Method and system for simulating fish school conditions in Zhoushan fishery under the influence of red tide

Abstract

The present invention discloses a method and system for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tides. The simulation method includes: obtaining red tide data to be predicted; inputting the distribution of the red tide data to be predicted into a red tide prediction model to obtain the spatial distribution of the red tide, wherein the red tide prediction model is constructed using a back-propagation neural network and trained using a training set containing data before and after the red tide occurs; obtaining the corresponding red tide type based on the red tide spatial distribution and average chlorophyll concentration; constructing a fish school behavior model, performing fish school simulation based on the fish school behavior model and the corresponding red tide type, and outputting the simulated fish school situation. The present invention can truly reflect the dynamic characteristics of fish schools under the influence of red tides.

Description

Technical Field

[0001] The invention belongs to the technical field of fish school simulation, and in particular relates to a method and system for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide. Background Art

[0002] The development and utilization of fishery resources and environmental protection have always attracted attention. The red tide phenomenon not only has a serious impact on the ocean structure, but also frequently destroys the original fishery structure and has a huge impact on fishery resources. The existing fish simulation system fails to fully consider the impact of environmental factors such as red tides on fish behavior and survival status; in addition, the existing red tide prediction and fish simulation systems are often independent of each other and lack effective integration, making it difficult to form a complete set of comprehensive simulation methods and systems that can be used for fishery resource management and protection. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention proposes a method and system for simulating the fish school situation in Zhoushan fishing grounds under the influence of red tide, which can truly reflect the dynamic characteristics of fish schools under the influence of red tide.

[0004] The present invention provides a method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide, comprising:

[0005] Obtain red tide data to be predicted;

[0006] Inputting the red tide data distribution to be predicted into a red tide prediction model to obtain the spatial distribution of the red tide, wherein the red tide prediction model is constructed by a back propagation neural network and obtained by training with a training set, wherein the training set includes data before and after the red tide occurs;

[0007] Obtain the corresponding red tide type based on the spatial distribution of red tide and the average chlorophyll concentration;

[0008] A fish school behavior model is constructed, fish school simulation is performed based on the fish school behavior model and the corresponding red tide type, and the fish school simulation situation is output.

[0009] Optionally, obtaining the training set includes:

[0010] Obtain data for a preset number of days before and after a red tide occurs;

[0011] The data are integrated into a space×time dataset, and NaN values ​​in the integrated dataset are replaced to obtain the training set.

[0012] Optionally, the fish school behavior model includes: a fish school swimming model and a fish school escape model;

[0013] The fish school swimming model is used to update the fish school's position and speed based on the fish school's current position, speed and model parameters;

[0014] The fish escape model is used to update the position and speed of the fish according to the current position and speed of the fish, the current position and speed of the predator and the model parameters.

[0015] Optionally, based on the current position, speed, and model parameters of the fish school, updating the fish school position and speed includes:

[0016] Based on the current position and velocity of the school of fish, a global alignment velocity is calculated based on the average velocity of the school of fish;

[0017] After calculating the global alignment speed, the approach speed, the gathering speed, and the repulsion speed of the fish school are initialized, the perception range of each fish's current position is traversed, the number of neighboring fish is found according to the perception distance, and the current fish is judged to be approaching or repelling, thereby obtaining a first judgment result;

[0018] Calculate the gathering force after the combined force according to the first judgment result and the number of neighboring fish;

[0019] Calculating the combined velocity of each fish in the school of fish based on the gathering force;

[0020] Based on the combined velocity, the fish school position and velocity are updated.

[0021] Optionally, the method to calculate the total speed of each fish in the school is:

[0022] Total speed = global alignment weight × global alignment speed + convergence weight × convergence speed + convergence weight × convergence force +

[0023] Repulsion weight × repulsion speed.

[0024] Optionally, updating the fish school's position and speed based on the fish school's current position and speed, the predator's current position and speed, and model parameters includes:

[0025] According to the current position and speed of the fish school, the current position and speed of the predator and the model parameters, a vector directly escaping from the predator is calculated as the escape direction;

[0026] judging the escape behavior according to the escape direction;

[0027] Based on the escape behavior, determining the distance between a certain fish in the school of fish and the predator to obtain a second determination result;

[0028] According to the second judgment result, the position and speed of the fish school are updated.

[0029] Optionally, updating the position and speed of the school of fish according to the second judgment result includes:

[0030] If a fish in the school is within the predator's hunting radius, an escape factor is obtained based on the distance between the fish in the school and the predator, and based on the escape factor, the position and speed of the fish are determined;

[0031] If the distance between a fish in the school and the predator is within a preset range, and the preset range is much smaller than the predation radius, random avoidance is performed, and a randomly generated angle is substituted into the rotation matrix to update the position and speed of the fish school.

[0032] Optionally, according to the escape factor, the method based on the position and speed of the fish school is:

[0033] Fish school speed = fish school initial speed + escape factor × (escape direction - fish school initial speed);

[0034] To perform random avoidance, randomly generate angles and substitute them into the rotation matrix. The method to update the position and speed of the fish school is as follows:

[0035] Update the speed of the fish school = rotation matrix × initial speed of the fish school.

