Method and system for controlling and optimizing breeding environment of lutjanus erythropterus fries

Abstract

The invention discloses a control optimization method and system for a lutjanus erythropterus fry cultivation environment. The method comprises the following steps: dividing a target breeding pool of lutjanus erythropterus fry into a plurality of three-dimensional sub-regions, and obtaining water quality change information of each sub-region in a preset time period and activity frequency change information of the lutjanus erythropterus fry in real time; based on the data, repulsion behaviors of the lutjanus erythropterus offspring seeds on the sub-regions under different water quality change conditions are analyzed, and a repulsion behavior-water quality characteristic database is established; determining a suitable cultivation environment condition of the lutjanus erythropterus fries according to the database; and finally, carrying out environment regulation and control on the cultivation pool according to the determined suitable environment conditions to form a scientific cultivation environment control optimization strategy. The method can dynamically sense the sensitive reaction of the fry to the water quality change, effectively improves the adaptability of the breeding environment and the fine management level, promotes the healthy growth of the fry, and improves the survival rate of the fry and the breeding benefits.

Description

Technical Field

[0001] The present invention relates to the technical field of aquaculture, and particularly relates to a method and system for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry. Background Art

[0002] Lutjanus erythropterus is an important marine economic fish, which has the advantages of fast growth rate, strong stress resistance, excellent meat quality, etc., and is deeply favored by the aquaculture industry and the market. With the increasing depletion of marine fishing resources, artificial aquaculture has become the main direction of the development of the Lutjanus erythropterus industry, and the healthy cultivation of fry is the key link to achieve efficient aquaculture.

[0003] In the prior art, the cultivation of Lutjanus erythropterus fry mostly relies on traditional artificial experience and water quality control means with a single parameter. For example, the survival rate of fry is improved by adjusting water temperature, dissolved oxygen or feeding frequency and other measures. However, the water quality change in the actual aquaculture environment is often a complex process of multi-factor superposition. The single-parameter control method is difficult to comprehensively reflect the true response of fry to the environment, resulting in insufficient timeliness and accuracy of water quality regulation, and thus affecting the healthy growth of fry.

[0004] In addition, there is currently a lack of systematic analysis means for the behavioral response of fry in the three-dimensional water body space. In particular, the correlation research between the active avoidance or aggregation behavior of fry and the local water quality change is relatively weak. Existing research mostly focuses on the observation of two-dimensional plane distribution characteristics, ignoring the activity differences of fry between different water layers, and it is difficult to effectively establish a dynamic adaptation relationship model between behavior and environment. At the same time, traditional water quality regulation is mostly static or manually set, lacking a data-driven intelligent optimization mechanism, and unable to meet the personalized needs of fry for water quality conditions at different times and different growth stages.

[0005] Therefore, there is an urgent need to provide a method and system for controlling the cultivation environment of fry that can comprehensively consider the behavioral changes of Lutjanus erythropterus fry, the multi-parameter linkage characteristics of water quality, and achieve dynamic regulation and intelligent optimization, so as to improve the scientificity, accuracy and automation level of fry cultivation, and promote the healthy growth of fry and the continuous improvement of aquaculture benefits. Summary of the Invention

[0006] In order to solve at least one of the above technical problems, the present invention proposes a method and system for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry.

[0007] The first aspect of the present invention provides a method for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry, including:

[0008] Dividing the target cultivation pool of Lutjanus erythropterus fry into N three-dimensional sub-regions, and obtaining the water quality change information and the activity frequency change information of Lutjanus erythropterus fry in each three-dimensional sub-region within a preset time period;

[0009] Determine the rejection behavior of Lutjanus erythopterus fry in the three-dimensional sub-regions with different water quality changes within a preset time period according to the water quality change information and activity frequency change information, and obtain a rejection behavior-water quality characteristic database;

[0010] Determine the suitable cultivation environmental conditions for Lutjanus erythopterus fry according to the rejection behavior-water quality characteristic database;

[0011] Optimize the cultivation environment of the target cultivation pond according to the suitable cultivation environmental conditions to obtain an optimized cultivation environment control strategy.

[0012] In this solution, dividing the target cultivation pond of Lutjanus erythopterus fry into N three-dimensional sub-regions, and obtaining the water quality change information and the activity frequency change information of Lutjanus erythopterus fry in each three-dimensional sub-region within a preset time period, specifically:

[0013] Divide the target cultivation pond of Lutjanus erythopterus fry into N three-dimensional sub-regions, and obtain the water quality change information in each three-dimensional sub-region within a preset time period based on water quality sensors, including water temperature, salinity, pH value, dissolved oxygen concentration, ammonia nitrogen content, and water particulate matter concentration change data;

[0014] Obtain the fry activity video data in each three-dimensional sub-region within a preset time period based on camera equipment, extract the video frame images of the fry activity video data, identify and count the number of Lutjanus erythopterus fry in each video frame image, and determine the activity frequency change information of Lutjanus erythopterus fry in each three-dimensional sub-region within a preset time period.

[0015] In this solution, determining the rejection behavior of Lutjanus erythopterus fry in the three-dimensional sub-regions with different water quality changes within a preset time period according to the water quality change information and activity frequency change information, and obtaining a rejection behavior-water quality characteristic database, specifically:

[0016] Extract the activity frequency information of each three-dimensional sub-region at each time node from the activity frequency change information according to a preset time step, perform normalization processing on the activity frequency information, and construct a three-dimensional space coordinate matrix for each time node, where each coordinate point contains the position parameters of the three-dimensional sub-region and the corresponding activity frequency value;

[0017] Construct a spatial clustering model for the three-dimensional sub-regions based on the BIRCH clustering algorithm, perform clustering operations on the three-dimensional space coordinate matrix according to the spatial clustering model, and generate an initial clustering sub-cluster of the fry activity frequency distribution in the three-dimensional space by iteratively constructing a clustering feature tree structure including a branching factor and a spatial radius threshold.

