A method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data

Abstract

The present invention relates to the field of fishery data analysis and prediction, and discloses a method for predicting the economic benefits of fish catches based on fishery acoustic data and multi-source data, comprising the following steps: collecting relevant data of fisheries, obtaining a class of fish catches at the same time, performing spatiotemporal alignment of the relevant data of fisheries to obtain spatiotemporal aligned fishery data, and constructing an updated fish catch economic benefit prediction model and an updated fish catch species quantity prediction model based on the spatiotemporal aligned fishery data, which are used to generate a fishing plan for a class of fish catches. The present invention can achieve the purpose of increasing economic benefits, improving fishermen's income, and reducing overfishing of sensitive species through accurate prediction and optimization of fishing vessel operation plans.

Description

Technical Field

[0001] The present invention relates to the field of fishery data analysis and prediction, in particular to a method for predicting economic benefits of fish catches based on fishery acoustic data and multi-source data. Background Art

[0002] Fishery resources are limited, and the growing population has led to an increasing demand for fishery resources. By predicting the economic benefits of fish catches, we can understand the economic value of different fishery resources, thereby guiding the rational allocation and utilization of fishery resources. This helps to avoid overfishing and waste and ensure the sustainable use of fishery resources. Fishery investment involves equipment purchase, farm construction, labor input and other aspects, which require a lot of money and resources. By predicting the economic benefits of fish catches, we can evaluate the potential returns and risks of different investment projects and provide investors with a scientific basis for decision-making. This helps to reduce investment risks and improve investment benefits. By predicting the economic benefits of fish catches, we can also understand the innovation points of fishery technology and the direction of industrial upgrading. For example, the prediction results may show that certain efficient and environmentally friendly fishing technologies and breeding models have higher economic benefits, thereby promoting the innovation of fishery technology and industrial upgrading, and improving the overall competitiveness of fishery production.

[0003] In the current fishery production, the randomness of fishing operations is high, and it is difficult to accurately predict the types, quantity and economic benefits of the catch, resulting in low fishing efficiency and serious waste of resources. At the same time, the dynamic changes in the environment and the market have a significant impact on the economic value of the catch, and an efficient prediction technology is urgently needed to guide scientific fishing and resource management. Therefore, a method for predicting the economic benefits of catch based on fishery acoustic data and multi-source data is proposed. Summary of the invention

[0004] The invention overcomes the shortcomings of the prior art and provides a method for predicting the economic benefits of fish catches based on fishery acoustic data and multi-source data.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] The first aspect of the present invention provides a method for predicting the economic benefits of fish catch based on fishery acoustic data and multi-source data, comprising the following steps:

[0007] S102: Determine the fishery detection area and collect fishery related data, including fishery acoustic data, fishery climate data and fishery market price information, and mark all types of fish catches that may exist in the fishery detection area as a type of fish catch;

[0008] S104: preprocessing the collected fishery acoustic data, and integrating the fishery acoustic data, fishery climate data, and market price information of different types of fish catches through a spatiotemporal alignment algorithm to obtain spatiotemporal aligned fishery data;

[0009] S106: Combined with the spatiotemporal aligned fishery data, a catch species quantity prediction model and a catch economic benefit prediction model are constructed, and the attention mechanism is introduced to optimize the sensitivity of key factors of the model to obtain an updated catch economic benefit prediction model and an updated catch species quantity prediction model;

[0010] S108: Generate a first fishing plan and a second fishing plan by updating the fish catch economic benefit prediction model and the fish catch species quantity prediction model.

[0011] Furthermore, in a preferred embodiment of the present invention, the step S102 is specifically:

[0012] Determining a fishery monitoring area, wherein the fishery monitoring area includes fish catch and seawater;

[0013] Placing a monitoring fishing vessel in the fishery detection area, installing a multi-beam echo sounder and a split-beam echo sounder on the side and bottom of the monitoring fishing vessel, and controlling the multi-beam echo sounder and the split-beam echo sounder to connect with a data recording system on the fishing vessel;

[0014] Wherein, the data recording system can process the data collected by the multi-beam echo sounder and the split-beam echo sounder;

[0015] Introducing a big data network, retrieving climate data of the fishery detection area based on the big data network, marking it as fishery climate data, and retrieving all types of fish catches that may exist in the fishery detection area through the big data network;

[0016] Determine in the big data network all detection parameters of the multi-beam echo sounder and the split-beam echo sounder when acting on all types of fish that may exist in the fishery detection area, calibrate them as target detection parameters, apply the target detection parameters to the multi-beam echo sounder and the split-beam echo sounder, and control the operation of the multi-beam echo sounder and the split-beam echo sounder to monitor the target intensity parameters and echo integral parameters in the fishery monitoring area;

[0017] The target intensity parameters and echo integral parameters are collectively referred to as fishery acoustic data, and all types of fish that may exist in the fishery detection area are calibrated as a type of fish. At the same time, the market price information of different types of type one fish is determined based on the big data network.