[0036] Optionally, before performing the fish school simulation based on the fish school behavior model and the corresponding red tide type, the method further includes:

[0037] Initialize the food intake, excretion, feeding rate, mortality rate of the entire fish population and the reproduction rate of plants, as well as the food intake and excretion of individual sample fish.

[0038] The present invention also provides a fish school situation simulation system in Zhoushan fishing grounds under the influence of red tide, comprising: a red tide prediction module and a fish school visualization simulation module;

[0039] The red tide prediction module is used to input the distribution of red tide data to be predicted into a red tide prediction model to obtain the spatial distribution of red tides, wherein the red tide prediction model is constructed by a back propagation neural network and obtained by training a training set, and the training set includes data before and after the red tide occurs;

[0040] The fish school visualization simulation module is used to obtain the corresponding red tide type according to the red tide spatial distribution and the average chlorophyll concentration; construct a fish school behavior model, perform fish school simulation based on the fish school behavior model and the corresponding red tide type, and output the fish school simulation situation.

[0041] Compared with the prior art, the present invention has the following advantages and technical effects:

[0042] The present invention trains a backpropagation neural network, generates predictions for different years, and can simulate fish schools based on different red tide conditions. Within the fish school simulation interface, the present invention defines the basic movement patterns and states of fish schools, accurately reflecting the dynamic characteristics of fish schools under the influence of red tides. The present invention organically integrates red tide prediction and fish school simulation functions to create a comprehensive simulation system with comprehensive functionality and ease of operation. Furthermore, while existing technologies may involve complex operational steps and parameter settings, the present invention offers an intuitive user interface and a simple operational process. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0044] Figure 1 4 is a flow chart of a method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide according to an embodiment of the present invention;

[0045] Figure 2 2024 red tide prediction results according to an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of the initialization of fish distribution according to an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the final iteration of a school of fish according to an embodiment of the present invention;

[0048] Figure 5 Schematic diagram of the basic framework of the platform according to an embodiment of the present invention;

[0049] Figure 6 1 is a schematic diagram of the prediction results on the fifth day after the occurrence of the red tide in the spring of 2003 according to an embodiment of the present invention;

[0050] Figure 7 1 is a schematic diagram of the prediction results on the third day after the occurrence of the red tide in the spring of 2004 according to an embodiment of the present invention;

[0051] Figure 8 2. It is a schematic diagram of a simulation parameter description interface according to an embodiment of the present invention;

[0052] Figure 9 is a schematic diagram of a fish school simulation result according to an embodiment of the present invention;

[0053] Figure 10 is a schematic diagram of a sample fish simulation result according to an embodiment of the present invention;

[0054] Figure 11 It is a schematic diagram of the overall interface operation result of an embodiment of the present invention. DETAILED DESCRIPTION

[0055] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0056] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0057] This embodiment proposes a method for simulating the fish school situation in Zhoushan fishing grounds under the influence of red tide. Figure 1 As shown, the specific steps include:

[0058] Obtain red tide data to be predicted;

[0059] The red tide data distribution to be predicted is input into the red tide prediction model to obtain the spatial distribution of the red tide. The red tide prediction model is constructed by a back propagation neural network and obtained by training with a training set that includes data before and after the red tide occurs.

[0060] Obtain the corresponding red tide type based on the spatial distribution of red tide and the average chlorophyll concentration;

[0061] Construct a fish school behavior model, perform fish school simulation based on the fish school behavior model and the corresponding red tide type, and output the fish school simulation situation.

[0062] Furthermore, obtaining a training set includes:

[0063] Obtain data for a preset number of days before and after a red tide occurs;

[0064] Integrate the data into a space × time dataset, replace NaN values ​​in the integrated dataset, and obtain the training set.

[0065] Furthermore, the fish school behavior model includes: a fish school swimming model and a fish school escape model;

[0066] Fish school swimming model, used to update the fish school's position and speed based on the current position, speed and model parameters of the fish school;

[0067] The fish escape model is used to update the position and speed of the fish school based on the current position and speed of the fish school, the current position and speed of the predator, and the model parameters.

[0068] Furthermore, updating the position and speed of the fish school according to the current position, speed and model parameters of the fish school includes:

[0069] Based on the current position and speed of the fish school, the global alignment speed is calculated according to the average speed of the fish school;

[0070] After calculating the overall alignment speed, the approach speed, gathering speed, and repulsion speed of the fish school are initialized. The perception range of each fish's current position is traversed. The number of neighboring fish is found based on the perception distance. The first judgment result is determined whether the current fish is approaching or repelling.

[0071] Based on the first judgment result and the number of neighboring fish, the gathering force after the combined force is calculated;

[0072] Calculate the combined speed of each fish in the school based on the gathering force;

[0073] Based on the combined velocity, update the fish position and velocity.

