Methods, apparatus and equipment for estimating fish stock based on machine learning models

Abstract

This invention discloses a method, apparatus, and equipment for estimating fish stock in ponds based on a machine learning model, relating to the field of fisheries counting technology. The method includes: acquiring pond data of the pond to be estimated; the pond data includes: meteorological data, basic fish data, fish feeding data, and water quality data; inputting the pond data of the pond to be estimated into a stock estimation model to obtain an estimated value of the fish stock in the pond; the stock estimation model is obtained by training a machine learning model using pond data from multiple training ponds and the corresponding actual values ​​of fish stock, and the machine learning model includes: a first sub-model, a second sub-model, a third sub-model, a fourth sub-model, and a fifth sub-model. This invention improves the accuracy of fish stock counting.

Description

Technical Field

[0001] This invention relates to the field of fishery counting technology, and in particular to a method, apparatus and equipment for estimating fish stock based on a machine learning model. Background Technology

[0002] In the aquaculture industry, accurately predicting fish stock is crucial for improving farming efficiency, optimizing feed supply, preventing overfishing, and enhancing economic benefits. However, in large-scale farming, accurately collecting data on fish population size and growth status is extremely difficult, and the impact of environmental factors such as water quality, climate change, and disease occurrence on fish populations is often overlooked in traditional methods. Traditional stock estimation methods, such as empirical judgment and simple calculations, are often not precise enough and cannot effectively address the complexity of environmental changes and fish behavior.

[0003] Therefore, the development of fish stock prediction technology urgently requires the introduction of more accurate methods to estimate fish growth rates and total populations. This includes real-time monitoring of the aquatic environment and fish population conditions, as well as the use of data analysis techniques for prediction. By comprehensively considering the impact of environmental factors on fish growth, the comprehensiveness and adaptability of prediction models can be improved, thereby effectively enhancing the overall efficiency of aquaculture and reducing negative environmental impacts. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus, and equipment for estimating fish stock based on a machine learning model, thereby improving the accuracy of fish stock counting.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for estimating fish stock based on a machine learning model, comprising:

[0007] Obtain fishpond data for the fishpond to be estimated; the fishpond data includes: meteorological data, basic fish data, fish feeding data, and water quality data;

[0008] The fishpond data of the fishpond to be estimated is input into the fish stock estimation model to obtain the estimated value of the fish stock in the fishpond to be estimated. The fish stock estimation model is obtained by training a machine learning model using fishpond data of multiple training fishponds and the corresponding actual values ​​of fish stock. The machine learning model includes: a first sub-model, a second sub-model, a third sub-model, a fourth sub-model, and a fifth sub-model.

[0009] Optionally, the meteorological data includes: daily air pressure, maximum temperature, minimum temperature, and weather; the basic fish data includes: fish age and average weight at sampling; the fish feeding data includes: feeding activity and amount of feed given; and the water quality data includes: dissolved oxygen, pH, and water temperature.

[0010] Optionally, the process of acquiring fish feeding data includes:

[0011] Acquire water surface environmental signals from the fish feeding area and non-fish feeding areas; the water surface environmental signals are water surface flow signals or water surface ripple signals.

[0012] The difference between the water surface environmental signal in the fish feeding area and the water surface environmental signal in the non-fish feeding area is determined as the feeding activity level of the fish.

[0013] The feeding activity of the fish population is used to control the feeding of the feeder and generate fish feeding data.

[0014] Optionally, the training process of the pond inventory estimation model includes:

[0015] Obtain fishpond data and corresponding actual fish stock values ​​from multiple training fishponds;

[0016] The machine learning model is constructed based on the first sub-model, the second sub-model, the third sub-model, the fourth sub-model, and the fifth sub-model.

[0017] Using the fishpond data of each training fishpond as input and the actual fish stock in the corresponding pond as output, the machine learning model is trained to obtain the stock estimation model.

