A method and system for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives.

Abstract

This invention provides a method and system for optimizing the deployment of nearshore fisheries resource monitoring stations and navigation routes based on multiple objectives. The invention first determines the optimization objectives, then selects the optimal coupling model through evaluation indicators, and obtains the true distribution data of resource density through the optimal coupling model. It employs the Delphi method combined with information entropy and coefficient of variation as dual objective weights, along with a target-oriented correction coefficient, to obtain the dynamic weights of the optimization objectives. Three-level dynamic stratified sampling is performed using a heterogeneity threshold, and the optimal number of stations is determined through precision-cost elasticity analysis. Finally, a multi-objective optimization function is constructed, integrating navigation distance, fuel cost, and operational time matching. The optimal navigation route is solved using SA-PSO. This invention effectively solves the problems of systematic bias and insufficient resource adaptability in traditional surveys, significantly improves sampling representativeness and navigation operation efficiency, reduces survey costs, and can be dynamically adjusted according to resource conservation and fisheries production needs.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and fisheries technology, and in particular to a method and system for the deployment and navigation route optimization of nearshore fisheries resource monitoring stations based on multiple objectives. Background Technology

[0002] Fisheries resource surveys are a crucial data source for fisheries scientific research. A preliminary fisheries resource survey system covering nearshore waters has been established. However, current surveys largely employ a combination of fixed stations and mobile surveys. The deployment of survey stations and routes lacks scientific and technical support, making it difficult to simultaneously achieve resource sampling accuracy, long-term data continuity, and cost control within limited vessel time, budget, and sea condition windows. Furthermore, current station deployment and optimization primarily focus on single objectives such as resource abundance and diversity, failing to address multiple fisheries management goals, including the supply of high-quality marine protein, total resource management, and resource conservation.

[0003] In addition, existing true value assessments of resource density only consider the spatiotemporal characteristics of resources, and the model fit deviates greatly from the actual distribution of resources, which cannot provide accurate basic data for subsequent sampling design. Moreover, the sampling stratification adopts a fixed geographical / environmental basis, and the number of stations within the stratum is allocated only according to a single heterogeneity coefficient, resulting in insufficient sampling accuracy in resource-complex areas and redundancy of stations in uniform areas.

[0004] Furthermore, existing route optimization does not take into account real-time sea conditions, station operation time windows, survey vessel performance, and other conditions, resulting in low feasibility of route optimization and a high risk of wasting ship time and funds.

[0005] Therefore, there is an urgent need to provide a scheme for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and system for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives, in order to determine the optimal number of survey stations, the best sampling method, and the optimal navigation route for each sea area.

[0007] In a first aspect, the present invention provides a method for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives, comprising the following steps:

[0008] S1) Select species richness, total resource density, and species resource density as optimization objectives;

[0009] S2) Calculate the spatial variation coefficient and temporal variation coefficient of resources based on historical fishery resource data and environmental factor data. At the same time, collect environmental factor data of the surveyed sea area and calculate the Pearson correlation coefficient between environmental factors and resource density.

[0010] S3) Select GLM, GAM, Kriging, and sdmTMB models as candidate models, use spatial variation coefficient and temporal variation coefficient as spatiotemporal feature terms, and use environmental factors with Pearson correlation coefficient greater than or equal to a preset threshold as environmental driving terms, and integrate them into the input layer of the candidate models to construct an environment-spatiotemporal coupled candidate model.

[0011] S4) Using the mean absolute error (MAE) and coefficient of determination Percentage Bias (PBIAS) and Spatiotemporal Environment Fit (TEF) are used as evaluation indicators to select the optimal coupling model; the true distribution data of resource density is obtained through the optimal coupling model, and the comprehensive heterogeneity coefficient of resource spatiotemporal environment is output.

[0012] S5) Obtain expert subjective weights through the Delphi method, calculate information entropy objective weights and coefficient of variation objective weights based on the true value data of resource density, fuse them according to a preset ratio to obtain the basic weights of the optimization target, and correct the basic weights of the optimization target.

