeds the preset value, it is considered that fishing activities have been carried out; Collect the marine environment data of the monitoring grid where the corresponding ship stays for more than the predetermined threshold and generate a fishing dataset.

6. A marine monitoring method based on big data mining according to claim 5, characterized in that The specific steps of training the marine monitoring point prediction model using the machine learning algorithm based on the fishing dataset include: Process the missing values and outliers in the fishing dataset and unify the data format; Divide the dataset into a training set and a test set; Use the training set to train the random forest model to obtain the marine monitoring point prediction model.

7. A marine monitoring method based on big data mining according to claim 6, characterized in that The specific steps of predicting the overfishing risk area and generating early warning information using the marine monitoring point prediction model based on the real-time monitored marine environment data include: Clean the collected real-time data to remove outliers and incorrect readings; Input the preprocessed real-time marine environment data and fishing boat activity data into the pre-trained marine monitoring point prediction model to obtain the fishing risk value of the monitoring grid in a future period; Generate corresponding early warning information when the fishing risk value exceeds the early warning value.

8. A marine monitoring system based on big data mining, adopting a marine monitoring method based on big data mining according to any one of claims 1 to 7, characterized in that It includes a fishing boat position acquisition module, a suspicious boat judgment module, a marine data collection module, a model training module and an early warning module; The fishing boat position acquisition module is used to obtain the travel trajectory and movement time data of the fishing boat generated by the positioning system set on the fishing boat; The suspicious boat judgment module is used to judge that if the fishing boat stays in a certain area for more than the predetermined threshold based on the travel trajectory and movement time data, mark it as a suspicious boat, and collect the corresponding marine environment data of the area; The marine data collection module is used to generate a fishing dataset from the collected marine environment data during the corresponding period when the suspicious boat is subsequently confirmed to carry out fishing activities; The model training module is used to train the marine monitoring point prediction model using the machine learning algorithm based on the fishing dataset; The early warning module is used to predict the overfishing risk area and generate early warning information using the marine monitoring point prediction model based on the real-time monitored marine environment data.

Images