[0074] Specifically, this function simulates the movement of fish based on approach, aggregation and repulsion. By inputting the current position, speed and various model parameters mentioned above of the fish school, it outputs the updated new position and speed of the fish school. I stipulate that every time the distribution position and speed of the fish school are input, the average speed direction of all the fish schools is the global alignment direction of the entire fish school. The function will traverse the information of each fish (including the neighboring fish within the perception range of each fish), determine the position of all neighboring fish through the perception range, and judge whether the fish will approach or repel in the next step. After the judgment, the gathering force after the joint force is calculated based on the number of neighboring fish. To calculate the combined speed, there is a formula:

[0075] Total speed = global alignment weight × global alignment speed + convergence weight × convergence speed + convergence weight × convergence force +

[0076] Repulsion weight × repulsion speed;

[0077] Furthermore, updating the position and speed of the fish school according to the current position and speed of the fish school, the current position and speed of the predator, and the model parameters includes:

[0078] According to the current position and speed of the fish school, the current position and speed of the predator and the model parameters, a vector directly escaping from the predator is calculated as the escape direction;

[0079] Determine the escape behavior based on the escape direction;

[0080] Based on the escape behavior, the distance between a certain fish in the school of fish and the predator is determined to obtain a second determination result;

[0081] Based on the second judgment result, the position and speed of the fish school are updated.

[0082] Further, based on the second judgment result, updating the fish school position and speed includes:

[0083] If a fish in the school is within the predator's hunting radius, the escape factor is obtained based on the distance between the fish in the school and the predator. Based on the escape factor, the position and speed of the fish are determined;

[0084] If the distance between a fish in the school and the predator is within a preset range, and the preset range is much smaller than the predation radius, random avoidance is performed, and a randomly generated angle is substituted into the rotation matrix to update the position and speed of the fish school.

[0085] Specifically, this function is based on the escape patterns of a school of fish. It takes as input the current position and velocity of the school of fish, the current position and velocity of the predator, and the various model parameters mentioned above, and outputs the updated position and velocity of the school of fish. The escape pattern of the school of fish is determined by the perception range and the actual distance between the school of fish and the predator. When the school of fish is within the predator's hunting radius, it adjusts its speed to consistently escape from the predator. When the school of fish is within the random avoidance radius, it adopts a random escape direction. When the school of fish is outside the predator's hunting radius, it swims according to the school of fish swimming model function.

[0086] Furthermore, before performing fish school simulation based on the fish school behavior model and the corresponding red tide type, the following steps are also included:

[0087] Initialize the food intake, excretion, feeding rate, mortality rate of the entire fish population and the reproduction rate of plants, as well as the food intake and excretion of individual sample fish.

[0088] This embodiment also provides a system for simulating fish school conditions in Zhoushan fishery under the influence of red tide, including: a red tide prediction module and a fish school visualization simulation module;

[0089] The red tide prediction module is used to input the red tide data distribution to be predicted into the red tide prediction model to obtain the spatial distribution of the red tide. The red tide prediction model is constructed by a back propagation neural network and obtained by training a training set, which includes data before and after the red tide occurs.

[0090] The fish school visualization simulation module is used to obtain the corresponding red tide type based on the spatial distribution of red tide and the average chlorophyll concentration; build a fish school behavior model, perform fish school simulation based on the fish school behavior model and the corresponding red tide type, and output the fish school simulation situation.

[0091] The following is combined with Figure 2-11 This embodiment is described in detail:

[0092] The technical solution of this embodiment includes three parts: back propagation neural network training data set, Matlab fish school simulation code writing, and Matlab Appdesigner platform construction. The following will explain these three aspects in detail.

[0093] In recent years, the rapid development of machine learning and neural networks has enabled researchers to better analyze the dynamics of marine ecosystems. While various methods are currently available for predicting red tides, neural network methods have been widely used in various nonlinear predictions due to their powerful nonlinear fitting and self-learning capabilities. This example, based on a backpropagation neural network method, trained a 26-year dataset and output the spatial distribution of red tides five days after they occurred, accurately capturing the occurrence patterns and spatial distribution characteristics of red tides.

[0094] 1. Data Description: This paper uses daily data from the Copernicus satellite observations of chlorophyll a mass concentration in seawater. The selected time range is April, May, and June 1998 to April, May, and June 2024; the spatial range is longitudes 121–123 and latitudes 30–31, with a grid size of 49 × 26.

[0095] 2. Dataset Processing: This paper predicts the spatial distribution of red tides five days after the occurrence of the spring of 2024. Therefore, the 10 days before the red tides occurred from the spring of 1998 to the spring of 2023 are used as the input data set, and the 5 days after the red tides occurred from the spring of 1998 to the spring of 2023 are used as the output data set for training. According to the China Ocean Information Network and the National Marine Environmental Forecast Center, the specific dates of red tides in the spring of 1998-2023 are obtained. For example, on June 16, 2005, a red tide occurred in the waters of Zhoushan, Zhejiang, with a maximum area of ​​approximately 1,000 square kilometers. The main red tide organisms were Thalassiosira circumflexa and Skeletonema costatum. The data from June 6, 2004 to June 15, 2004 were used as the input data for the 10 days before the red tide, and the data from June 16, 2004 to June 20, 2004 were used as the output data for the 5 days after the red tide. Similarly, for the 26 years from 1998 to 2023, the input dataset size is 49 × 26 × 10 × 26, and the output dataset size is 49 × 26 × 5 × 26. 49 × 26 represents the spatial distribution, while 10 × 26 and 5 × 26 represent the temporal distribution. To facilitate training, we combine them into a spatial × temporal dataset, with an input dataset size of 1274 × 260 and an output dataset size of 1274 × 130.