[0018] Furthermore, in a preferred embodiment of the present invention, the step S104 is specifically:

[0019] Preliminary screening is performed on the fishery acoustic data, wherein the preliminary screening step is to propose data whose target intensity parameters and echo integral parameters in the fishery acoustic data are greater than a preset range, and retain data that is not greater than the preset range, to obtain the fishery acoustic data after preliminary screening, and mark it as preliminary screened fishery acoustic data;

[0020] For the preliminary screening fishery acoustic data, an adaptive filter is introduced, and the preliminary screening fishery acoustic data is imported into the adaptive filter for noise filtering and weak current erasing, so as to obtain the preliminary screening fishery acoustic data after adaptive filtering, which is calibrated as filtered fishery acoustic data;

[0021] Extracting time-related features of the filtered fishery acoustic data, including the data duration and data amplitude of the filtered fishery acoustic data, performing Fourier transform on the filtered fishery acoustic data, and extracting frequency-domain-related features in the Fourier-transformed filtered fishery acoustic data, including the spectrum distribution state and the main frequency;

[0022] Through the spatiotemporal alignment algorithm, the time-related characteristics and frequency-domain related characteristics of filtered fishery acoustic data are combined to achieve the spatiotemporal alignment of filtered fishery acoustic data and related data.

[0023] Further, in a preferred embodiment of the present invention, the time-space alignment of the filtered fishery acoustic data and related data is achieved by combining the time-related characteristics and frequency-domain related characteristics of the filtered fishery acoustic data through the time-space alignment algorithm, specifically:

[0024] Obtaining the acquisition timeline of the fishery climate data, and constructing the timestamp of the fishery climate data according to the acquisition timeline of the fishery climate data, and constructing the timestamp of the market price information of the first type of fish catch of different types, and obtaining the timestamp of the market price of the first type of fish catch;

[0025] Based on the time-related characteristics and frequency-domain-related characteristics of the filtered fishery acoustic data, a timestamp of the filtered fishery acoustic data is constructed, wherein the timestamps of the filtered fishery acoustic data corresponding to the fishery acoustic data collected at different locations in the fishery monitoring area are different;

[0026] Time-align all timestamps of filtered fishery acoustic data, timestamps of fishery climate data and timestamps of market prices of a type of fish catch to obtain time-aligned fishery data;

[0027] Obtaining filtered fishery acoustic data, fishery climate data, and market price information of different types of first-class catches for collection geographic coordinate analysis, and introducing a spatial mapping algorithm based on the collection geographic coordinates of filtered fishery acoustic data, fishery climate data, and market price information of different types of first-class catches, mapping different data in the same spatial framework for spatial alignment, and obtaining spatially aligned fishery data;

[0028] The time-aligned fishery data and space-aligned fishery data are integrated into a unified dataset and standardized and transformed to obtain time-space aligned fishery data.

[0029] Furthermore, in a preferred embodiment of the present invention, the step S106 is specifically as follows:

[0030] A random forest algorithm is introduced, and based on the random forest algorithm, an initial decision tree model is preset, and fishery acoustic data and fishery climate data after spatiotemporal alignment are extracted from the spatiotemporal aligned fishery data, and are calibrated as spatiotemporal aligned fishery sub-data;

[0031] The spatiotemporally aligned fishery sub-data are imported into the initial decision tree model, and a root node is determined in the initial decision tree model. The root node is used as the origin, and the spatiotemporally aligned fishery sub-data are recursively divided into subsets through the initial decision tree model, wherein a division threshold and a maximum number of divisions are preset, and the subset recursive division is to split the spatiotemporally aligned fishery sub-data based on the division threshold until the number of divisions is equal to the maximum number of divisions;

[0032] The initial decision tree model after the subset recursive partitioning is defined as a preliminary partitioning decision tree model, in which the splitting points of the spatiotemporally aligned fishery sub-data are marked as leaf nodes, and based on the leaf nodes, the preliminary partitioning decision tree model is post-pruned to obtain a target decision tree model, and all target decision tree models are integrated in combination with a random forest algorithm to obtain a model that can predict the number of a type of catch, which is marked as a catch number prediction model;

[0033] In the catch species and quantity prediction model, the spatiotemporal aligned fishery data is input, wherein the spatiotemporal aligned fishery data includes market price information of a class of catches of different species after spatiotemporal alignment;

[0034] In the fish species quantity prediction model, an iterative profit calculation algorithm is preset to traverse and calculate the total profit after combining the quantity of different types of fish species and the corresponding market price information, and a profit map of the quantity of different types of fish species is constructed based on the total profit. Based on the profit map of the quantity of different types of fish species, the model parameters of the fish species quantity prediction model are updated to obtain a fish economic benefit prediction model;

[0035] The attention mechanism is introduced into the fishery economic benefit prediction model and the fishery species quantity prediction model to optimize the sensitivity of the key factors of the model, and to obtain an updated fishery economic benefit prediction model and an updated fishery species quantity prediction model.

[0036] Furthermore, in a preferred embodiment of the present invention, the attention mechanism is optimized in the fish catch economic benefit prediction model and the fish catch species quantity prediction model to achieve the optimization of the sensitivity of the key factors of the model, and to obtain an updated fish catch economic benefit prediction model and an updated fish catch species quantity prediction model, specifically:

[0037] Obtain the attention mechanism in the fish catch economic benefit prediction model and the fish catch species quantity prediction model, and mark them as the attention mechanism to be analyzed;

[0038] In the attention mechanism to be analyzed, the attention monitoring parameter is adjusted to a sparse attention monitoring parameter, and the prediction weight table of the current prediction of the catch in the catch economic benefit prediction model and the catch species quantity prediction model is obtained, and calibrated as the current prediction weight table;

[0039] Running a fish catch economic benefit prediction model and a fish catch species quantity prediction model to obtain a fish catch species quantity ranking table and a fish catch economic benefit ranking table, and based on the fish catch species quantity ranking table and the fish catch economic benefit ranking table, updating the weights of the current prediction weight table in a reverse order so that the weight distribution ratio is equal to the ratio of the number of fish catch species and the economic benefit in the fish catch species quantity ranking table and the fish catch economic benefit ranking table, and obtaining a target prediction weight table;

[0040] Based on the target prediction weight table, the weight of the attention mechanism to be analyzed whose attention monitoring parameter is equal to the sparse attention monitoring parameter is updated to achieve sensitivity adjustment of the catch economic benefit prediction model and the catch species quantity prediction model, and obtain an updated catch economic benefit prediction model and an updated catch species quantity prediction model.