[0018] Perform density clustering on the initial clustering sub-clusters, merge adjacent sub-clusters within the spatial radius threshold range to form a set of micro-clusters representing the activity heat in different regional ranges, calculate the three-dimensional space volume covered by each micro-cluster and the average value of the fry activity frequency it contains, and generate a spatial activity heat distribution map of Lutjanus erythropterus fry in the target cultivation pond at each time node;

[0019] Perform time serialization processing on the spatial activity heat distribution map at each time node, and calculate the temporal decay characteristic values of the activity heat in each three-dimensional sub-region through the dynamic time warping algorithm;

[0020] If the temporal decay characteristic value exceeds the preset threshold, it is determined that Lutjanus erythropterus has repulsive behavior within the preset time period, and the three-dimensional sub-region of Lutjanus erythropterus repulsion is marked as a repulsive sub-region;

[0021] Analyze the water quality change information of the repulsive sub-region based on the random forest algorithm, determine the repulsion correlation of Lutjanus erythropterus to different water quality changes, and construct a repulsion behavior-water quality characteristic database.

[0022] In this solution, the analysis of the water quality change information of the repulsive sub-region based on the random forest algorithm, determining the repulsion correlation of Lutjanus erythropterus to different water quality changes, and constructing a repulsion behavior-water quality characteristic database are specifically as follows:

[0023] Calculate the difference between the temporal decay characteristic value of each repulsive sub-region and the preset threshold, and determine the repulsion intensity of Lutjanus erythropterus to the water quality change information corresponding to each repulsive sub-region according to the difference;

[0024] Take the water quality change information of the repulsive sub-region and the corresponding repulsion intensity as the positive sample data set, and the water quality change information of the non-repulsive sub-region as the negative sample data set;

[0025] Construct a random forest classification model, take the positive sample data set and the negative sample data set as the input feature variables of the model, take the repulsion behavior label as the first output variable of the model and the repulsion intensity as the second output variable of the model, use the Gini coefficient as the node splitting criterion, and adopt the method of sampling with replacement to extract sample subsets from the training set, randomly select sample subsets to split at each node, and obtain a preset number of decision trees;

[0026] Train the random forest classification model based on the preset number of decision trees, use the cross-validation method to adjust the model hyperparameters, calculate the feature importance scores of each water quality parameter in the model, and screen out the water quality parameter combinations whose correlation with the repulsion behavior exceeds the preset threshold according to the feature importance scores;

[0027] Construct an exclusion behavior - water quality characteristic database with water quality parameters as the characteristic dimension and the intensity of exclusion behavior as the label according to the selected water quality parameter combinations and their corresponding exclusion behavior labels. The database stores the mapping relationship between different water quality parameter combinations and exclusion behaviors and the intensity of exclusion behaviors.

[0028] In this solution, determining the suitable cultivation environmental conditions for Lutjanus erythopterus fries according to the exclusion behavior - water quality characteristic database is specifically as follows:

[0029] Extract the water quality parameter combinations with the intensity of exclusion behavior lower than the preset intensity threshold from the exclusion behavior - water quality characteristic database as the candidate suitable parameter set;

[0030] Perform dimensionality reduction processing on the water quality parameters of the candidate suitable parameter set through the principal component analysis method. Calculate the covariance matrix of each water quality parameter, decompose the covariance matrix to obtain the eigenvectors and variance contribution rates, and screen out the principal component vectors representing water quality characteristics according to the variance contribution rates;

[0031] Reconstruct the principal component projection space of water quality parameters based on the principal component vectors. Sort according to the absolute value of the load coefficient of each water quality parameter in the principal component projection space, and screen out the dominant water quality parameter combinations with the load coefficient exceeding the preset load threshold;

[0032] Construct an environmental fitness evaluation function for Lutjanus erythopterus fries based on the dominant water quality parameter combinations. Input the environmental fitness evaluation function into the genetic algorithm, and perform global optimization and solution through population initialization, fitness calculation, selection, crossover, and mutation operations to obtain the optimal solution set of each dominant water quality parameter;

[0033] Determine the constraint boundary conditions of each water quality parameter in the dominant water quality parameter combination according to the optimal solution set, and combine the statistical distribution characteristics of non - dominant water quality parameters in the candidate suitable parameter set to calculate the fluctuation range of non - dominant water quality parameters at different quantiles, generate the water quality suitable parameter interval for the target cultivation pond of Lutjanus erythopterus fries, and obtain the suitable cultivation environmental conditions.