[0096] 3. Training the Neural Network: The input dataset is 1274×260 pixels, containing 1274 feature points and 260 samples. The output dataset is 1274×130 pixels, containing 1274 feature points and 130 samples. It's important to note that NaN values ​​cannot be present in backpropagation neural network training. Therefore, NaN values ​​for continents must be excluded. Here, all NaN values ​​are replaced with 0. In Matlab, training is performed using the Neural Network Fitting Tool 14.4, and the trained neural network is output.

[0097] 4. 2024 Red Tide Forecast Results: The data of the first 10 days of the 2024 spring red tide are integrated into space × time data, input into the trained neural network, and then the output data is restored to three dimensions, namely longitude × latitude × 5 days. The results are as follows: Figure 2 shown.

[0098] 2. Matlab fish school simulation code writing

[0099] When it comes to fish school simulation, researchers typically use ecological principles and behavioral theories to construct fish school behavior models using computer simulation technology. These models simulate the movement patterns and distribution characteristics of fish in natural environments, such as adaptability, predation, reproduction, and mortality. This paper uses Matlab simulation technology to simulate fish schools in Zhoushan fishing grounds, reflecting the behavioral states of fish schools under the influence of red tides. The dynamic changes of the entire school are simulated by analyzing the behavior of individual fish and the entire school. The fish schools here follow basic swimming patterns, providing an intuitive and dynamic visualization process.

[0100] 1. Model Description: The "organisms" involved in this invention include two types: fish schools and "predators". The predators here can refer to organisms that actually prey on fish schools, or they can refer to external factors that randomly threaten individual fish schools, leading to a sudden drop in energy or death.

[0101] (1) Fish school swimming model: The artificial fish school algorithm proposes that the fish school swimming process will follow four swimming rules and ultimately achieve global optimization:

[0102] ① Each fish will move closer to the center of its neighbors and keep in line with their direction;

[0103] ② When fish are too close, they will repel each other to avoid collisions with their neighbors;

[0104] ③ With a certain fish as the center, neighboring fish within the sensing range act as a small school, tending to gather together and maintain the same direction;

[0105] ④ The entire school of fish will have a globally aligned direction, and at the end of the iteration the entire school of fish will move in that direction together.

[0106] (2) Fish escape model: Here I stipulate that fish follow three escape rules:

[0107] ① When a fish senses a predator within its perception range, an escape factor is calculated based on the ratio of the distance between the perception range and the predator. Under this escape factor, the fish will deviate from its previous swimming direction to escape.

[0108] ② If a fish is very close to a predator, it will perform a random evasive maneuver, and its direction will be randomly deflected (some fish may be very close to the predator at the beginning, and die due to the sudden loss of energy caused by random avoidance);

[0109] ③If a fish is outside the escape radius, it maintains its current direction and speed and follows the swimming model.

[0110] 2. Parameter settings are shown in Table 1:

[0111] Table 1

[0112]

[0113]

[0114] 3. Code Logic: The fish school simulation code of this invention mainly consists of three functions: the fish school swimming model function, the fish school escape model function, and the 3D visualization dynamic simulation drawing function. The following will introduce each function from the perspective of model introduction and code structure.

[0115] (1) Fish school swimming model function:

[0116] ① Model introduction: This function simulates the movement of fish based on approach, aggregation and repulsion. By inputting the current position, speed and various model parameters mentioned above of the fish school, it outputs the updated new position and speed of the fish school. I stipulate that every time the distribution position and speed of the fish school are input, the average speed direction of all the fish schools is the global alignment direction of the entire fish school. The function will traverse the information of each fish (including the neighboring fish within the perception range of each fish), determine the position of all neighboring fish through the perception range, and judge whether the fish will approach or repel in the next step. After the judgment, the gathering force after the joint force is calculated based on the number of neighboring fish. To calculate the combined speed, there is a formula:

[0117] Total speed = global alignment weight × global alignment speed + convergence weight × convergence speed + convergence weight × convergence force +

[0118] Repulsion weight × repulsion speed;

[0119] It should be noted that after the speed is converged at each step in this function, the speed needs to be normalized and then substituted into the swimming speed range of the fish school.

[0120] ②Code structure: First, obtain the current position and speed of all fish schools, and calculate the global alignment speed from the average speed; initialize the approach speed, gathering speed, and repulsion speed of the fish school; traverse the perception range of the current position of the fish school, and find the number of neighboring fish according to the distance. If there is a fish that is too close to the fish, it is affected by a repulsion factor, which is determined by the ratio of the difference between the minimum distance and the actual distance; if they are all outside the collision distance, the ratio of the gathering speed of the small fish school within the perception range to the number of neighboring fish is used as the aggregation force. Here, the repulsion speed and gathering force should be normalized to avoid accumulation; finally, the combined speed of each fish is calculated according to the above model. Note that the fish school speeds must be normalized after calculation, and then the maximum and minimum swimming speeds of the fish school in the model parameters are used to limit them.

[0121] (2) Fish escape model function:

[0122] ① Model Introduction: This function is based on the escape patterns of a school of fish. It takes as input the current position and velocity of the school of fish, the current position and velocity of the predator, and the various model parameters mentioned above, and outputs the updated position and velocity of the school of fish. The escape pattern of the school of fish is determined by the perception range and the actual distance between the school of fish and the predator. When the school of fish is within the predator's hunting radius, it adjusts its speed to consistently escape from the predator. When the school of fish is within the random avoidance radius, it adopts a random escape direction. When the school of fish is outside the predator's hunting radius, it swims according to the school of fish swimming model function.