[0018] Optionally, the machine learning model is trained using fishpond data from each training fishpond as input and the actual fish stock count as output to obtain the stock count estimation model, including:

[0019] Determine the initialization parameters of the machine learning model;

[0020] Using the fishpond data and corresponding actual fish stock values ​​of each training fishpond, starting with the initial parameters, the machine learning model is iteratively updated multiple times to obtain the fish stock estimation model; wherein, the update process of the machine learning model at the current iteration number includes:

[0021] For any training fishpond:

[0022] The meteorological data is input into the first sub-model of the current generation to obtain the first training feature vector of the current generation.

[0023] The basic fish data is input into the second sub-model of the current generation to obtain the second training feature vector of the current generation.

[0024] Fish feeding data is input into the third sub-model of the current generation to obtain the third training feature vector of the current generation;

[0025] The water quality data is input into the fourth sub-model of the current generation to obtain the fourth training feature vector of the current generation.

[0026] The first, second, third, and fourth training feature vectors of the current generation are input into the fifth sub-model of the current generation to obtain the estimated fish stock of the current generation.

[0027] Based on the estimated and actual fish stocks in all training ponds at the current generation, calculate the loss at the current generation.

[0028] Determine whether the stopping condition is met; the stopping condition is that the loss in the current generation is less than a preset value or the preset number of training iterations has been reached.

[0029] If so, the machine learning model at the current iteration number will be determined as the inventory estimation model.

[0030] If not, update the parameters of the machine learning model for the current iteration and return "Input the meteorological data into the first sub-model for the current iteration to obtain the first training feature vector for the current iteration".

[0031] Optionally, the first sub-model, the second sub-model, the third sub-model, the fourth sub-model, and the fifth sub-model are all one of neural networks, recurrent neural networks, convolutional networks, tree models, and support vector machines.

[0032] A device for estimating fish stock based on a machine learning model includes: an input unit, a prediction unit, and an output unit connected in sequence; the prediction unit is configured with a stock estimation model.

[0033] The input unit is used to receive fishpond data of the fishpond to be estimated; the fishpond data includes: meteorological data, basic fish data, fish feeding data, and water quality data;

[0034] The prediction unit is used to input the fishpond data of the fishpond to be estimated into the stock estimation model to obtain the estimated value of the fish stock of the fishpond to be estimated; the stock estimation model is obtained by training a machine learning model using fishpond data of multiple training fishponds and the corresponding actual values ​​of fish stock; the machine learning model includes: a first sub-model, a second sub-model, a third sub-model, a fourth sub-model and a fifth sub-model.

[0035] The output unit is used to output an estimated value of the fish stock in the pond to be estimated.

[0036] A fish stock estimation device based on a machine learning model includes: one or more sensors, one or more microcontrollers, and a cloud server, wherein the cloud server is used to execute the steps of the fish stock estimation method based on the machine learning model described above.

[0037] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the fish stock estimation method based on a machine learning model as described above.

[0038] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the fish stock estimation method based on a machine learning model as described above.

[0039] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0040] This invention discloses a method, apparatus, and device for estimating fish stock in ponds based on a machine learning model. First, fishpond data is acquired, including meteorological data, basic fish data, fish feeding data, and water quality data. Then, the fishpond data is input into the stock estimation model to obtain an estimated fish stock. The stock estimation model is trained using fishpond data from multiple training ponds and corresponding actual fish stock values. The machine learning model includes a first sub-model, a second sub-model, a third sub-model, a fourth sub-model, and a fifth sub-model. This invention utilizes a machine learning model to estimate fish stock, improving the accuracy of fish stock counting. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of the fish stock estimation method based on a machine learning model provided in Embodiment 1 of the present invention.

[0043] Figure 2 This is a schematic diagram of a machine learning model architecture;

[0044] Figure 3 This is a schematic diagram of a neural network structure;

[0045] Figure 4 A schematic diagram illustrating the process of estimating fish stock using neural networks;

[0046] Figure 5 A schematic diagram of a fish stock estimation device based on a machine learning model;

[0047] Figure 6 A schematic diagram of a fish stock estimation device based on a machine learning model;

[0048] Figure 7 This is a diagram of the internal structure of a computer device. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] The purpose of this invention is to provide a method, apparatus, and equipment for estimating fish stock based on a machine learning model, aiming to improve the accuracy of fish stock counting.