[0041] Furthermore, in a preferred embodiment of the present invention, the step S108 is specifically:

[0042] Running the updated fish catch economic benefit prediction model and the updated fish catch species quantity prediction model to obtain an updated fish catch species quantity ranking table and a fish catch economic benefit ranking table, which are marked as the updated fish catch species quantity ranking table and the updated fish catch economic benefit ranking table;

[0043] Determine the total number of Class I fish catches, and analyze them in combination with the updated fish catch economic benefit ranking table to generate an initial Class I fish catch fishing plan, wherein the initial Class I fish catch fishing plan is to generate the initial catch quantity of different types of Class I fish catches within the total number of Class I fish catches and in the fishery monitoring area based on the economic benefit ranking of the fish catches obtained in the fish catch economic benefit ranking table, and control the sum of the initial catch quantity of all types of Class I fish catches to be equal to the total number of Class I fish catches;

[0044] Analyze the updated list of the number of species of fish to determine the total number of the first category of fish of different species in the fishery monitoring area, and analyze it in combination with the initial catch number of the first category of fish of different species;

[0045] If the initial catch quantity of a type of fish of all species is not greater than the corresponding total quantity in the fishery monitoring area, the initial catch fishing plan shall be marked as the first fishing plan;

[0046] If the initial catch quantity of a type of fish is greater than the corresponding total quantity in the fishery monitoring area, the corresponding type of fish will be marked as type 2 fish, and a second fishing plan will be generated;

[0047] Among them, the second fishing plan is based on the first fishing plan, controlling the catch quantity of the second type of fish to be equal to the corresponding total number in the fishery monitoring area.

[0048] The second aspect of the present invention further provides a fish catch economic benefit prediction system based on fishery acoustic data and multi-source data, the fish catch economic benefit prediction system comprises a memory and a processor, the memory stores a fish catch economic benefit prediction method, and when the fish catch economic benefit prediction method is executed by the processor, the following steps are implemented:

[0049] Determine the fishery detection area and collect fishery related data, including fishery acoustic data, fishery climate data and fishery market price information, and mark all types of fish catches that may exist in the fishery detection area as a type of fish catch;

[0050] The collected fishery acoustic data are preprocessed, and the fishery acoustic data, fishery climate data and market price information of different types of fish are integrated through the spatiotemporal alignment algorithm to obtain spatiotemporal aligned fishery data;

[0051] Combined with the spatiotemporal aligned fishery data, a catch species quantity prediction model and a catch economic benefit prediction model were constructed, and the attention mechanism was introduced to optimize the sensitivity of key factors of the model, thus obtaining an updated catch economic benefit prediction model and an updated catch species quantity prediction model.

[0052] By updating the economic benefit prediction model of the catch and updating the catch species and quantity prediction model, the first fishing plan and the second fishing plan are generated.

[0053] The present invention solves the technical defects existing in the background technology, and has the following beneficial effects: collecting relevant data of fisheries, obtaining a type of fish catch at the same time, performing spatiotemporal alignment of the relevant data of fisheries, obtaining spatiotemporal aligned fishery data, and constructing an updated fish catch economic benefit prediction model and an updated fish catch species and quantity prediction model based on the spatiotemporal aligned fishery data, which are used to generate a fishing plan for a type of fish catch. The present invention can achieve the purpose of increasing economic benefits, improving fishermen's income, and reducing overfishing of sensitive species through accurate prediction and optimization of fishing vessel operation plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, drawings of other embodiments can be obtained based on these drawings without paying creative work.

[0055] Figure 1 A flow chart of a method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data is shown;

[0056] Figure 2 A flow chart of a method for constructing and updating a fish catch economic benefit prediction model and a fish catch species quantity prediction model is shown;

[0057] Figure 3 The program view of the catch economic benefit prediction system based on fishery acoustic data and multi-source data is shown. DETAILED DESCRIPTION

[0058] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0059] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0060] Figure 1 A flow chart of a method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data is shown, comprising the following steps:

[0061] S102: Determine the fishery detection area and collect fishery related data, including fishery acoustic data, fishery climate data and fishery market price information, and mark all types of fish catches that may exist in the fishery detection area as a type of fish catch;

[0062] S104: preprocessing the collected fishery acoustic data, and integrating the fishery acoustic data, fishery climate data, and market price information of different types of fish catches through a spatiotemporal alignment algorithm to obtain spatiotemporal aligned fishery data;

[0063] S106: Combined with the spatiotemporal aligned fishery data, a catch species quantity prediction model and a catch economic benefit prediction model are constructed, and the attention mechanism is introduced to optimize the sensitivity of key factors of the model to obtain an updated catch economic benefit prediction model and an updated catch species quantity prediction model;

[0064] S108: Generate a first fishing plan and a second fishing plan by updating the fish catch economic benefit prediction model and the fish catch species quantity prediction model.