[0013] S6) The entire survey sea area is divided into non-overlapping micro-regions based on geographical zoning, environmental zoning, and heterogeneity zoning, and the number of sampling stations in each micro-region is dynamically allocated; under the optimal sampling method for the entire area, the optimal number of survey stations is determined by the accuracy-cost elasticity coefficient.

[0014] S7) Construct a comprehensive optimization objective function that minimizes the navigation distance, fuel cost, and operation time matching degree; use the particle swarm optimization (PSO) algorithm for global search to generate an initial solution group, and use the simulated annealing algorithm for local fine optimization to output the optimal navigation route that satisfies all constraints.

[0015] Preferably, in step S4), the optimal coupling model is selected based on the evaluation index, as follows:

[0016] First, the reconstructed GLM, GAM, Kriging, and sdmTMB models were screened using percentage bias PBIAS, retaining candidate models whose percentage bias PBIAS falls within the range of [-25%, 25%].

[0017] Then, the Spatiotemporal Environment Fit Factor (TEF) is used to perform a second screening of the candidate models selected in the first screening, and candidate models with a Spatiotemporal Environment Fit Factor (TEF) ≥ 0.55 are retained.

[0018] Using the coefficient of determination The candidate models selected in the second round of screening were then screened a third time, retaining the coefficient of determination. Candidate models;

[0019] The candidate models selected in the third screening were screened for the fourth time using the mean absolute error (MAE), and the candidate model with the smallest mean absolute error (MAE) was selected as the optimal coupling model.

[0020] If only one candidate model remains after a certain screening, it is directly determined to be the optimal coupling model.

[0021] Preferably, in step S4), the spatial interpolation and prediction of the optimal coupling model are used to obtain the true distribution data of fishery resource density for the entire surveyed sea area, as follows:

[0022] The surveyed sea area is divided into grids with equal intervals of latitude and longitude, generating a set of latitude and longitude coordinates for each grid node, with each grid node serving as a prediction unit.

[0023] The latitude and longitude coordinates, standardized values ​​of spatiotemporal feature terms, and standardized values ​​of environmental driving terms of the grid nodes are input into the optimal coupling model to calculate the predicted value of fishery resource density for each grid node.

[0024] The latitude and longitude coordinates of the grid nodes are matched with the corresponding true resource density values ​​to generate a dataset of true resource density distribution in the surveyed sea area, and a spatial distribution map of the true resource density values ​​is drawn.

[0025] Secondly, the present invention provides a multi-objective nearshore fisheries resource monitoring station deployment and navigation route optimization system, comprising:

[0026] The optimization objective determination module is used to select species richness, total resource density, and species resource density as optimization objectives.

[0027] The quantification module is used to calculate the spatial and temporal variation coefficients of resources based on historical fishery resource data and environmental factor data. At the same time, it collects environmental factor data of the surveyed sea area and calculates the Pearson correlation coefficient between environmental factors and resource density to quantitatively characterize the spatiotemporal and environmental characteristics of resources.

[0028] The model reconstruction module is used to select GLM, GAM, Kriging, and sdmTMB models as candidate models, use spatial variation coefficient and temporal variation coefficient as spatiotemporal feature terms, and use environmental factors with Pearson correlation coefficient greater than or equal to a preset threshold as environmental driving terms, and integrate them into the input layer of the candidate model to construct an environment-spatiotemporal coupled candidate model.

[0029] The model selection module utilizes the mean absolute error (MAE) and coefficient of determination. Percentage Bias (PBIAS) and Spatiotemporal Environment Fit (TEF) are used as evaluation indicators to select the optimal coupling model; the true distribution data of resource density is obtained through the optimal coupling model, and the comprehensive heterogeneity coefficient of resource spatiotemporal environment is output.

[0030] The optimization weight calculation module obtains expert subjective weights through the Delphi method, calculates objective weights of information entropy and coefficient of variation based on true resource density data, merges them according to a preset ratio to obtain the basic weights of the optimization target, and then corrects the basic weights of the optimization target.