[0034] In this solution, optimizing the cultivation environment of the target cultivation pond according to the suitable cultivation environmental conditions to obtain the optimized cultivation environment control strategy is specifically as follows:

[0035] Optimize the cultivation environment of the target cultivation pond of Lutjanus erythopterus fries according to the suitable cultivation environmental conditions, and monitor the growth rate of the activity frequency of Lutjanus erythopterus fries in the excluded sub - area after environmental optimization in real time;

[0036] Evaluate the optimization effect of the cultivation environment according to the growth rate of the activity frequency. If the optimization effect is greater than the preset effect value, when the water quality parameters after optimizing the cultivation environment do not change within the preset monitoring time, calibrate the corresponding water quality parameters as stable water quality parameters and suspend the operation of optimizing the cultivation environment;

[0037] Real-time monitor the deterioration trend of the water quality parameters in each three-dimensional sub-region after suspending the operation of optimizing the cultivation environment, and predict the change value of the water quality parameters in each three-dimensional sub-region according to the deterioration trend of the water quality parameters;

[0038] According to the change value of the water quality parameters, when the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, record the time length after suspending the operation of optimizing the cultivation environment until the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, and calibrate it as the first time length;

[0039] When the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, optimize the cultivation environment again, record the time length when the water quality parameters reach the stable water quality parameters, and calibrate it as the second time length. Perform periodic control operations on the water quality control equipment of the target cultivation pond according to the first time length and the second time length to obtain the first control optimization strategy for the cultivation environment;

[0040] If the optimization effect is not greater than the preset effect value, determine that the water quality control equipment of the target cultivation pond has reached the optimization upper limit, and perform an update operation on the water quality control equipment that has reached the optimization upper limit to obtain the second control optimization strategy.

[0041] The second aspect of the present invention also provides a control optimization system for the cultivation environment of Lutjanus erythropterus fry, which includes: a memory and a processor. The memory includes a program for the control optimization method of the cultivation environment of Lutjanus erythropterus fry. When the program for the control optimization method of the cultivation environment of Lutjanus erythropterus fry is executed by the processor, the following steps are implemented:

[0042] Divide the target cultivation pond of Lutjanus erythropterus fry into N three-dimensional sub-regions, and obtain the water quality change information and the activity frequency change information of Lutjanus erythropterus fry in each three-dimensional sub-region within a preset time period;

[0043] Determine the rejection behavior of Lutjanus erythropterus fry in three-dimensional sub-regions with different water quality changes within a preset time period according to the water quality change information and the activity frequency change information, and obtain a rejection behavior - water quality characteristic database;

[0044] Determine the suitable cultivation environment conditions for Lutjanus erythropterus fry according to the rejection behavior - water quality characteristic database;

[0045] Optimize the cultivation environment of the target cultivation pond according to the suitable cultivation environment conditions to obtain a control optimization strategy for the cultivation environment.

[0046] The present invention discloses a method and system for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry. The method includes: dividing the target cultivation pond of Lutjanus erythropterus fry into multiple three-dimensional sub-regions, and obtaining in real time the water quality change information and the change information of the activity frequency of Lutjanus erythropterus fry in each sub-region within a preset time period; analyzing the repulsion behaviors of Lutjanus erythropterus fry in different water quality change conditions towards each sub-region based on the above data, and establishing a repulsion behavior-water quality characteristic database; then determining the suitable cultivation environment conditions for Lutjanus erythropterus fry according to this database; and finally implementing environmental regulation on the cultivation pond according to the determined suitable environment conditions to form a scientific cultivation environment control and optimization strategy. The present invention can dynamically sense the sensitive reaction of fry to water quality changes, effectively improve the adaptability and refined management level of the cultivation environment, thereby promoting the healthy growth of fry, and improving the survival rate of seedling raising and the breeding benefit. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The flowchart of a method for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry according to the present invention is shown;

[0048] Figure 2 The flowchart of obtaining the water quality change information and the change information of the activity frequency according to the present invention is shown;

[0049] Figure 3 The flowchart of constructing the repulsion behavior-water quality characteristic database according to the present invention is shown;

[0050] Figure 4 The block diagram of a system for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry according to the present invention is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] In order to more clearly understand the above objects, features and advantages of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0052] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0053] Figure 1 The flowchart of a method for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry according to the present invention is shown.

[0054] As Figure 1 shown, the first aspect of the present invention provides a method for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry, including:

[0055] S102, divide the target cultivation pond for Lutjanus erythopterus fries into N three-dimensional sub-regions, and obtain the water quality change information and the activity frequency change information of Lutjanus erythopterus fries in each three-dimensional sub-region within a preset time period;

[0056] S104, determine the rejection behavior of Lutjanus erythopterus fries to the three-dimensional sub-regions with different water quality changes within a preset time period according to the water quality change information and the activity frequency change information, and obtain a rejection behavior-water quality characteristic database;

[0057] S106, determine the suitable cultivation environment conditions for Lutjanus erythopterus fries according to the rejection behavior-water quality characteristic database;

[0058] S108, optimize the cultivation environment of the target cultivation pond according to the suitable cultivation environment conditions to obtain an optimized cultivation environment control strategy.

[0059] It should be noted that by dividing the target cultivation pond for Lutjanus erythopterus fries into N three-dimensional sub-regions and obtaining the water quality change information and the fry activity frequency change information in each sub-region within a preset time period, the refined perception of the water environment and the dynamic monitoring of the fry behavior are realized, effectively improving the spatial resolution of environmental data collection and the time accuracy of behavior data; secondly, by analyzing the corresponding relationship between water quality changes and activity frequencies, the rejection behavior of Lutjanus erythopterus fries under specific water quality conditions is identified, and a rejection behavior-water quality characteristic database is constructed, which can quantify the adaptive differences of fries to different water quality changes; thirdly, according to this database, the water quality parameter combinations with lower rejection intensity are identified, and the suitable cultivation environment conditions are screened out by combining multi-dimensional data dimensionality reduction and optimization algorithms, realizing the scientific evaluation and prediction of the water quality environment with the best adaptability for fries; finally, the environment of the target cultivation pond is regulated according to the determined suitable environment conditions, and the regulation effect is feedback evaluated by combining the fry activity frequency, so as to construct a dynamic closed-loop cultivation environment control optimization strategy, effectively improving the responsiveness and intelligent level of water quality regulation, and finally realizing a significant increase in the health rate and survival rate of Lutjanus erythopterus fries and enhancing the stability and production efficiency of the aquaculture system.