[0051] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] Example 1

[0053] like Figure 1 As shown, the fish stock estimation method based on a machine learning model in this embodiment includes:

[0054] Step 101: Obtain fishpond data for the fishpond to be estimated; fishpond data includes: meteorological data, basic fish data, fish feeding data, and water quality data.

[0055] As an optional implementation method, meteorological data includes: daily air pressure, maximum temperature, minimum temperature and weather; basic fish data includes: fish age and average weight at sampling; fish feeding data includes: feeding activity and amount of feed given; water quality data includes: dissolved oxygen, pH and water temperature.

[0056] As an optional implementation method, the process of acquiring fish feeding data includes:

[0057] Step 1011: Obtain the water surface environment signal of the fish feeding area and the water surface environment signal of the non-fish feeding area; the water surface environment signal is the water surface flow signal or the water surface ripple signal.

[0058] Step 1012: Determine the difference between the water surface environmental signal in the fish feeding area and the water surface environmental signal in the non-fish feeding area as the feeding activity level of the fish.

[0059] Step 1013: Use the feeding activity of the fish school to control the feeding machine and generate fish feeding data.

[0060] Step 102: Input the fishpond data of the fishpond to be estimated into the stock estimation model to obtain the estimated value of the fish stock in the fishpond to be estimated.

[0061] The fish stock estimation model is trained using data from multiple training fishponds and the corresponding actual fish stock values. The machine learning model includes five sub-models: a first sub-model, a second sub-model, a third sub-model, a fourth sub-model, and a fifth sub-model. The structure of the machine learning model is as follows: Figure 2 As shown. Where X1-X N Given N inputs, Y1-Y N For the corresponding N outputs.

[0062] As an optional implementation method, the training process of the stock estimation model includes:

[0063] Step 1021: Obtain fishpond data and the corresponding actual fish population of multiple training fishponds.

[0064] Step 1022: Construct a machine learning model based on the first sub-model, the second sub-model, the third sub-model, the fourth sub-model, and the fifth sub-model.

[0065] Step 1023: Using the fishpond data of each training fishpond as input and the actual value of the corresponding fish stock as output, train the machine learning model to obtain the stock estimation model.

[0066] As an optional implementation, step 1023 includes:

[0067] Determine the initialization parameters of the machine learning model.

[0068] Using fishpond data and corresponding actual fish stocks from each training pond, starting with initialized parameters, the machine learning model is iteratively updated multiple times to obtain a fish stock estimation model. The update process of the machine learning model at the current iteration number includes:

[0069] For any training fishpond:

[0070] The meteorological data is input into the first sub-model of the current generation to obtain the first training feature vector of the current generation.

[0071] The basic fish data is input into the second sub-model of the current generation to obtain the second training feature vector of the current generation.

[0072] Fish feeding data is input into the third sub-model of the current generation to obtain the third training feature vector of the current generation.

[0073] The water quality data is input into the fourth sub-model of the current generation to obtain the fourth training feature vector of the current generation.

[0074] The first, second, third, and fourth training feature vectors of the current generation are input into the fifth sub-model of the current generation to obtain the estimated fish stock of the current generation.

[0075] Based on the estimated and actual fish stocks in all training ponds at the current generation, calculate the loss at the current generation.

[0076] Determine if the stopping condition is met; the stopping condition is that the loss in the current generation is less than the preset value or the preset number of training iterations has been reached.

[0077] If so, the machine learning model for the current iteration number will be determined as the inventory estimation model.

[0078] If not, update the parameters of the machine learning model for the current iteration and return "Input the meteorological data into the first sub-model for the current iteration to obtain the first training feature vector for the current iteration".

[0079] As an optional implementation, the first sub-model, the second sub-model, the third sub-model, the fourth sub-model, and the fifth sub-model are all one of the following: neural network, recurrent neural network, convolutional network, tree model, and support vector machine.

[0080] Specifically, we will take the first, second, third, fourth, and fifth sub-models, all of which are artificial neural networks, as examples for illustration.