[0031] The site determination module divides the entire survey sea area into non-overlapping micro-regions based on geographical zoning, environmental zoning, and heterogeneity zoning, and dynamically allocates the number of sampling sites in each micro-region; under the optimal sampling method for the entire area, the optimal number of survey sites is determined by the accuracy-cost elasticity coefficient.

[0032] The route optimization module is used to construct a comprehensive optimization objective function that minimizes the flight distance, fuel cost, and operation time matching. It uses the particle swarm optimization (PSO) algorithm for global search to generate an initial solution group, and then uses simulated annealing algorithm for local fine-tuning to output the optimal flight route that satisfies all constraints.

[0033] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations.

[0034] The beneficial technical effects of this invention are as follows:

[0035] 1. This invention is based on the comprehensive heterogeneity coefficient of resource spatiotemporal environment, and dynamically stratifies according to three levels: geographical zoning, environmental zoning, and heterogeneity classification. It combines heterogeneity and accuracy to allocate the number of stations, so as to achieve dense sampling in high heterogeneity areas and fine sampling in low heterogeneity areas, and accurately match the differences in resource distribution. At the same time, the optimal fitting model is screened layer by layer through quantitative indicators such as PBIAS, R², and MAE to ensure the reliability of the true value estimation of resource density.

[0036] 2. This invention assigns weights to the optimization objective based on expert subjective weights, information entropy, and objective weights of the variation system, making the survey results closer to reality;

[0037] 3. This invention takes the shortest sailing distance, the lowest fuel cost, and the highest operation time matching degree as the optimization objective function, and uses the real-time sea state resistance coefficient, the channel detour coefficient, and the operation time window matching coefficient as constraints. It solves the problem through the SA-PSO hybrid algorithm. The PSO algorithm completes the global initial solution search, and the SA algorithm is used to achieve local fine optimization. Combined with the Metropolis criterion, it escapes the local optimum, which greatly improves the feasibility and efficiency of the survey and navigation.

[0038] 4. This invention determines the optimal number of survey stations by using multiple sampling methods and the precision-cost elasticity coefficient, which greatly improves the accuracy of station deployment and enhances the economic efficiency of fishery resource surveys. Attached Figure Description

[0039] Figure 1 This is a schematic flowchart of the method of the present invention;

[0040] Figure 2 This is a schematic diagram of the process for selecting the optimal coupling model in an embodiment of the present invention;

[0041] Figure 3 This is a schematic diagram of the process for solving the optimal navigation route in an embodiment of the present invention. Detailed Implementation

[0042] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0043] like Figure 1 As shown in the figure, this embodiment provides a method for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives, including the following steps:

[0044] S1) Select species richness, total resource density, and species resource density as optimization objectives;

[0045] S2) Calculate the spatial variation coefficient and temporal variation coefficient of resources based on historical fishery resource data and environmental factor data. At the same time, collect environmental factor data of the surveyed sea area and calculate the Pearson correlation coefficient between environmental factors and resource density to quantitatively characterize the spatiotemporal and environmental basic characteristics of resources.

[0046] In this embodiment, the following is adopted: The criteria perform outlier checks on resource density data and environmental factor data. If a site's data value exceeds... If the value falls outside the specified range, it is considered an outlier, and all data corresponding to that site is removed; among them, The mean of the dataset. The standard deviation of the dataset;

[0047] Kriging interpolation combined with station spatial coordinates was used to complete the missing values; if the missing percentage was > 5%, the corresponding station data for that survey year / period was removed.

[0048] After standardization, a standardized resource-environment fusion dataset is obtained, which includes station number, latitude and longitude, survey time, standardized total resource density, standardized resource density of major economic species, standardized species richness, standardized water depth, standardized salinity, and standardized water temperature.

[0049] In this embodiment, the resource spatial variation coefficient is used to quantify the degree of difference in the distribution of fishery resource density at different spatial stations in the surveyed sea area within the same survey period; by selecting standardized resource density data for a certain survey period, grouping them according to the survey station, and recording them as datasets. ,in The number of effective survey stations was determined; and the mean resource density for each survey period was calculated. and standard deviation The ratio of the standard deviation to the mean was used as the spatial variation coefficient of resources during the survey period. ,Right now:

[0050]

[0051] In the formula, , These represent the mean and standard deviation of resource density during the survey period, respectively.