[0123] ② Code structure: First, obtain the current position and speed of all fish and the current position and speed of the predator. First, calculate the vector that directly escapes from the predator as the escape direction and normalize the vector. Then, determine the escape behavior. If the fish is within the predator's hunting radius, use the ratio of the difference between the perception range and the actual distance between the fish and the predator as the escape factor to update the speed of the fish. The formula is as follows:

[0124] Fish school speed = fish school initial speed + escape factor × (escape direction - fish school initial speed)

[0125] If the fish is within the random avoidance radius, the escape direction is not accumulated. Instead, a random avoidance is performed. The randomly generated angle is substituted into the rotation matrix to update the speed of the fish school. The formula is as follows:

[0126] Update the speed of the fish school = rotation matrix × initial speed of the fish school'

[0127] (3) 3D visualization dynamic simulation drawing function:

[0128] ① Model introduction: This function is a drawing function. By inputting the current position and speed of the fish school and the current position and speed of the predator, the fish school behavior is drawn with each update iteration, presenting a three-dimensional dynamic effect.

[0129] ② Code structure: First, input the current position and speed of the fish school, and the current position and speed of the predator; clear the previous graphics window content, set the 3D view and proportional coordinate axes, draw the predator point and fish school vector in this window, and dynamically draw with each iteration of information input.

[0130] ③ Visual display of fish school model: First, the fish school and its speed direction will be randomly distributed, such as Figure 3 As shown in , as the number of iterations of the fish school increases, most of the fish schools will follow the fish school gathering and global alignment in the swimming process, and the final direction tends to be unified, as shown in Figure 4 shown.

[0131] 3. Matlab Appdesigner platform construction

[0132] This invention, developed using the MATLAB Appdesigner tool, can be exported as a standalone app. It organically integrates red tide prediction and fish school simulation for the first time, creating a comprehensive simulation system with comprehensive functionality and ease of operation. This system effectively simulates fish school conditions in Zhoushan fishing grounds under the influence of red tide, enabling information integration. Compared to independent red tide models and fish school simulations, this invention provides visual analysis for both fishery resource management and environmental protection, helping to reduce management costs and improve decision-making efficiency.

[0133] 1. Construction of the basic framework of the platform:

[0134] The present invention simulates the fish school situation in Zhoushan fishing grounds under red tide, so it will be divided into four modules, namely red tide prediction module, fish school visualization simulation module, fish school simulation result module, and explanation and operation.

[0135] The red tide prediction module requires a direct comparison of the predicted and actual spatial distribution visualization interfaces. In addition to predicting the red tide distribution in the spring of 2024, the neural network of the present invention can also perform a red tide prediction view for other years in the training set, output different red tide distributions and chlorophyll concentrations, and simulate different fish school simulation effects and results for the sea area where the red tide occurs based on the different chlorophyll concentrations output; in addition to visual dynamic simulation, the fish school simulation interface can also build a view of the simulation results of individual fish and the entire fish school, so that users can have a more complete understanding of the fish school simulation process. The fish school simulation result interface shows the situation of individual fish and the entire fish school, including feeding conditions, excretion conditions, degree of influence of red tides, and current status; in addition, there are different types of red tides occurring in the sea area. I have divided three different types of red tides, namely other red tides, harmful red tides and toxic red tides. For different types of red tides, the fish school will finally present different simulation results; the final description and operation module introduces the method technology, use and concise operation process of the invention.

[0136] Existing red tide prediction and ecological simulation technologies may involve complex operation steps and parameter settings. Overall, this invention has an intuitive user interface and operation process, fully considering the impact of environmental factors such as red tides on fish behavior and survival status, and can adjust the feeding intensity, excretion intensity, mortality rate, etc. of fish under different red tide types. Figure 5 shown.

[0137] 2. Module control logic:

[0138] According to the platform construction, the present invention has four modules: red tide prediction module, fish school visualization simulation module, fish school simulation result module, and instructions and operations. The instructions and operations are only for the purpose of demonstrating the invention description and operation process to enhance user understanding, and do not contain actual control logic, so they are not considered. The red tide prediction module uses a dataset predicted by a backpropagation neural network as a "red tide spatial distribution library." This library contains red tides from the five training years and the 2024 red tide. The spatial distribution of the 2024 red tide is primarily based on the five days following the 2024 red tide. Based on this, the average chlorophyll a concentration (a) is calculated as a key parameter for fish school simulation. The fish school visualization simulation first uses the previously described model to ensure that the fish follow basic swimming rules during the visualization process. The addition of threatening external factors, namely "predators," increases the randomness of behavioral mutations in the fish's habitat. This external factor can cause a sudden drop in fish energy, or even death. This sudden drop in fish energy can lead to increased phytoplankton intake. If the red tide is harmful or toxic, the sudden drop in fish energy will lead to a more severe red tide due to excessive feeding. The fish school simulation result module uses the chlorophyll a concentration during the previous red tide as a key parameter, enabling information transfer between red tide prediction and fish school simulation. Finally, throughout the control process, different simulation effects are achieved for the three different red tide types based on subsequent assumptions.