[0081] like Figure 3As shown, an artificial neural network consists of layers of neurons, divided into input layers, hidden layers, and output layers. Input signals or features are first received by the input layer and then passed layer by layer to the hidden layers. During this process, each neuron processes the signal. Finally, these signals are passed to the output layer. Between different layers, the application of activation functions enhances the network's non-linear expressive power. This entire process from input to output is called forward propagation. The artificial neural network receives data through the input layer, where each input feature corresponds to a neuron. The data is then passed to one or more hidden layers, which are the core of the network structure and responsible for weighting and non-linearly transforming the input data. The number of neurons and the activation function in each hidden layer can be adjusted according to the specific problem. The output layer of the artificial neural network outputs the final calculation result according to the nature of the task. Each neuron in the artificial neural network performs the following calculations:

[0082]

[0083] Where o is the output of the neuron; f(·) is the activation function used to introduce nonlinear factors; w i For the i-th input x i The corresponding weights; b is the bias term; n is the total number of inputs.

[0084] like Figure 4 As shown, the first sub-model is a first neural network, which receives meteorological data and outputs a first feature vector; the second sub-model is a second neural network, which receives basic fish data and outputs a second feature vector; the third sub-model is a third neural network, which receives fish feeding data and outputs a third feature vector; the fourth sub-model is a fourth neural network, which receives water quality data and outputs a fourth feature vector; and the fifth sub-model is a fifth neural network, which receives the first, second, third, and fourth feature vectors and outputs an estimated value of the fish population in the pond.

[0085] To evaluate and optimize model performance, this invention employs a loss function to measure the difference between the network output (estimated fish stock) and the actual fish stock. In the regression problem, the mean squared error (MSE) is used as the loss function, calculated as follows:

[0086]

[0087] Among them, y j Let J be the true value of the fish stock in the j-th training pond. is the estimated fish population in the j-th training pond; m is the total number of training ponds.

[0088] To minimize the loss function, this example uses gradient descent to adjust the network parameters (weights and biases). The parameter update formula is as follows:

[0089]

[0090] Where θ is the parameter (weight or bias); η is the learning rate; This represents the gradient of the loss function with respect to the parameters.

[0091] Example 2

[0092] like Figure 5 As shown, the fish stock estimation device based on a machine learning model in this embodiment includes: an input unit 210, a prediction unit 220, and an output unit 230 connected in sequence; the prediction unit 230 is internally configured with a stock estimation model.

[0093] The input unit 210 is used to receive fishpond data of the fishpond to be estimated; the fishpond data includes: meteorological data, basic fish data, fish feeding data and water quality data.

[0094] The prediction unit 220 is used to input the fishpond data of the fishpond to be estimated into the stock estimation model to obtain the estimated value of the fish stock in the fishpond to be estimated. The stock estimation model is obtained by training a machine learning model using fishpond data of multiple training fishponds and the corresponding actual values ​​of fish stock. The machine learning model includes: a first sub-model, a second sub-model, a third sub-model, a fourth sub-model, and a fifth sub-model.

[0095] Output unit 230 is used to output the estimated value of the fish stock in the fishpond to be estimated.

[0096] Example 3

[0097] like Figure 6 As shown, the fish stock estimation device based on the machine learning model in this embodiment includes: one or more sensors 310, one or more microcontrollers 320, and a cloud server 330. The cloud server 330 is used to execute the steps of the fish stock estimation method based on the machine learning model in embodiment 1.

[0098] Specifically, in terms of data collection, the fish stock estimation device based on machine learning models is equipped with various sensors 310, including temperature sensors, dissolved oxygen sensors, and feeding desire sensors. These sensors monitor water quality parameters (such as water temperature and dissolved oxygen content) and fish behavior in real time. The sensors 310 continuously send data to a central processing system to ensure the timeliness and accuracy of the information. Meanwhile, to guarantee high data quality, the calibration and maintenance of the sensors follow strict standards.

[0099] In the data processing stage, the fish stock estimation device based on machine learning models is equipped with a cloud server 330 and various machine learning algorithms. These algorithms extract key information from the large amount of raw data collected and transform it into feature vectors, providing a scientific basis for decision-making. During this process, the data undergoes preprocessing, including cleaning, noise reduction, and standardization, to improve the accuracy and reliability of the analysis results.