[0052] In this embodiment, the time variation coefficient is used to quantify the dynamic changes in fishery resource density at the same spatial station or the entire surveyed sea area during different survey periods. By using standardized resource density data from all valid stations in the surveyed sea area at different survey periods, the mean resource density of the sea area for each period is calculated to form a time series dataset. , The number of valid survey periods; the coefficient of variation of time. Represented as:

[0053]

[0054] In the formula, , These represent the mean and standard deviation of resource density for the survey time series, respectively.

[0055] In this embodiment, the Pearson correlation coefficient is used to quantify the degree of linear correlation between environmental factors (water depth, salinity, water temperature) and fishery resource density. The Pearson correlation coefficient is calculated as follows:

[0056] Select the preprocessed, standardized resource-environment fusion dataset, and construct paired datasets for resource density and single environmental factors, denoted as . ,in, To standardize the resource density dataset, To standardize a single environmental factor, The number of valid stations for synchronization;

[0057] Calculate separately The mean, sample covariance, and sample standard deviation are used to obtain the Pearson correlation coefficient. for:

[0058]

[0059] In the formula, >0 indicates a positive correlation. <0 indicates a negative correlation; express The sample covariance; for The sample standard deviation; , The first Standardized resource density values ​​and standardized environmental factor values ​​for each station; , They are respectively The mean;

[0060] In this embodiment, if ≥0.7 indicates strong linear correlation, 0.3≤ <0.7 indicates moderate linear correlation. A value <0.3 indicates a weak linear correlation.

[0061] S3) Select GLM, GAM, Kriging, and sdmTMB models as candidate models, use spatial variation coefficient and temporal variation coefficient as spatiotemporal feature terms, and use environmental factors with Pearson correlation coefficient ≥ 0.3 as environmental driving terms, and integrate them into the input layer of the candidate models to construct an environment-spatiotemporal coupled candidate model.

[0062] In this embodiment, the reconstructed GLM model is represented as:

[0063]

[0064] In the formula, Indicates the join function; For response variables The expected value of the condition; For the intercept term; , These are the linear regression coefficients for spatial variation and time variation, respectively; Indicates the first Environmental driver items The linear regression coefficients; The number of environmental drivers; This is the random error term;

[0065] In this embodiment, the reconstructed GAM model is represented as:

[0066]

[0067] In the formula, This represents the measured value of resource density; For the intercept term; , It is a smoothing function; For the first Environmental driver items The smoothing function;

[0068] In this embodiment, for the Kriging model, the spatiotemporal feature term is used as a spatial covariate and the environment-driven term is used as a collaborative variable. The reconstructed Kriging model is expressed as follows:

[0069]

[0070] In the formula, Unknown position The optimal interpolated estimate of resource density at that location; For known station locations Measured value of resource density at the location; The number of known measured stations; These are the Kriging weighting coefficients; For known station locations Measured values ​​of environmental driving factors at the location. Number of stations with known environmental factors; These are the weighting coefficients for environmental covariates; , These are the spatial variation coefficients. Coefficient of variation over time The correction factor.

[0071] In this embodiment, the reconstructed sdmTMB model is represented as:

[0072]

[0073] In the formula, This represents the measured value of resource density; For the intercept term; It is a two-dimensional smoothing function; Indicates the first A smoothing function for each environment-driven term; For Matérn Gaussian random fields; This is an independent random error term.

[0074] The basic parameters of the reconstructed GLM, GAM, Kriging, and sdmTMB models were calibrated and fitted to obtain preliminary candidate models of environment-spatiotemporal coupling.

[0075] S4) Using the mean absolute error (MAE) and coefficient of determination Percentage Bias (PBIAS) and Spatiotemporal Environment Fit (TEF) are used as evaluation indicators to select the optimal coupling model; the true distribution data of resource density is obtained through the optimal coupling model, and the comprehensive heterogeneity coefficient of resource spatiotemporal environment is output.