[0065] Furthermore, in a preferred embodiment of the present invention, the step S102 is specifically:

[0066] Determining a fishery monitoring area, wherein the fishery monitoring area includes fish catch and seawater;

[0067] Placing a monitoring fishing vessel in the fishery detection area, installing a multi-beam echo sounder and a split-beam echo sounder on the side and bottom of the monitoring fishing vessel, and controlling the multi-beam echo sounder and the split-beam echo sounder to connect with a data recording system on the fishing vessel;

[0068] Wherein, the data recording system can process the data collected by the multi-beam echo sounder and the split-beam echo sounder;

[0069] Introducing a big data network, retrieving climate data of the fishery detection area based on the big data network, marking it as fishery climate data, and retrieving all types of fish catches that may exist in the fishery detection area through the big data network;

[0070] Determine in the big data network all detection parameters of the multi-beam echo sounder and the split-beam echo sounder when acting on all types of fish that may exist in the fishery detection area, calibrate them as target detection parameters, apply the target detection parameters to the multi-beam echo sounder and the split-beam echo sounder, and control the operation of the multi-beam echo sounder and the split-beam echo sounder to monitor the target intensity parameters and echo integral parameters in the fishery monitoring area;

[0071] The target intensity parameters and echo integral parameters are collectively referred to as fishery acoustic data, and all types of fish that may exist in the fishery detection area are calibrated as a type of fish. At the same time, the market price information of different types of type one fish is determined based on the big data network.

[0072] It should be noted that in order to predict the economic benefits of fish catches, it is first necessary to determine the types of fish catches, market prices and other information in the detection area, and determine the number of different types of fish catches. The types of fish catches can be determined by searching the big data network, while the analysis of their quantity requires a combined analysis of acoustic data, climate data, etc. Acoustic data refers to target intensity parameters and echo integral parameters, which describe the distribution of fish underwater. Climate data refers to seawater temperature, salinity and dissolved oxygen. Different climate data have different effects on different types of fish catches. The distribution of fish catches can be roughly judged based on climate data. In order to determine the target intensity parameters and echo integral parameters, multi-beam echo sounders and split-beam echo sounders need to be installed to obtain them. The purpose of obtaining market price information of different types of fish catches is that since the market price information of different fish catches is inconsistent, the benefits brought are also inconsistent. In order to predict the economic benefits of fish catches, it is necessary to combine the market price information of different types of fish catches.

[0073] Furthermore, in a preferred embodiment of the present invention, the step S104 is specifically:

[0074] Preliminary screening is performed on the fishery acoustic data, wherein the preliminary screening step is to propose data whose target intensity parameters and echo integral parameters in the fishery acoustic data are greater than a preset range, and retain data that is not greater than the preset range, to obtain the fishery acoustic data after preliminary screening, and mark it as preliminary screened fishery acoustic data;

[0075] For the preliminary screening fishery acoustic data, an adaptive filter is introduced, and the preliminary screening fishery acoustic data is imported into the adaptive filter for noise filtering and weak current erasing, so as to obtain the preliminary screening fishery acoustic data after adaptive filtering, which is calibrated as filtered fishery acoustic data;

[0076] Extracting time-related features of the filtered fishery acoustic data, including the data duration and data amplitude of the filtered fishery acoustic data, performing Fourier transform on the filtered fishery acoustic data, and extracting frequency-domain-related features in the Fourier-transformed filtered fishery acoustic data, including the spectrum distribution state and the main frequency;

[0077] Through the spatiotemporal alignment algorithm, the time-related characteristics and frequency-domain related characteristics of filtered fishery acoustic data are combined to achieve the spatiotemporal alignment of filtered fishery acoustic data and related data.

[0078] It should be noted that after obtaining the relevant data of fisheries, these data need to be preprocessed, because there may be noise and outliers in the data when it is collected, and filtering and outliers need to be eliminated to achieve the purpose of improving data accuracy. First, analyze the acoustic data of fisheries, first perform adaptive filtering of the data, perform noise filtering and weak current erasure, and then extract the relevant features of time and space from the filtered acoustic data of fisheries, because time and space alignment is required. The purpose of time alignment is to ensure that data from different time points or different collection cycles can be compared and analyzed in the same time frame, and the purpose of space alignment is to ensure that data from different locations or different spatial distributions can be compared and analyzed in the same spatial frame. The combination of the two is time and space alignment, realizing data fusion and obtaining accurate and comprehensive fishery data. Time and space provide data support for the relevant features of time and space.

[0079] Further, in a preferred embodiment of the present invention, the time-space alignment of the filtered fishery acoustic data and related data is achieved by combining the time-related characteristics and frequency-domain related characteristics of the filtered fishery acoustic data through the time-space alignment algorithm, specifically:

[0080] Obtaining the acquisition timeline of the fishery climate data, and constructing the timestamp of the fishery climate data according to the acquisition timeline of the fishery climate data, and constructing the timestamp of the market price information of the first type of fish catch of different types, and obtaining the timestamp of the market price of the first type of fish catch;

[0081] Based on the time-related characteristics and frequency-domain-related characteristics of the filtered fishery acoustic data, a timestamp of the filtered fishery acoustic data is constructed, wherein the timestamps of the filtered fishery acoustic data corresponding to the fishery acoustic data collected at different locations in the fishery monitoring area are different;

[0082] Time-align all timestamps of filtered fishery acoustic data, timestamps of fishery climate data and timestamps of market prices of a type of fish catch to obtain time-aligned fishery data;

[0083] Obtaining filtered fishery acoustic data, fishery climate data, and market price information of different types of first-class catches for collection geographic coordinate analysis, and introducing a spatial mapping algorithm based on the collection geographic coordinates of filtered fishery acoustic data, fishery climate data, and market price information of different types of first-class catches, mapping different data in the same spatial framework for spatial alignment, and obtaining spatially aligned fishery data;

[0084] The time-aligned fishery data and space-aligned fishery data are integrated into a unified dataset and standardized and transformed to obtain time-space aligned fishery data.