[0139] Each of my modules relies on buttons to operate. Therefore, to perform predictions or simulations, I need to add callback functions to the button code. Simply call the name of a coordinate axis, plot on that axis, and press the button to perform the prediction or simulation. The control logic for the first three modules will be explained below.

[0140] (1) Red tide prediction module:

[0141] The red tide prediction module has only two parts of code logic: one is the "Random Verification Set" button, and the other is the "24-year Forecast" button. Add callback functions to these two buttons and start writing.

[0142] ① For the random validation set, you first need to load the neural network output, input data, and real chlorophyll a concentration data. The neural network requires a two-dimensional space × time dataset as input. This input should be processed externally or in the callback code after input. Note that NaN values ​​cannot be present during backpropagation neural network training; I replaced all NaN values ​​with 0. After processing, the neural network is fed into the output, which is a two-dimensional space × time data. The dataset is then processed back to a longitude × latitude × 5 days × 26 years structure. The upper left corner shows the specific days 1–5. Before visualizing the spatial distribution of chlorophyll, the value of this "day modifier" is read, corresponding to the specific day in the longitude × latitude × (1, 2, 3, 4, 5) days × 26 years. Regarding the selection of years, I only selected the predictions for 5 years (1998, 2000, 2003, 2010, and 2014). These years have different spatial distribution characteristics and chlorophyll concentrations, which can produce different simulation effects for subsequent fish simulation work. These years are numbered and randomly extracted. The specific extracted years are converted into strings. The year text of the spatial distribution coordinate axis is updated to the string each time the "Random Validation Set" is pressed, and the two-dimensional data of longitude × latitude of the year and the predicted days are selected for plotting. The processing of the actual spatial distribution data of chlorophyll a concentration is simpler, because it is also longitude × latitude × 5 days × 26 years data. Just select the corresponding year and the longitude × latitude spatial distribution data of the corresponding days. The results of running the "Random Validation Set" button are as follows. Figure 6 As shown in the figure, the forecast is the result of the fifth day after the red tide occurred in the spring of 2003.

[0143] ② The principle of the "24-year prediction" button and the "Random Validation Set" button is the same. You only need to switch the data set. The "Random Validation Set" button needs to load 26 years of input data sets and 26 years of real data. The "24-year prediction" button only needs the input data sets of 10 days before the red tide in 2024 and the real data sets of 5 days after the red tide. Similarly, the dimensions of the input data set should be processed as space × time, and the NaN values ​​of the continent should be removed before the neural network prediction. After the prediction is completed, the dimensions are modified to longitude × latitude × 5 days, and finally the specific number of days in the "day adjuster" is read to select the longitude × latitude spatial distribution of days 1 to 5. Figure 7 Shown is the spatial distribution of red tide readings on the third day.

[0144] (2) Fish school visualization simulation module:

[0145] The red tide prediction module has 5 buttons, but only one button has code logic, which is the "Start" button. The "Pause", "Continue", and "Close" buttons only need to execute uiwate, uiresume, and delete commands on the entire panel. There is no actual control logic, so they are not considered. The "Simulation Parameter Description" is to write a new platform interface externally, write the simulation parameters used for fish simulation, and there is no callback function. After it is built, you only need to add a button in the main callback to call it. The simulation parameter description interface is as follows Figure 8 shown.

[0146] The "Start" button requires the most information to be called up, including the type of red tide, overall simulation information of fish schools, including fish feeding rate, fish mortality rate, total fish feeding volume, fish excretion volume, plant reproduction rate, and sample fish simulation information, including the current feeding volume of sample fish, the current excretion volume of sample fish, the degree of impact of red tide on sample fish, and the current status of sample fish.

[0147] Before the simulation begins, the average chlorophyll-a concentration obtained from the spatial distribution of the red tide forecast will be read. This value will serve as an important basis for the overall fish intake and plant reproduction rate. First, the overall fish intake, excretion, feeding rate, mortality rate, and plant reproduction rate will be initialized. The intake and excretion of individual sample fish will be accumulated based on the initialization at each iteration. I have proposed four assumptions about fish feeding, and the subsequent fish feeding and excretion simulation will be based on these assumptions:

[0148] Assumption 1: Each fish is of the same size and weight, approximately 80g, and eats at a constant rate;

[0149] Hypothesis 2: Each fish eats about 5% of its body weight, 50% of which comes from phytoplankton. The mass of chlorophyll a accounts for 1% to 2% of the dry weight of phytoplankton, and 90% of phytoplankton is water.

[0150] Assumption 3: The situation of Zhoushan fishery will be based on the simulated fish population within the simulation area of ​​200×200×200.

[0151] Hypothesis 4: When no red tide occurs, the sea area will follow the Lotka-Volterra model, that is, the algae reproduction situation × the number of algae and the feeding situation × the number of fish are balanced, and the fish mortality situation × the number of fish and the feeding reproduction situation × the number of algae are balanced. Therefore, the average chlorophyll a concentration predicted above should be read as an important basis for the number of algae; in the case of other red tides, the plant reproduction rate will grow at a rate of about twice the previous reproduction rate; in the case of harmful red tides and toxic red tides, the plant reproduction rate and mortality rate ratio will increase proportionally with the decrease in the feeding rate ratio.

[0152] The following will explain the red tide type, the overall simulation information of the fish school, and the simulation information of the sample fish.