[0076] In this embodiment, the optimal coupling model is selected based on evaluation metrics, as follows:

[0077] First, the reconstructed GLM, GAM, Kriging, and sdmTMB models were screened using percentage bias PBIAS, retaining candidate models whose percentage bias PBIAS falls within the range of [-25%, 25%].

[0078] Then, the Spatiotemporal Environment Fit Factor (TEF) is used to perform a second screening of the candidate models selected in the first screening, and candidate models with a Spatiotemporal Environment Fit Factor (TEF) ≥ 0.55 are retained.

[0079] Using the coefficient of determination The candidate models selected in the second round of screening were then screened a third time, retaining the coefficient of determination. Candidate models;

[0080] The candidate models selected in the third screening were screened for the fourth time using the mean absolute error (MAE), and the candidate model with the smallest mean absolute error (MAE) was selected as the optimal coupling model.

[0081] If only one candidate model remains after a certain screening, it is directly determined to be the optimal coupling model.

[0082] In this embodiment, spatial interpolation and prediction using the optimal coupling model are used to obtain the true distribution data of fishery resource density across the entire surveyed sea area, as detailed below:

[0083] The surveyed sea area is divided into grids with equal intervals of latitude and longitude, generating a set of latitude and longitude coordinates for each grid node, with each grid node serving as a prediction unit.

[0084] The latitude and longitude coordinates, standardized values ​​of spatiotemporal feature terms, and standardized values ​​of environmental driving terms of the grid nodes are input into the optimal coupling model to calculate the predicted value of fishery resource density for each grid node.

[0085] The latitude and longitude coordinates of the grid nodes are matched with the corresponding true resource density values ​​to generate a dataset of true resource density distribution in the surveyed sea area, and a spatial distribution map of the true resource density values ​​is drawn.

[0086] In this embodiment, the resource spatiotemporal environment comprehensive heterogeneity coefficient for:

[0087] ;

[0088] in, The coefficient of variation of environmental factors. ; The percentage of the absolute value of the Pearson correlation coefficient between environmental factors and resource density; For the first The environmental factor variation coefficient of each environmental driver.

[0089] In this embodiment, the mean absolute error (MAE) is expressed as:

[0090]

[0091] In the formula, , They represent the first Model fitted values ​​and measured values ​​of fishery resource density for each station.

[0092] In this embodiment, the determination coefficient Represented as:

[0093]

[0094]

[0095] In the formula, This represents the average of the measured values ​​of resource density.

[0096] In this embodiment, the percentage deviation PBIAS is expressed as:

[0097]

[0098] In this embodiment, the spatiotemporal environment fitting degree (TEF) is expressed as:

[0099]

[0100] In the formula, The spatiotemporal fit degree; For environmental fit; To adjust the coefficient, b = 0.4 in this embodiment;

[0101]

[0102]

[0103] In the formula, The mean absolute error of fitting the spatiotemporal feature terms individually; The mean absolute error of the individual fit for the environment-driven term; These represent the maximum and minimum values ​​of the measured resource density, respectively.

[0104] S5) Obtain expert subjective weights through the Delphi method, calculate information entropy objective weights and coefficient of variation objective weights based on the true value data of resource density, fuse them according to a preset ratio to obtain the basic weights of the optimization target, and correct the basic weights of the optimization target.

[0105] In this embodiment, for each target Calculate the first The proportion of a sample's standardized value to the sum of all standardized values ​​of the corresponding target samples; and through information entropy. To measure the uncertainty of the target data, and through information entropy Calculate the weight coefficients of the optimization objective. ,Right now:

[0106]

[0107]

[0108]

[0109] In the formula, The proportion of the target sample; For the first The sample at the th Standardized values ​​under an optimization objective.

[0110] In this embodiment, the coefficient of variation Reflecting the relative dispersion of the target data, through the coefficient of variation Normalization is performed to obtain the objective weights of the coefficient of variation. ,Right now:

[0111]

[0112]

[0113] In the formula, , The target The sample mean and sample standard deviation.