[0085] It should be noted that timestamp is the key data for spatiotemporal alignment of data. Time synchronization of data can be achieved based on the timestamps between different data, and the timestamps of different data sources can be adjusted to the same reference point. Spatial registration is the process of eliminating the differences in spatial distribution of data obtained by different sensors or acquisition devices. Based on time registration and spatial registration, spatiotemporal aligned fishery data can be obtained.

[0086] Furthermore, in a preferred embodiment of the present invention, the step S108 is specifically:

[0087] Running the updated fish catch economic benefit prediction model and the updated fish catch species quantity prediction model to obtain an updated fish catch species quantity ranking table and a fish catch economic benefit ranking table, which are marked as the updated fish catch species quantity ranking table and the updated fish catch economic benefit ranking table;

[0088] Determine the total number of Class I fish catches, and analyze them in combination with the updated fish catch economic benefit ranking table to generate an initial Class I fish catch fishing plan, wherein the initial Class I fish catch fishing plan is to generate the initial catch quantity of different types of Class I fish catches within the total number of Class I fish catches and in the fishery monitoring area based on the economic benefit ranking of the fish catches obtained in the fish catch economic benefit ranking table, and control the sum of the initial catch quantity of all types of Class I fish catches to be equal to the total number of Class I fish catches;

[0089] Analyze the updated list of the number of species of fish to determine the total number of the first category of fish of different species in the fishery monitoring area, and analyze it in combination with the initial catch number of the first category of fish of different species;

[0090] If the initial catch quantity of a type of fish of all species is not greater than the corresponding total quantity in the fishery monitoring area, the initial catch fishing plan shall be marked as the first fishing plan;

[0091] If the initial catch quantity of a type of fish is greater than the corresponding total quantity in the fishery monitoring area, the corresponding type of fish will be marked as type 2 fish, and a second fishing plan will be generated;

[0092] Among them, the second fishing plan is based on the first fishing plan, controlling the catch quantity of the second type of fish to be equal to the corresponding total number in the fishery monitoring area.

[0093] It should be noted that after obtaining the updated fish catch economic benefit prediction model and the updated fish catch species quantity prediction model, the updated fish catch species quantity ranking table and the updated fish catch economic benefit ranking table can be obtained by running the model. The updated fish catch species quantity ranking table and the updated fish catch economic benefit ranking table can predict the quantity of different types of catches in the monitoring area and the corresponding economic benefits. First, the total number of catches needs to be determined, which is determined according to the order. The number of fish caught must not exceed the total number, so now according to the economic benefits of the catch, the catch quantity of different types of catches is allocated, and it is determined that the total number of catches of all types of catches is not greater than the total number, so as to obtain a preliminary fishing plan. Secondly, the catch quantity of different types of fish is analyzed to determine whether there is a target catch quantity greater than the maximum catch quantity. The maximum catch quantity can be obtained by updating the catch type quantity sorting table. If not, the preliminary fishing plan is directly output. If so, the fishing plan needs to be updated. On the basis of the first fishing plan, the catch quantity of the second type of fish is controlled to be equal to the corresponding total number in the fishery monitoring area, that is, in the second fishing plan, the maximum catch of the second type of fish can only be the corresponding total number in the fishery monitoring area, and the catch quantity of other fish is equal to the first fishing plan.

[0094] Figure 2 A flow chart of a method for constructing and updating a fish catch economic benefit prediction model and a fish catch species quantity prediction model is shown, comprising the following steps:

[0095] S202: Combine the spatiotemporal aligned fishery data to construct a catch species quantity prediction model and a catch economic benefit prediction model, and introduce an attention mechanism to optimize the sensitivity of key factors of the model to obtain an updated catch economic benefit prediction model and an updated catch species quantity prediction model;

[0096] S204: Optimizing the attention mechanism in the fishery economic benefit prediction model and the fishery species quantity prediction model to achieve sensitivity optimization of key factors of the model, and obtaining an updated fishery economic benefit prediction model and an updated fishery species quantity prediction model.

[0097] Furthermore, in a preferred embodiment of the present invention, the step S202 is specifically:

[0098] A random forest algorithm is introduced, and based on the random forest algorithm, an initial decision tree model is preset, and fishery acoustic data and fishery climate data after spatiotemporal alignment are extracted from the spatiotemporal aligned fishery data, and are calibrated as spatiotemporal aligned fishery sub-data;

[0099] The spatiotemporally aligned fishery sub-data are imported into the initial decision tree model, and a root node is determined in the initial decision tree model. The root node is used as the origin, and the spatiotemporally aligned fishery sub-data are recursively divided into subsets through the initial decision tree model, wherein a division threshold and a maximum number of divisions are preset, and the subset recursive division is to split the spatiotemporally aligned fishery sub-data based on the division threshold until the number of divisions is equal to the maximum number of divisions;

[0100] The initial decision tree model after the subset recursive partitioning is defined as a preliminary partitioning decision tree model, in which the splitting points of the spatiotemporally aligned fishery sub-data are marked as leaf nodes, and based on the leaf nodes, the preliminary partitioning decision tree model is post-pruned to obtain a target decision tree model, and all target decision tree models are integrated in combination with a random forest algorithm to obtain a model that can predict the number of a type of catch, which is marked as a catch number prediction model;