[0060] Figure 2 The flowchart showing the water quality change information and the activity frequency change information obtained by the present invention is shown.

[0061] According to an embodiment of the present invention, the step of dividing the target cultivation pond for Lutjanus erythopterus fries into N three-dimensional sub-regions and obtaining the water quality change information and the activity frequency change information of Lutjanus erythopterus fries in each three-dimensional sub-region within a preset time period is specifically as follows:

[0062] S202. Divide the target cultivation pond of Lutjanus erythopterus fry into N three-dimensional sub-regions, and based on water quality sensors, obtain the water quality change information of each three-dimensional sub-region within a preset time period, including water temperature, salinity, pH value, dissolved oxygen concentration, ammonia nitrogen content, and water particulate matter concentration change data;

[0063] S204. Based on the camera equipment, obtain the fry activity video data of each three-dimensional sub-region within a preset time period, extract the video frame images of the fry activity video data, identify and count the number of Lutjanus erythopterus fry in each video frame image, and determine the activity frequency change information of Lutjanus erythopterus fry in each three-dimensional sub-region within a preset time period.

[0064] It should be noted that the activity frequency is the fry appearance frequency information in the three-dimensional sub-region per unit time.

[0065] According to the embodiments of the present invention, determining the rejection behavior of Lutjanus erythopterus fry in three-dimensional sub-regions with different water quality changes within a preset time period based on the water quality change information and activity frequency change information, and obtaining a rejection behavior - water quality feature database, specifically:

[0066] Extract the activity frequency information of each three-dimensional sub-region at each time node from the activity frequency change information according to a preset time step, perform normalization processing on the activity frequency information, and construct a three-dimensional space coordinate matrix for each time node, where each coordinate point contains the position parameters of the three-dimensional sub-region and the corresponding activity frequency value;

[0067] Based on the BIRCH clustering algorithm, construct a spatial clustering model for the three-dimensional sub-regions, perform clustering operations on the three-dimensional space coordinate matrix according to the spatial clustering model, and generate an initial clustering sub-cluster of the fry activity frequency distribution in the three-dimensional space by iteratively constructing a clustering feature tree structure including a branching factor and a spatial radius threshold;

[0068] Perform density clustering on the initial clustering sub-clusters, merge adjacent sub-clusters within the spatial radius threshold range to form a micro-cluster set representing the activity heat in different regional ranges, calculate the three-dimensional space volume covered by each micro-cluster and the average value of the fry activity frequency it contains, and generate a spatial activity heat distribution map of Lutjanus erythopterus fry in the target cultivation pond for each time node;

[0069] Perform time serialization processing on the spatial activity heat distribution maps of each time node, and calculate the temporal decay characteristic values of the activity heat of each three-dimensional sub-region through the dynamic time warping algorithm;

[0070] If the temporal decay characteristic value exceeds the preset threshold, it is determined that Lutjanus erythopterus has a rejection behavior within a preset time period, and the three-dimensional sub-region rejected by Lutjanus erythopterus is marked as a rejection sub-region;

[0071] Analyze the water quality change information of the exclusion sub-region based on the random forest algorithm, determine the exclusion correlation of Lutjanus erythopterus to different water quality changes, and construct an exclusion behavior-water quality characteristic database.

[0072] It should be noted that since the BIRCH algorithm has the ability to process large-scale spatial data, efficiently identify dense regions, and construct a clustering feature tree, it can hierarchically cluster the activity frequency values in the three-dimensional sub-region while maintaining spatial continuity. By continuously iteratively aggregating similar activity frequency points, an initial sub-cluster reflecting the actual activity concentration of the fry is formed, and then a set of micro-clusters is formed through density clustering merging, accurately depicting the spatial distribution of the activity heat of the fry in the three-dimensional water body at different times. This spatial activity heat distribution map can reflect the tendency or avoidance behavior of the fry in different sub-regions by visually quantifying the aggregation degree of the activity frequency. Further, as time goes by, the spatial heat maps at consecutive time nodes are constructed into a time series, and the temporal decay characteristic values of the activity heat in each sub-region are calculated through the dynamic time warping algorithm, which can effectively identify the significant downward trend of the activity frequency. If the activity heat in a certain region continues to decay and exceeds the preset threshold, it indicates that the fry avoids activities in this region for a long time, showing an exclusion behavior, and thus this region can be marked as an exclusion sub-region. For example, if the fry shows activities within the range of area A within a certain time range, but the activity heat indicates that the activity heat of the fry within the range of area A decreases after a certain time, it can be determined that the fry has an exclusion behavior towards area A.

[0073] Figure 3 The flowchart of constructing the exclusion behavior-water quality characteristic database of the present invention is shown.

[0074] According to an embodiment of the present invention, the analysis of the water quality change information of the exclusion sub-region based on the random forest algorithm, determining the exclusion correlation of Lutjanus erythopterus to different water quality changes, and constructing an exclusion behavior-water quality characteristic database are specifically as follows:

[0075] S302, calculate the difference between the temporal decay characteristic value of each exclusion sub-region and the preset threshold, and determine the exclusion intensity of Lutjanus erythopterus to the water quality change information corresponding to each exclusion sub-region according to the difference;

[0076] S304, use the water quality change information of the exclusion sub-region and the corresponding exclusion intensity as the positive sample data set, and the water quality change information of the non-exclusion sub-region as the negative sample data set;