[0100] The cloud server 330 serves as the data processing and control unit. This unit is responsible for receiving and processing feeding activity signals and, based on these signals, sending start / stop control signals to the microcontroller 320. This unit also receives data from water quality sensors and weighing data. It includes a high-performance processor and necessary software support to generate pond stock predictions based on machine learning algorithms.

[0101] The microcontroller is responsible for receiving electrical signals from various sensors and communicating with the cloud server 330 via network. This unit is also responsible for receiving control signals from the cloud server 330 and the feed weight signal from the feed feeder 340. It is also responsible for controlling the switching on and off of the feed feeder 340.

[0102] The feeder 340 is responsible for feeding and weighing the feed. After feeding is completed, it sends the feed weight signal back to the microcontroller 320.

[0103] Specifically, meteorological data is obtained through a meteorological API. After the fish farmer completes sampling, the fish age and average weight of the samples are entered into the estimation device. The device conducts trial feedings at feeding times, controlling the start and stop of the trial feedings based on the water surface fluctuation data of the fish population. Feeding is stopped when the water surface fluctuations become calm. The amount of feed given is obtained by weighing the fish and the water surface fluctuation data. Water quality data during feeding is acquired and uploaded by temperature sensors, dissolved oxygen sensors, and pH sensors in the device.

[0104] Example 4

[0105] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the fish stock estimation method based on a machine learning model in Embodiment 1.

[0106] Example 5

[0107] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the fish stock estimation method based on a machine learning model in Embodiment 1.

[0108] Example 6

[0109] A computer program product includes a computer program that, when executed by a processor, implements the steps of the fish stock estimation method based on a machine learning model in Example 1.

[0110] Example 7

[0111] A computer device, which may be a database, may have an internal structure diagram as shown below. Figure 7 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores pending transactions. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements the fish stock estimation method based on a machine learning model in Example 1.

[0112] Advantages of this invention:

[0113] 1. Improved estimation accuracy: Traditional methods rely on empirical estimation or simple calculations, which are easily affected by subjective judgment. This invention, by using a machine learning model, can effectively integrate and analyze various key data, including meteorological data, basic fish data, fish feeding data, and water quality data. The core advantage of this method lies in its comprehensive data utilization capability, which can provide accurate and reliable fish stock estimates under complex natural environmental conditions, thereby greatly improving the estimation accuracy.

[0114] 2. Real-time data processing and analysis: Compared to traditional methods that may rely on delayed or historical data, this invention can collect and analyze data in real time, ensuring the timeliness of information. This is crucial for rapidly responding to environmental changes and adjusting aquaculture strategies in a timely manner.

[0115] 3. Automated and intelligent operation: Traditional methods often require manual intervention. This invention automates data collection and processing, reduces manual intervention, and improves the level of intelligent operation.

[0116] 4. Adaptability and Flexibility: Machine learning models can adapt to different aquaculture environments and fish species, exhibiting strong universal applicability and flexibility. Compared to traditional methods, this allows for wider application in aquaculture scenarios of different scales and types.

[0117] 5. Environmentally friendly: More accurate estimation of fish stock can reduce feed waste and over-farming, thereby reducing negative impacts on the aquatic environment and promoting sustainable aquaculture.