[0101] In the catch species and quantity prediction model, the spatiotemporal aligned fishery data is input, wherein the spatiotemporal aligned fishery data includes market price information of a class of catches of different species after spatiotemporal alignment;

[0102] In the fish species quantity prediction model, an iterative profit calculation algorithm is preset to traverse and calculate the total profit after combining the quantity of different types of fish species and the corresponding market price information, and a profit map of the quantity of different types of fish species is constructed based on the total profit. Based on the profit map of the quantity of different types of fish species, the model parameters of the fish species quantity prediction model are updated to obtain a fish economic benefit prediction model;

[0103] The attention mechanism is introduced into the fishery economic benefit prediction model and the fishery species quantity prediction model to optimize the sensitivity of the key factors of the model, and to obtain an updated fishery economic benefit prediction model and an updated fishery species quantity prediction model.

[0104] It should be noted that the random forest algorithm evolved from the decision tree algorithm, and the random forest model is formed by multiple trained decision tree models. The random forest model can realize data prediction and is used to build a model for the number of fish species and an economic benefit prediction model. To build a model for the number of fish species prediction, it is necessary to automatically generate a decision tree for the fishery acoustic data and fishery climate data that are aligned in time and space, that is, starting from the root node, recursively build a decision tree according to the result of feature selection. At each node, the best split feature is selected, and the data set is divided into several subsets according to the different values ​​of the feature, that is, according to the partition threshold, the subset is split until the termination condition is met. After the subset split is completed, in order to avoid overfitting, the decision tree needs to be pruned, the purpose is to simplify the decision tree model and improve its generalization ability. Post-pruning processing is adopted, that is, after the decision tree is fully generated, some nodes are deleted to simplify the model, and multiple decision tree models are combined to generate a model for the number of fish species prediction. At the same time, it is necessary to build a fish catch economic benefit prediction model. By inputting the market price information of a type of fish catch of different types after time and space alignment in the fish catch type and quantity prediction model, the market economic benefits of the market price information of different types of fish catch corresponding to the quantity can be determined. An iterative algorithm is designed to traverse the predicted types and quantities of each type of fish catch, calculate the total income or total profit according to the corresponding market price, and finally obtain the fish catch economic benefit prediction model.

[0105] Furthermore, in a preferred embodiment of the present invention, the step 204 is specifically:

[0106] Obtain the attention mechanism in the fish catch economic benefit prediction model and the fish catch species quantity prediction model, and mark them as the attention mechanism to be analyzed;

[0107] In the attention mechanism to be analyzed, the attention monitoring parameter is adjusted to a sparse attention monitoring parameter, and the prediction weight table of the current prediction of the catch in the catch economic benefit prediction model and the catch species quantity prediction model is obtained, and calibrated as the current prediction weight table;

[0108] Running a fish catch economic benefit prediction model and a fish catch species quantity prediction model to obtain a fish catch species quantity ranking table and a fish catch economic benefit ranking table, and based on the fish catch species quantity ranking table and the fish catch economic benefit ranking table, updating the weights of the current prediction weight table in a reverse order so that the weight distribution ratio is equal to the ratio of the number of fish catch species and the economic benefit in the fish catch species quantity ranking table and the fish catch economic benefit ranking table, and obtaining a target prediction weight table;

[0109] Based on the target prediction weight table, the weight of the attention mechanism to be analyzed whose attention monitoring parameter is equal to the sparse attention monitoring parameter is updated to achieve sensitivity adjustment of the catch economic benefit prediction model and the catch species quantity prediction model, and obtain an updated catch economic benefit prediction model and an updated catch species quantity prediction model.

[0110] It should be noted that the purpose of improving the sensitivity of the attention mechanism in the catch economic benefit prediction model and the catch species quantity prediction model is to improve the prediction accuracy and stability of the model and provide a more scientific and reliable basis for fishery management and resource allocation. First, the current attention mechanism is obtained, and the monitoring parameters are replaced in the current attention mechanism, and sparse attention monitoring parameters are used to reduce the complexity of the calculation. Then, the weight distribution of the attention mechanism needs to be adjusted so that the model pays more attention to the factors that have a greater impact on the prediction results. The weight distribution is distributed according to the number of catch species and the market price information of the catch. The larger the number of species, the higher the economic benefits brought by the market, and the catch is ranked in front, and the weight ratio is greater. According to the above-mentioned weight distribution ratio, that is, the ratio of the number of catch species and the economic benefits of a class of catch species in the catch economic benefit ranking table and the catch economic benefit ranking table, the target prediction weight table can be obtained. In the attention mechanism, the weight is updated in combination with the target prediction weight table to achieve sensitivity adjustment, that is, the attention focus of the catch is adjusted, and the updated catch economic benefit prediction model and the updated catch species quantity prediction model are obtained.

[0111] like Figure 3 As shown, the second aspect of the present invention further provides a catch economic benefit prediction system based on fishery acoustic data and multi-source data, the catch economic benefit prediction system comprises a memory 31 and a processor 32, the memory 31 stores a catch economic benefit prediction method, and when the catch economic benefit prediction method is executed by the processor 32, the following steps are implemented:

[0112] Determine the fishery detection area and collect fishery related data, including fishery acoustic data, fishery climate data and fishery market price information, and mark all types of fish catches that may exist in the fishery detection area as a type of fish catch;

[0113] The collected fishery acoustic data are preprocessed, and the fishery acoustic data, fishery climate data and market price information of different types of fish are integrated through the spatiotemporal alignment algorithm to obtain spatiotemporal aligned fishery data;

[0114] Combined with the spatiotemporal aligned fishery data, a catch species quantity prediction model and a catch economic benefit prediction model were constructed, and the attention mechanism was introduced to optimize the sensitivity of key factors of the model, thus obtaining an updated catch economic benefit prediction model and an updated catch species quantity prediction model.