[0077] S306. Build a random forest classification model. Use the positive sample dataset and the negative sample dataset as the input feature variables of the model, use the exclusion behavior label as the first output variable of the model, and the exclusion intensity as the second output variable of the model. Use the Gini coefficient as the node splitting criterion, and use the sampling with replacement method to extract a sample subset from the training set. Randomly select a sample subset for splitting at each node to obtain a preset number of decision trees;

[0078] S308. Train the random forest classification model based on the preset number of decision trees. Use the cross-validation method to adjust the model hyperparameters, calculate the feature importance scores of each water quality parameter in the model, and screen out the water quality parameter combinations whose correlation with the exclusion behavior exceeds the preset threshold according to the feature importance scores;

[0079] S310. According to the screened water quality parameter combinations and their corresponding exclusion behavior labels, build an exclusion behavior-water quality feature database with the water quality parameters as the feature dimension and the exclusion behavior intensity as the label. The database stores the mapping relationship between different water quality parameter combinations and the exclusion behavior and the exclusion behavior intensity.

[0080] It should be noted that due to the powerful multi-variable feature modeling ability and non-linear classification performance of random forest, it can mine the implicit relationship between variables and labels in complex high-dimensional data. The larvae of Lutjanus erythopterus show different degrees of repulsion behavior under specific water quality conditions, and this behavior may be comprehensively affected by various water quality parameters, such as water temperature, salinity, pH value, dissolved oxygen, ammonia nitrogen content, etc., which cannot be explained by a single parameter. Random forest avoids model overfitting and improves the model's perception ability of the interaction between features by integrating multiple decision trees. In each tree training, a subset of features is randomly selected from the feature set, and samples are sampled with replacement. In this method, by constructing a positive sample data set with the water quality change information in the repulsive sub-region and the corresponding repulsion intensity, and using the data in the non-repulsive sub-region as negative samples to form a binary classification or regression problem, random forest can effectively distinguish the feature differences between the two types of data. During the training process, the model constructs a large number of decision trees and combines the Gini coefficient for node splitting to continuously learn the splitting rules between water quality parameters and repulsion behavior. In the construction of each tree, the model evaluates the contribution of different water quality parameters to the splitting result, and finally realizes the classification judgment of new samples through the voting or average prediction results of all trees. More importantly, the random forest model can provide the feature importance score of each input feature after training, that is, measure the influence degree of each water quality parameter on the classification result in the model. This enables us to screen out the water quality parameter combinations highly related to the repulsion behavior based on feature importance, eliminate redundant or noise parameters, and thus construct a "repulsion behavior - water quality feature database". The hyperparameters include the number of decision trees, the maximum depth of each tree, the minimum number of samples per node, and the maximum number of features.

[0081] According to an embodiment of the present invention, determining the suitable cultivation environmental conditions for the larvae of Lutjanus erythopterus according to the repulsion behavior - water quality feature database specifically includes:

[0082] Extracting the water quality parameter combinations with a repulsion behavior intensity lower than a preset intensity threshold from the repulsion behavior - water quality feature database as a candidate suitable parameter set;

[0083] Performing dimensionality reduction processing on the water quality parameters of the candidate suitable parameter set by the principal component analysis method, calculating the covariance matrix of each water quality parameter, decomposing the covariance matrix to obtain the eigenvectors and variance contribution rates, and screening out the principal component vectors representing the water quality characteristics according to the variance contribution rates;

[0084] Reconstructing the principal component projection space of the water quality parameters based on the principal component vectors, sorting according to the absolute value of the load coefficient of each water quality parameter in the principal component projection space, and screening out the dominant water quality parameter combinations with a load coefficient exceeding the preset load threshold;

[0085] Construct an environmental fitness evaluation function for the juvenile Lutjanus erythropterus based on the combination of the dominant water quality parameters, input the environmental fitness evaluation function into the genetic algorithm, and perform global optimization and solution through population initialization, fitness calculation, selection, crossover, and mutation operations to obtain the optimal solution set of each dominant water quality parameter;

[0086] Determine the constraint boundary conditions of each water quality parameter in the combination of the dominant water quality parameters according to the optimal solution set, and combine the statistical distribution characteristics of the non-dominant water quality parameters in the candidate suitable parameter set to calculate the fluctuation range of the non-dominant water quality parameters at different quantiles, generate the water quality suitable parameter interval of the target cultivation pond for the juvenile Lutjanus erythropterus, and obtain the suitable cultivation environmental conditions.

[0087] It should be noted that during the cultivation process of the juvenile Lutjanus erythropterus, different water quality parameters show complex and non-linear effects on the behavior of the juveniles. Especially when there are multiple interrelated water quality variables, directly analyzing the relationship between each parameter and the rejection behavior is often affected by problems such as data redundancy, dimensionality disaster, and parameter collinearity, making it difficult to effectively extract representative and practical environmental characteristics. Therefore, after extracting the combination of water quality parameters with low rejection intensity based on the rejection behavior-water quality characteristic database, it is of great practical significance to introduce the principal component analysis (PCA) method for dimensionality reduction. As a classic linear dimensionality reduction technique, principal component analysis can convert multiple highly correlated water quality variables into a set of independent principal components by calculating the covariance matrix of the original water quality parameters and performing eigen-decomposition on it, thereby maximizing the retention of the effective information of the original data, reducing the data dimension, and eliminating redundant information. In this way, the parameter dimension that has the most significant impact on the behavior of Lutjanus erythropterus in the overall water quality change can be accurately identified, that is, the dominant water quality parameters. These dominant parameters are variables with larger load coefficients in PCA, representing their strong explanatory power and driving role in water quality changes. After screening these dominant water quality parameters, an environmental fitness evaluation function for the juveniles can be further constructed based on them and input into the genetic algorithm for global optimization to search for the optimal parameter combination. Finally, combined with statistical analysis, the reasonable fluctuation range of the non-dominant water quality parameters is determined, and a complete water quality suitable parameter interval is comprehensively constructed to determine the suitable cultivation environmental conditions for the juvenile Lutjanus erythropterus. The environmental fitness evaluation function is the sum of the scores of each dominant water quality parameter multiplied by the corresponding weights. The statistical constraint boundary conditions refer to determining the reasonable value range based on the statistical distribution characteristics of the water quality parameters in the low-rejection samples, usually including the maximum value, minimum value, mean value, standard deviation, and upper and lower limits corresponding to different quantiles.