[0153] ① Red tide type:

[0154] I have set up three types of red tides, namely other red tides, harmful red tides, and toxic red tides. The China Marine Disaster Bulletin (https: / / www.nmdis.org.cn / hygb / zghyzhgb / ) states that other red tides refer to red tides that do not produce toxins and have not yet caused damage to marine natural resources or the marine economy, but may have potential impacts on the marine ecosystem. Therefore, I have stipulated in my invention that during the process of other red tides, the impact on fish schools is almost zero, and phytoplankton will be provided as part of their food, which will not affect the excretion, death, poisoning, etc. of fish schools; harmful red tides refer to red tides that do not pose a direct threat to humans, but may cause damage to marine natural resources or the marine economy through physical, chemical and other means. Therefore, I have stipulated in my invention that during the process of harmful red tides, Although phytoplankton will be provided as part of the food, "poisoning" will occur as the amount of food consumed increases, which is divided into three conditions: mild, moderate, and poisoning. It will also reduce the feeding rate and excretion rate to a certain extent, increasing the risk of death. Toxic red tide refers specifically to the red tide that can cause poisoning or even death in humans. Among the three types of red tides, it is the red tide that has the greatest impact on human health and the most serious damage to fish schools. Therefore, I stipulate in the invention that during the process of toxic red tide, the feeding rate and excretion of fish schools will be lower than other red tides and harmful red tides, and the mortality rate will increase. The "poisoning" of the fish school is also divided into three conditions: mild, moderate, and poisoning. But it also represents the human attitude towards screening this type of fish school.

[0155] ② Overall simulation information of fish school:

[0156] To obtain the overall simulation information of the fish school, we first need to initialize the fish school and predators, using the previously built fish school simulation model as the main program, and then add some other initialization data, such as fish school energy, total food intake, and total excretion.

[0157] After determining the model assumptions of fish feeding and excretion, the red tide type is judged. According to the text description of the "Red Tide Type Drop-down Box", different red tide situations are simulated with different standards; in the fish simulation, the simulation of each fish needs to be traversed, so the energy demand, feeding demand, and excretion demand only need to be added to the main program of the fish simulation, and the feeding and excretion amounts can be gradually increased with iteration; here I set the initial energy value of each fish to be 40J~70J, the feeding threshold to be 65J, the fish swimming will consume 1~2J, each feeding will increase by 2~2.5J, and the fish escape will consume 3~4J. During the escape process, the fish cannot eat and excrete. The excretion is set to excrete once after feeding 5 times, and the excretion amount is 3%~5% of the total feeding amount. The feeding amount and excretion amount of each iteration are accumulated and output to their respective result boxes. The simulation results are as follows Figure 9 shown.

[0158] ①Sample fish simulation information:

[0159] To obtain the simulation information of sample fish, we actually take out the 1st to 5th value fish as sample fish before the fish school simulation iteration. The advantage of sample fish is that the food intake, excretion and status of each fish can be viewed from the perspective of a single fish. First, we need to determine the type of red tide. Different types of red tides have different standards for fish schools, and the final simulation results of the sample fish will also be different. Then, through the perception range of the sample fish, we can determine the current state. If the fish school is not within the predator's hunting range, it is in the "steady forward" state. If the fish school is within the predator's hunting range, it is in the "escape" state. If it always maintains the "escape" state, then the energy will eventually decrease to 0, which is the "death" state. Similarly, the sample fish will update the food intake and excretion at each iteration of the new speed and position, and accumulate and output them in their respective result boxes. Under other red tide types, the sample fish will be "not affected" by the red tide, but harmful red tides and toxic red tides will cause fish schools to be polluted by red tides. I stipulate that the impact of this type of pollution on fish schools depends on the food intake. When the food intake of the sample fish reaches 15mg, it is lightly polluted, 30mg for moderate pollution, and 45mg for severe pollution. For example Figure 10 These are the simulated conditions of sample fish of other red tides, harmful red tides, and toxic red tides respectively.

[0160] This completes the technical solution of this embodiment. Overall, the invention, primarily based on the MATLAB AppDesigner tool, simulates the behavior of fish schools during a red tide in the Zhoushan fishing grounds, where the fish follow basic swimming patterns. The method uses the spatial distribution of red tides from 1998 to 2023 as a training set to predict the 2024 red tide. Based on this red tide pattern and average chlorophyll density, a fish simulation platform is constructed to analyze the state of fish affected by the red tide, including three types: other red tides, harmful red tides, and toxic red tides.

[0161] Taking the situation on the third day after the red tide in spring 2024 as an example, the predicted chlorophyll concentration is 6.357 mg / m 3 The input for fish school simulation is that harmful red tides frequently occur in Zhejiang waters, so the red tide type is selected as harmful red tide and the fish school simulation is started. The simulation results are as follows: Figure 11 shown.