[0115] By updating the economic benefit prediction model of the catch and updating the catch species and quantity prediction model, the first fishing plan and the second fishing plan are generated.

[0116] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data, characterized in that: The following steps are involved: S102: Determine the fishery detection area and collect fishery related data, including fishery acoustic data, fishery climate data and fishery market price information, and mark all types of fish catches that may exist in the fishery detection area as a type of fish catch; S104: preprocessing the collected fishery acoustic data, and integrating the fishery acoustic data, fishery climate data, and market price information of different types of fish catches through a spatiotemporal alignment algorithm to obtain spatiotemporal aligned fishery data; S106: Combined with the spatiotemporal aligned fishery data, a catch species quantity prediction model and a catch economic benefit prediction model are constructed, and the attention mechanism is introduced to optimize the sensitivity of key factors of the model to obtain an updated catch economic benefit prediction model and an updated catch species quantity prediction model; S108: Generate a first fishing plan and a second fishing plan by updating the fish catch economic benefit prediction model and the fish catch species quantity prediction model; Wherein, the S106 is specifically: A random forest algorithm is introduced, and based on the random forest algorithm, an initial decision tree model is preset, and fishery acoustic data and fishery climate data after spatiotemporal alignment are extracted from the spatiotemporal aligned fishery data, and are calibrated as spatiotemporal aligned fishery sub-data; The spatiotemporally aligned fishery sub-data are imported into the initial decision tree model, and a root node is determined in the initial decision tree model. The root node is used as the origin, and the spatiotemporally aligned fishery sub-data are recursively divided into subsets through the initial decision tree model, wherein a division threshold and a maximum number of divisions are preset, and the subset recursive division is to split the spatiotemporally aligned fishery sub-data based on the division threshold until the number of divisions is equal to the maximum number of divisions; The initial decision tree model after the subset recursive partitioning is defined as a preliminary partitioning decision tree model, in which the splitting points of the spatiotemporally aligned fishery sub-data are marked as leaf nodes, and based on the leaf nodes, the preliminary partitioning decision tree model is post-pruned to obtain a target decision tree model, and all target decision tree models are integrated in combination with a random forest algorithm to obtain a model that can predict the number of a type of catch, which is marked as a catch number prediction model; In the catch species and quantity prediction model, the spatiotemporal aligned fishery data is input, wherein the spatiotemporal aligned fishery data includes market price information of a class of catches of different species after spatiotemporal alignment; In the fish species quantity prediction model, an iterative profit calculation algorithm is preset to traverse and calculate the total profit after combining the quantity of different types of fish species and the corresponding market price information, and a profit map of the quantity of different types of fish species is constructed based on the total profit. Based on the profit map of the quantity of different types of fish species, the model parameters of the fish species quantity prediction model are updated to obtain a fish economic benefit prediction model; Introducing the attention mechanism into the fish catch economic benefit prediction model and the fish catch species quantity prediction model to optimize the sensitivity of the key factors of the model, and obtaining an updated fish catch economic benefit prediction model and an updated fish catch species quantity prediction model; The method of optimizing the attention mechanism in the fish catch economic benefit prediction model and the fish catch species quantity prediction model to achieve the optimization of the sensitivity of the key factors of the model, and obtaining an updated fish catch economic benefit prediction model and an updated fish catch species quantity prediction model, is specifically as follows: Obtain the attention mechanism in the fish catch economic benefit prediction model and the fish catch species quantity prediction model, and mark them as the attention mechanism to be analyzed; In the attention mechanism to be analyzed, the attention monitoring parameter is adjusted to a sparse attention monitoring parameter, and the prediction weight table of the current prediction of the catch in the catch economic benefit prediction model and the catch species quantity prediction model is obtained, and calibrated as the current prediction weight table; Running a fish catch economic benefit prediction model and a fish catch species quantity prediction model to obtain a fish catch species quantity ranking table and a fish catch economic benefit ranking table, and based on the fish catch species quantity ranking table and the fish catch economic benefit ranking table, updating the weights of the current prediction weight table in a reverse order so that the weight distribution ratio is equal to the ratio of the number of fish catch species and the economic benefit in the fish catch species quantity ranking table and the fish catch economic benefit ranking table, and obtaining a target prediction weight table; Based on the target prediction weight table, the weight of the attention mechanism to be analyzed whose attention monitoring parameter is equal to the sparse attention monitoring parameter is updated to achieve sensitivity adjustment of the catch economic benefit prediction model and the catch species quantity prediction model, and obtain an updated catch economic benefit prediction model and an updated catch species quantity prediction model.