[0088] According to the embodiment of the present invention, the cultivation environment of the target cultivation pond is optimized according to the suitable cultivation environmental conditions to obtain a cultivation environment control optimization strategy, specifically:

[0089] Optimize the cultivation environment of the target cultivation pond for Lutjanus erythopterus fries according to the suitable cultivation environment conditions, and monitor in real time the growth rate of the activity frequency of Lutjanus erythopterus fries in the excluded sub-region after the environmental optimization.

[0090] Evaluate the optimization effect of the cultivation environment according to the growth rate of the activity frequency. If the optimization effect is greater than the preset effect value, when the water quality parameters after the cultivation environment optimization do not change within the preset monitoring time, calibrate the corresponding water quality parameters as stable water quality parameters and suspend the cultivation environment optimization operation.

[0091] Monitor in real time the deterioration trend of the water quality parameters in each three-dimensional sub-region after suspending the cultivation environment optimization operation, and predict the change value of the water quality parameters in each three-dimensional sub-region according to the deterioration trend of the water quality parameters.

[0092] According to the change value of the water quality parameters, when the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, record the time length from when the cultivation environment optimization operation is suspended to when the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, and calibrate it as the first time length.

[0093] When the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, optimize the cultivation environment again, record the time length from when the water quality parameters reach the stable water quality parameters, and calibrate it as the second time length. Perform periodic control operations on the water quality control equipment of the target cultivation pond according to the first time length and the second time length to obtain the first control optimization strategy for the cultivation environment.

[0094] If the optimization effect is not greater than the preset effect value, determine that the water quality control equipment of the target cultivation pond has reached the optimization upper limit, and perform an update operation on the water quality control equipment that has reached the optimization upper limit to obtain the second control optimization strategy.

[0095] It should be noted that during the process of optimizing the cultivation environment, continuous water quality optimization will cause great power consumption and equipment loss. And when the equipment performance reaches the physical limit, the existing system cannot effectively identify the upper limit of equipment optimization, and still continuously executes invalid control instructions, resulting in waste of resources and inability to trigger equipment update decisions in a timely manner; by real-time monitoring the growth rate of the activity frequency of fry in the excluded sub-region after environmental optimization, combined with the prediction of the deterioration trend of water quality parameters and the calibration of time length, the dynamic coupling analysis of the control effect and the equipment state is realized. The first control optimization strategy constructs a periodic operation model of the water quality control equipment by recording the deterioration time of water quality parameters (the first time length) after pausing optimization and the time (the second time length) to re-optimize to a stable state, and can adaptively adjust the start-stop cycle of the equipment according to the deterioration rate of water quality, reduce the invalid operation time of the equipment while maintaining water quality stability, reduce energy consumption and extend the equipment life; the second control optimization strategy solves the problem of control failure caused by equipment performance decline in the traditional system by determining the forced update operation after the equipment reaches the optimization upper limit, ensuring the continuous effectiveness of the water quality control ability. The periodic operation control is to stop the work of the cultivation environment optimization equipment in the first time length and work in the second time length.

[0096] Figure 4 The block diagram of a control optimization system for the cultivation environment of Lutjanus erythropterus fry according to the present invention is shown.

[0097] In the second aspect of the present invention, a control optimization system 4 for the cultivation environment of Lutjanus erythropterus fry is further provided. The system includes: a memory 41 and a processor 42. The memory includes a program for the control optimization method of the cultivation environment of Lutjanus erythropterus fry. When the program for the control optimization method of the cultivation environment of Lutjanus erythropterus fry is executed by the processor, the following steps are realized:

[0098] Divide the target cultivation pond of Lutjanus erythropterus fry into N three-dimensional sub-regions, and obtain the water quality change information and the activity frequency change information of Lutjanus erythropterus fry in each three-dimensional sub-region within a preset time period;

[0099] Determine the exclusion behavior of Lutjanus erythropterus fry in different three-dimensional sub-regions with water quality changes within a preset time period according to the water quality change information and the activity frequency change information, and obtain an exclusion behavior-water quality feature database;

[0100] Determine the suitable cultivation environment conditions of Lutjanus erythropterus fry according to the exclusion behavior-water quality feature database;

[0101] Optimize the cultivation environment of the target cultivation pond according to the suitable cultivation environment conditions to obtain a cultivation environment control optimization strategy.

[0102] The present invention discloses a method and system for controlling and optimizing the cultivation environment of Lutjanus erythropterus fry. The method includes: dividing the target cultivation pond of Lutjanus erythropterus fry into multiple three-dimensional sub-regions, and acquiring in real time the water quality change information and the activity frequency change information of Lutjanus erythropterus fry in each sub-region within a preset time period; analyzing, based on the above data, the rejection behaviors of Lutjanus erythropterus fry in different water quality change conditions towards each sub-region, and establishing a rejection behavior-water quality characteristic database; then determining the suitable cultivation environment conditions for Lutjanus erythropterus fry according to this database; and finally implementing environmental regulation on the cultivation pond according to the determined suitable environment conditions to form a scientific cultivation environment control and optimization strategy. The present invention can dynamically sense the sensitive reaction of fry to water quality changes, effectively improve the adaptability and refined management level of the cultivation environment, thereby promoting the healthy growth of fry and improving the survival rate of seedling raising and the aquaculture benefit.