[0118] It should be noted that the object information (including but not limited to object device information, object personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this invention are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0119] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided by this invention may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided by this invention may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0120] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0121] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for estimating fish stock in ponds based on a machine learning model, characterized in that, The method includes: Obtain fishpond data for the fishpond to be estimated; the fishpond data includes: meteorological data, basic fish data, fish feeding data, and water quality data; The fishpond data of the fishpond to be estimated is input into the fish stock estimation model to obtain the estimated value of the fish stock in the fishpond to be estimated; the fish stock estimation model is obtained by training a machine learning model using fishpond data of multiple training fishponds and the corresponding actual values ​​of fish stock; the machine learning model includes: a first sub-model, a second sub-model, a third sub-model, a fourth sub-model and a fifth sub-model. Meteorological data includes: daily air pressure, maximum temperature, minimum temperature, and weather conditions; the basic fish data includes: fish age and average weight at sampling; the fish feeding data includes: feeding activity level and amount of feed given; the water quality data includes: dissolved oxygen, pH, and water temperature. The process of obtaining fish feeding data includes: Acquire water surface environmental signals from the fish feeding area and non-fish feeding areas; the water surface environmental signals are water surface flow signals or water surface ripple signals. The difference between the water surface environmental signal in the fish feeding area and the water surface environmental signal in the non-fish feeding area is determined as the feeding activity level of the fish. The feeding activity of the fish school is used to control the feeding machine and generate fish feeding data. The training process of the pond inventory estimation model includes: Obtain fishpond data and corresponding actual fish stock values ​​from multiple training fishponds; The machine learning model is constructed based on the first sub-model, the second sub-model, the third sub-model, the fourth sub-model, and the fifth sub-model. Using the fishpond data of each training fishpond as input and the actual value of the corresponding fish stock as output, the machine learning model is trained to obtain the stock estimation model. Using fishpond data from each training fishpond as input and the actual fish stock count as output, the machine learning model is trained to obtain the fish stock estimation model, including: Determine the initialization parameters of the machine learning model; Using the fishpond data and corresponding actual fish stock values ​​of each training fishpond, starting with the initial parameters, the machine learning model is iteratively updated multiple times to obtain the fish stock estimation model; wherein, the update process of the machine learning model at the current iteration number includes: For any training fishpond: The meteorological data is input into the first sub-model of the current generation to obtain the first training feature vector of the current generation. The basic fish data is input into the second sub-model of the current generation to obtain the second training feature vector of the current generation. Fish feeding data is input into the third sub-model of the current generation to obtain the third training feature vector of the current generation; The water quality data is input into the fourth sub-model of the current generation to obtain the fourth training feature vector of the current generation. The first, second, third, and fourth training feature vectors of the current generation are input into the fifth sub-model of the current generation to obtain the estimated fish stock of the current generation. Based on the estimated and actual fish stocks in all training ponds at the current generation, calculate the loss at the current generation. Determine whether the stopping condition is met; the stopping condition is that the loss in the current generation is less than a preset value or the preset number of training iterations has been reached. If so, the machine learning model at the current iteration number will be determined as the inventory estimation model. If not, update the parameters of the machine learning model at the current iteration number and return "Input the meteorological data into the first sub-model at the current iteration number to obtain the first training feature vector at the current iteration number".

2. The method for estimating fish stock based on a machine learning model according to claim 1, characterized in that, The first sub-model, the second sub-model, the third sub-model, the fourth sub-model, and the fifth sub-model are all one of the following: neural network, recurrent neural network, convolutional network, tree model, and support vector machine.

3. A fish stock estimation device based on a machine learning model, used to implement the fish stock estimation method based on a machine learning model as described in claims 1-2, characterized in that, The device includes: an input unit, a prediction unit, and an output unit connected in sequence; the prediction unit is internally configured with a stock estimation model. The input unit is used to receive fishpond data of the fishpond to be estimated; the fishpond data includes: meteorological data, basic fish data, fish feeding data, and water quality data; The prediction unit is used to input the fishpond data of the fishpond to be estimated into the stock estimation model to obtain the estimated value of the fish stock of the fishpond to be estimated; the stock estimation model is obtained by training a machine learning model using fishpond data of multiple training fishponds and the corresponding actual values ​​of fish stock; the machine learning model includes: a first sub-model, a second sub-model, a third sub-model, a fourth sub-model and a fifth sub-model. The output unit is used to output an estimated value of the fish stock in the pond to be estimated.

4. A fish stock estimation device based on a machine learning model, characterized in that, The device includes: one or more sensors, one or more microcontrollers, and a cloud server, wherein the cloud server is used to perform the steps of implementing the fish stock estimation method based on the machine learning model according to any one of claims 1-2.

5. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the fish stock estimation method based on a machine learning model as described in any one of claims 1-2.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the fish stock estimation method based on a machine learning model as described in any one of claims 1-2.

Images