2. The method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data according to claim 1, characterized in that: The S102 is specifically: Determining a fishery monitoring area, wherein the fishery monitoring area includes fish catch and seawater; Placing a monitoring fishing vessel in the fishery detection area, installing a multi-beam echo sounder and a split-beam echo sounder on the side and bottom of the monitoring fishing vessel, and controlling the multi-beam echo sounder and the split-beam echo sounder to connect with a data recording system on the fishing vessel; Wherein, the data recording system can process the data collected by the multi-beam echo sounder and the split-beam echo sounder; Introducing a big data network, retrieving climate data of the fishery detection area based on the big data network, marking it as fishery climate data, and retrieving all types of fish catches that may exist in the fishery detection area through the big data network; Determine in the big data network all detection parameters of the multi-beam echo sounder and the split-beam echo sounder when acting on all types of fish that may exist in the fishery detection area, calibrate them as target detection parameters, apply the target detection parameters to the multi-beam echo sounder and the split-beam echo sounder, and control the operation of the multi-beam echo sounder and the split-beam echo sounder to monitor the target intensity parameters and echo integral parameters in the fishery monitoring area; The target intensity parameters and echo integral parameters are collectively referred to as fishery acoustic data, and all types of fish that may exist in the fishery detection area are calibrated as a type of fish. At the same time, the market price information of different types of type one fish is determined based on the big data network.

3. The method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data according to claim 1, characterized in that: The S104 is specifically: Preliminary screening is performed on the fishery acoustic data, wherein the preliminary screening step is to propose data whose target intensity parameters and echo integral parameters in the fishery acoustic data are greater than a preset range, and retain data that is not greater than the preset range, to obtain the fishery acoustic data after preliminary screening, and mark it as preliminary screened fishery acoustic data; For the preliminary screening fishery acoustic data, an adaptive filter is introduced, and the preliminary screening fishery acoustic data is imported into the adaptive filter for noise filtering and weak current erasing, so as to obtain the preliminary screening fishery acoustic data after adaptive filtering, which is calibrated as filtered fishery acoustic data; Extracting time-related features of the filtered fishery acoustic data, including the data duration and data amplitude of the filtered fishery acoustic data, performing Fourier transform on the filtered fishery acoustic data, and extracting frequency-domain-related features in the Fourier-transformed filtered fishery acoustic data, including the spectrum distribution state and the main frequency; Through the spatiotemporal alignment algorithm, the time-related characteristics and frequency-domain related characteristics of filtered fishery acoustic data are combined to achieve the spatiotemporal alignment of filtered fishery acoustic data and related data.

4. The method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data according to claim 3 is characterized in that: The time-space alignment algorithm is used to combine the time-related characteristics and frequency-domain related characteristics of the filtered fishery acoustic data to achieve the time-space alignment of the filtered fishery acoustic data and related data, specifically: Obtaining the acquisition timeline of the fishery climate data, and constructing the timestamp of the fishery climate data according to the acquisition timeline of the fishery climate data, and constructing the timestamp of the market price information of the first type of fish catch of different types, so as to obtain the timestamp of the market price of the first type of fish catch; Based on the time-related characteristics and frequency-domain-related characteristics of the filtered fishery acoustic data, a timestamp of the filtered fishery acoustic data is constructed, wherein the timestamps of the filtered fishery acoustic data corresponding to the fishery acoustic data collected at different locations in the fishery monitoring area are different; Time-align all timestamps of filtered fishery acoustic data, timestamps of fishery climate data and timestamps of market prices of a type of fish catch to obtain time-aligned fishery data; Obtaining filtered fishery acoustic data, fishery climate data, and market price information of different types of first-class catches for collection geographic coordinate analysis, and introducing a spatial mapping algorithm based on the collection geographic coordinates of filtered fishery acoustic data, fishery climate data, and market price information of different types of first-class catches, mapping different data in the same spatial framework for spatial alignment, and obtaining spatially aligned fishery data; The time-aligned fishery data and space-aligned fishery data are integrated into a unified dataset and standardized and transformed to obtain time-space aligned fishery data.

5. The method for predicting economic benefits of fish catch based on fishery acoustic data and multi-source data according to claim 1, characterized in that: The S108 is specifically: Running the updated fish catch economic benefit prediction model and the updated fish catch species quantity prediction model to obtain an updated fish catch species quantity ranking table and a fish catch economic benefit ranking table, which are marked as the updated fish catch species quantity ranking table and the updated fish catch economic benefit ranking table; Determine the total number of Class I fish catches, and analyze them in combination with the updated fish catch economic benefit ranking table to generate an initial Class I fish catch fishing plan, wherein the initial Class I fish catch fishing plan is to generate the initial catch quantity of different types of Class I fish catches within the total number of Class I fish catches and in the fishery monitoring area based on the economic benefit ranking of the fish catches obtained in the fish catch economic benefit ranking table, and control the sum of the initial catch quantity of all types of Class I fish catches to be equal to the total number of Class I fish catches; Analyze the updated list of the number of species of fish to determine the total number of the first category of fish of different species in the fishery monitoring area, and analyze it in combination with the initial catch number of the first category of fish of different species; If the initial catch quantity of a type of fish of all species is not greater than the corresponding total quantity in the fishery monitoring area, the initial catch fishing plan shall be marked as the first fishing plan; If the initial catch quantity of a type of fish is greater than the corresponding total quantity in the fishery monitoring area, the corresponding type of fish will be marked as type 2 fish, and a second fishing plan will be generated; Among them, the second fishing plan is based on the first fishing plan, controlling the catch quantity of the second type of fish to be equal to the corresponding total number in the fishery monitoring area.

6. A fish catch economic benefit prediction system based on fishery acoustic data and multi-source data, characterized in that: The fishery economic benefit prediction system includes a memory and a processor, wherein the fishery economic benefit prediction method program is stored in the memory, and when the fishery economic benefit prediction method program is executed by the processor, the fishery economic benefit prediction method steps as described in any one of claims 1-5 are implemented.

Images