[0103] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed with each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be electrical, mechanical, or other forms.

[0104] The units described above as separate components may or may not be physically separated. The components shown as units may or may not be physical units; they can be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0105] In addition, in each embodiment of the present invention, each functional unit can be all integrated in a processing unit, or each unit can be separately used as a unit, or two or more units can be integrated in a unit; the above integrated units can be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0106] Those of ordinary skill in the art can understand that all or part of the steps to implement the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including those of the above method embodiments. The foregoing storage medium includes various media that can store program codes, such as removable storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0107] Alternatively, if the above integrated unit of the present invention is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. The foregoing storage medium includes various media that can store program codes, such as removable storage devices, ROM, RAM, magnetic disks, or optical discs.

[0108] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for controlling and optimizing the cultivation environment of Lutjanus erythopterus fry, characterized in that It includes the following steps: Divide the target cultivation pond of Lutjanus erythropterus fry into N three-dimensional sub-regions, and obtain the water quality change information and the activity frequency change information of Lutjanus erythropterus fry in each three-dimensional sub-region within a preset time period; Determine the repulsion behavior of Lutjanus erythropterus fry in different water quality change three-dimensional sub-regions within a preset time period according to the water quality change information and the activity frequency change information, and obtain a repulsion behavior - water quality characteristic database; Determine the suitable cultivation environmental conditions of Lutjanus erythropterus fry according to the repulsion behavior - water quality characteristic database; Optimize the cultivation environment of the target cultivation pond according to the suitable cultivation environmental conditions to obtain an optimized cultivation environment control strategy.

2. The control optimization method for the cultivation environment of Lutjanus erythopterus fry according to claim 1, wherein, The step of dividing the target cultivation pond of Lutjanus erythropterus fry into N three-dimensional sub-regions and obtaining the water quality change information and the activity frequency change information of Lutjanus erythropterus fry in each three-dimensional sub-region within a preset time period is specifically as follows: Divide the target cultivation pond of Lutjanus erythropterus fry into N three-dimensional sub-regions, and obtain the water quality change information in each three-dimensional sub-region within a preset time period based on water quality sensors, including water temperature, salinity, pH value, dissolved oxygen concentration, ammonia nitrogen content, and water body particulate matter concentration change data; Based on the camera equipment, obtain the fry activity video data in each three-dimensional sub-region within a preset time period, extract the video frame images of the fry activity video data, identify and count the number of Lutjanus erythropterus fry in each video frame image, and determine the activity frequency change information of Lutjanus erythropterus fry in each three-dimensional sub-region within a preset time period.

3. The control optimization method for the cultivation environment of Lutjanus erythropterus fry according to claim 1, characterized in that, The step of determining the repulsion behavior of Lutjanus erythropterus fry in different water quality change three-dimensional sub-regions within a preset time period according to the water quality change information and the activity frequency change information, and obtaining a repulsion behavior - water quality characteristic database is specifically as follows: Extract the activity frequency information of each three-dimensional sub-region at each time node from the activity frequency change information according to a preset time step, perform normalization processing on the activity frequency information, and construct a three-dimensional space coordinate matrix for each time node, where each coordinate point contains the position parameters of the three-dimensional sub-region and the corresponding activity frequency value; Based on the BIRCH clustering algorithm, construct a spatial clustering model of the three-dimensional sub-regions, perform clustering operations on the three-dimensional space coordinate matrix according to the spatial clustering model, and generate an initial clustering sub-cluster of the fry activity frequency distribution in the three-dimensional space by iteratively constructing a clustering feature tree structure including a branching factor and a spatial radius threshold; Perform density clustering on the initial clustering sub-clusters, merge adjacent sub-clusters within the spatial radius threshold range to form a micro-cluster set representing the activity heat of different regional ranges, calculate the three-dimensional space volume covered by each micro-cluster and the average value of the fry activity frequency contained therein, and generate a spatial activity heat distribution map of Lutjanus erythropterus fry in the target cultivation pond at each time node; Perform time serialization processing on the spatial activity heat distribution map of each time node, and calculate the temporal decay characteristic value of the activity heat of each three-dimensional sub-region through the dynamic time warping algorithm; If the temporal decay characteristic value exceeds the preset threshold, it is determined that Lutjanus erythropterus has a repulsion behavior within a preset time period, and the three-dimensional sub-region repelled by Lutjanus erythropterus is marked as a repulsion sub-region; Analyze the water quality change information in the exclusion sub-region based on the random forest algorithm, determine the exclusion correlation of Lutjanus erythopterus to different water quality changes, and construct an exclusion behavior-water quality characteristic database.