[0162] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide, characterized in that: include: Obtain red tide data to be predicted; Inputting the red tide data distribution to be predicted into a red tide prediction model to obtain the spatial distribution of the red tide, wherein the red tide prediction model is constructed by a back propagation neural network and obtained by training with a training set, wherein the training set includes data before and after the red tide occurs; Obtain the corresponding red tide type based on the spatial distribution of red tide and the average chlorophyll concentration; Constructing a fish school behavior model, performing fish school simulation based on the fish school behavior model and the corresponding red tide type, and outputting the fish school simulation situation; The fish school behavior model includes: a fish school swimming model and a fish school escape model; The fish school swimming model is used to update the fish school's position and speed based on the fish school's current position, speed and model parameters; The fish escape model is used to update the position and speed of the fish according to the current position and speed of the fish, the current position and speed of the predator and the model parameters; Based on the current position, speed and model parameters of the fish school, updating the fish school position and speed includes: Based on the current position and velocity of the school of fish, a global alignment velocity is calculated based on the average velocity of the school of fish; After calculating the global alignment speed, the approach speed, the gathering speed, and the repulsion speed of the fish school are initialized, the perception range of each fish's current position is traversed, the number of neighboring fish is found according to the perception distance, and the current fish is judged to be approaching or repelling, thereby obtaining a first judgment result; Calculate the gathering force after the combined force according to the first judgment result and the number of neighboring fish; Calculating the combined velocity of each fish in the school of fish based on the gathering force; Based on the combined velocity, updating the position and velocity of the school of fish; The method to calculate the total speed of each fish in the school is: Combined speed = global alignment weight × global alignment speed + approach weight × approach speed + convergence weight × convergence force + repulsion weight × repulsion speed; Among them, the ratio of the gathering speed of small fish schools within the sensing range to the number of neighboring fish is used as the gathering force.

2. The method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide according to claim 1, characterized in that: Obtaining the training set includes: Obtain data for a preset number of days before and after a red tide occurs; The data are integrated into a space×time dataset, and NaN values ​​in the integrated dataset are replaced to obtain the training set.

3. The method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide according to claim 1, characterized in that: Based on the current position and speed of the fish school, the current position and speed of the predator, and the model parameters, updating the position and speed of the fish school includes: According to the current position and speed of the fish school, the current position and speed of the predator and the model parameters, a vector directly escaping from the predator is calculated as the escape direction; judging the escape behavior according to the escape direction; Based on the escape behavior, determining the distance between a certain fish in the school of fish and the predator to obtain a second determination result; According to the second judgment result, the position and speed of the fish school are updated.

4. The method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide according to claim 3, characterized in that: According to the second judgment result, updating the fish school position and speed includes: If a fish in the school is within the predator's hunting radius, an escape factor is obtained based on the distance between the fish in the school and the predator, and based on the escape factor, the position and speed of the fish are determined; If the distance between a fish in the school and the predator is within a preset range, and the preset range is much smaller than the predation radius, random avoidance is performed, and a randomly generated angle is substituted into the rotation matrix to update the position and speed of the fish school.

5. The method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide according to claim 4, characterized in that: According to the escape factor, the method based on the position and speed of the fish school is: Fish school speed = fish school initial speed + escape factor × (escape direction - fish school initial speed); To perform random avoidance, randomly generate angles and substitute them into the rotation matrix. The method to update the position and speed of the fish school is as follows: Update the speed of the fish school = rotation matrix × initial speed of the fish school.

6. The method for simulating fish school conditions in Zhoushan fishing grounds under the influence of red tide according to claim 1, characterized in that: Before performing fish school simulation based on the fish school behavior model and the corresponding red tide type, the following steps are further included: Initialize the food intake, excretion, feeding rate, mortality rate of the entire fish population and the reproduction rate of plants, as well as the food intake and excretion of individual sample fish.

7. The fish school situation simulation system in Zhoushan fishing grounds under the influence of red tide is characterized by: include: Red tide prediction module and fish school visualization simulation module; The red tide prediction module is used to input the distribution of red tide data to be predicted into a red tide prediction model to obtain the spatial distribution of red tides, wherein the red tide prediction model is constructed by a back propagation neural network and obtained by training a training set, and the training set includes data before and after the red tide occurs; The fish school visualization simulation module is used to obtain the corresponding red tide type according to the red tide spatial distribution and the average chlorophyll concentration; construct a fish school behavior model, perform fish school simulation based on the fish school behavior model and the corresponding red tide type, and output the fish school simulation situation; The fish school behavior model includes: a fish school swimming model and a fish school escape model; The fish school swimming model is used to update the fish school's position and speed based on the fish school's current position, speed and model parameters; The fish escape model is used to update the position and speed of the fish according to the current position and speed of the fish, the current position and speed of the predator and the model parameters; Based on the current position, speed and model parameters of the fish school, updating the fish school position and speed includes: Based on the current position and velocity of the school of fish, a global alignment velocity is calculated based on the average velocity of the school of fish; After calculating the global alignment speed, the approach speed, the gathering speed, and the repulsion speed of the fish school are initialized, the perception range of each fish's current position is traversed, the number of neighboring fish is found according to the perception distance, and the current fish is judged to be approaching or repelling, thereby obtaining a first judgment result; Calculate the gathering force after the combined force according to the first judgment result and the number of neighboring fish; Calculating the combined velocity of each fish in the school of fish based on the gathering force; Based on the combined velocity, updating the position and velocity of the school of fish; The method to calculate the total speed of each fish in the school is: Combined speed = global alignment weight × global alignment speed + approach weight × approach speed + convergence weight × convergence force + repulsion weight × repulsion speed; Among them, the ratio of the gathering speed of small fish schools within the sensing range to the number of neighboring fish is used as the gathering force.

Images