4. The control optimization method for the breeding environment of Lutjanus erythropterus fry according to claim 3, characterized in that, The analysis of the water quality change information in the exclusion sub-region based on the random forest algorithm, determining the exclusion correlation of Lutjanus erythopterus to different water quality changes, and constructing an exclusion behavior-water quality characteristic database are specifically as follows: Calculate the difference between the temporal decay eigenvalue of each exclusion sub-region and the preset threshold, and determine the exclusion intensity of Lutjanus erythopterus to the water quality change information corresponding to each exclusion sub-region according to the difference; Take the water quality change information of the exclusion sub-region and the corresponding exclusion intensity as the positive sample data set, and the water quality change information of the non-exclusion sub-region as the negative sample data set; Construct a random forest classification model. Take the positive sample data set and the negative sample data set as the input feature variables of the model, take the exclusion behavior label as the first output variable of the model, and the exclusion intensity as the second output variable of the model. Use the Gini coefficient as the node splitting criterion, and use the sampling method with replacement to extract sample subsets from the training set. Randomly select sample subsets for splitting at each node to obtain a preset number of decision trees; Train the random forest classification model based on the preset number of decision trees, use the cross-validation method to adjust the model hyperparameters, calculate the feature importance scores of each water quality parameter in the model, and screen out the water quality parameter combinations whose correlation with the exclusion behavior exceeds the preset threshold according to the feature importance scores; According to the selected water quality parameter combinations and their corresponding exclusion behavior labels, construct an exclusion behavior-water quality characteristic database with water quality parameters as the feature dimension and exclusion behavior intensity as the label. The database stores the mapping relationship between different water quality parameter combinations and exclusion behaviors and the exclusion behavior intensity.

5. The control optimization method for the cultivation environment of Lutjanus erythopterus fry, according to claim 1, is characterized in that, The determination of the suitable cultivation environment conditions for Lutjanus erythopterus fry according to the exclusion behavior-water quality characteristic database is specifically as follows: Extract the water quality parameter combinations with exclusion behavior intensity lower than the preset intensity threshold from the exclusion behavior-water quality characteristic database as the candidate suitable parameter set; Perform dimensionality reduction processing on the water quality parameters of the candidate suitable parameter set through the principal component analysis method. Calculate the covariance matrix of each water quality parameter, decompose the covariance matrix to obtain the eigenvector and variance contribution rate, and screen out the principal component vectors representing water quality characteristics according to the variance contribution rate; Reconstruct the principal component projection space of the water quality parameters based on the principal component vectors, sort according to the absolute value of the loading coefficient of each water quality parameter in the principal component projection space, and screen out the dominant water quality parameter combinations with the loading coefficient exceeding the preset loading threshold; Construct an environmental fitness evaluation function for Lutjanus erythopterus fry based on the dominant water quality parameter combinations, input the environmental fitness evaluation function into the genetic algorithm, and perform global optimization and solution through population initialization, fitness calculation, selection, crossover, and mutation operations to obtain the optimal solution set of each dominant water quality parameter; Determine the constraint boundary conditions of each water quality parameter in the dominant water quality parameter combination according to the optimal solution set, and combine the statistical distribution characteristics of the non-dominant water quality parameters in the candidate suitable parameter set to calculate the fluctuation range of the non-dominant water quality parameters at different quantiles, generate the water quality suitable parameter interval of the target cultivation pond for Lutjanus erythropterus fry, and obtain the suitable cultivation environment conditions.

6. The control optimization method for the cultivation environment of Lutjanus erythopterus fry, according to claim 1, is characterized in that Optimize the cultivation environment of the target cultivation pond according to the suitable cultivation environment conditions to obtain the optimized cultivation environment control strategy, specifically: Optimize the cultivation environment of the target cultivation pond for Lutjanus erythropterus fry according to the suitable cultivation environment conditions, and monitor the growth rate of the activity frequency of Lutjanus erythropterus fry in the exclusion sub-region after the environmental optimization in real time; Evaluate the optimization effect of the cultivation environment according to the growth rate of the activity frequency. If the optimization effect is greater than the preset effect value, when the water quality parameters after the cultivation environment optimization do not change within the preset monitoring time, mark the corresponding water quality parameters as stable water quality parameters and suspend the cultivation environment optimization operation; Monitor the deterioration trend of the water quality parameters in each three-dimensional sub-region in real time after suspending the cultivation environment optimization operation, and predict the change value of the water quality parameters in each three-dimensional sub-region according to the deterioration trend of the water quality parameters; According to the change value of the water quality parameters, when the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, record the time length when the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number after suspending the cultivation environment optimization operation, and mark it as the first time length; When the water quality parameters of the preset number of three-dimensional sub-regions reach the preset number, optimize the cultivation environment again, record the time length when the water quality parameters reach the stable water quality parameters, and mark it as the second time length. Perform periodic control operations on the water quality control equipment of the target cultivation pond according to the first time length and the second time length to obtain the first control optimization strategy for the cultivation environment; If the optimization effect is not greater than the preset effect value, determine that the water quality control equipment of the target cultivation pond has reached the optimization upper limit, and perform an update operation on the water quality control equipment that has reached the optimization upper limit to obtain the second control optimization strategy.

7. A control and optimization system for the cultivation environment of Lutjanus erythopterus fry, characterized in that, The control optimization system for the cultivation environment of Lutjanus erythropterus fry includes a memory and a processor. The memory includes a control optimization method program for the cultivation environment of Lutjanus erythropterus fry. When the control optimization method program for the cultivation environment of Lutjanus erythropterus fry is executed by the processor, the following steps are implemented: Divide the target cultivation pond for Lutjanus erythropterus fry into N three-dimensional sub-regions, and obtain the water quality change information and the activity frequency change information of Lutjanus erythropterus fry in each three-dimensional sub-region within a preset time period; Determine the exclusion behavior of Lutjanus erythropterus fry in the three-dimensional sub-regions with different water quality changes within a preset time period according to the water quality change information and the activity frequency change information, and obtain the exclusion behavior-water quality characteristic database; Determine the suitable cultivation environment conditions of Lutjanus erythropterus fry according to the exclusion behavior-water quality characteristic database; Optimize the cultivation environment of the target cultivation pond according to the suitable cultivation environment conditions to obtain the optimized cultivation environment control strategy.

Images