[0114] The weighted linear fusion method is used to fuse expert subjective weights, information entropy objective weights, and variation coefficient objective weights according to a preset ratio to obtain the basic weights of the optimization objective. ,Right now:

[0115] ;

[0116] In the formula, , , These are the proportional coefficients for the corresponding weights. + + ; Subjective weighting by experts; For the first The basic weights of each optimization objective;

[0117] Set target-oriented correction coefficient The weights of the corrected optimization objective are... Represented as:

[0118]

[0119] In the formula, .

[0120] S6) The entire survey sea area is divided into non-overlapping micro-regions based on geographical, environmental, and heterogeneous zoning, and the number of sampling stations in each micro-region is dynamically allocated; under the optimal sampling method for the entire area, the optimal number of survey stations is determined by the accuracy-cost elasticity coefficient; specifically, the following steps are included:

[0121] S61) The entire survey sea area is divided according to geographical features into Each non-overlapping main area ;

[0122] S62) Each main area is divided according to significant environmental factors. Divided into Sub-district ;

[0123] S63) Extract each sub-region Resource spatiotemporal environment comprehensive heterogeneity coefficient of all grid nodes Calculate the average heterogeneity coefficient of the sub-region. And according to the preset heterogeneity classification threshold, the sub-regions are divided into... Divided into highly heterogeneous micro-regions Medium heterogeneous micro-regions Low heterogeneity micro-regions ;Right now:

[0124]

[0125] like This indicates a highly heterogeneous micro-region. ;like This is a micro-region of medium heterogeneity. ;like This indicates a low heterogeneity micro-region. ;

[0126] Finally obtained Micro-area Number of grid nodes in each micro-region ;

[0127] S64), for each gradient of total number of stations Calculate each micro-region Site weight allocation Each gradient corresponds to a set of candidate total stations; that is:

[0128]

[0129] In the formula, For micro-regions The average heterogeneity coefficient; For micro-regions The accuracy requirement threshold;

[0130] Therefore, the number of sampling stations in each micro-region for:

[0131] ;

[0132] In the formula, Indicates the rounding operation;

[0133] S65), for each gradient of total number of stations We conducted K=1000 computer simulations using simple random sampling, systematic sampling, and dynamic stratified sampling respectively; and calculated the relative estimation error REE and the absolute value of the relative deviation |RB|; as follows:

[0134] The simple random sampling described above randomly selects nodes from the entire grid. Each site is sampled without replacement; the systematic sampling sorts the entire grid by latitude and longitude, randomly determines the starting point, and samples are taken at fixed intervals. Each site; the dynamic stratified sampling in each micro-region Inside, from Randomly selected from each grid node Each site, after all micro-areas are sampled, is summarized as follows: One sampling site.

[0135] In this embodiment, for each optimization objective Based on the results of K resampling, calculate the relative estimation error REE and the absolute value of the relative deviation |RB|:

[0136] ;

[0137] In the formula, For the first Secondary sampling for the first Estimates of the optimization objective; For the first The optimization objective is the true value of the entire domain; For the first The relative estimation error REE of each optimization objective;

[0138]

[0139] In the formula, For the first The absolute value of the relative deviation of each optimization objective |RB|;

[0140] S66) Combining the weights of the modified optimization objective Construct the comprehensive evaluation indices CREE and C|RB|, namely:

[0141]

[0142]

[0143] For each gradient of total number of stations By comparing the CREE or C|RB| values ​​of the three sampling methods, the sampling scheme with the smallest value is selected as the optimal sampling under the corresponding gradient.

[0144] The frequency of optimal sampling under all total station number gradients is counted, and the optimal sampling method with the highest frequency is selected as the optimal sampling method for the entire domain.

[0145] S67) Calculate the elasticity coefficients of the gradient of the total number of stations. And filter the elasticity coefficients of all total station number gradients. The total number of stations gradient is ≥1, and the smallest total number of stations gradient is selected. As the optimal number of survey sites;

[0146]

[0147] In the formula, Gradient of total number of stations from Increase to The rate of change of accuracy over time; CREE was selected as the accuracy. ; Gradient of total number of stations from Increase to Cost change rate at time; Gradient of total number of stations The total cost of the survey.

[0148] S7) Construct a comprehensive optimization objective function that minimizes the navigation distance, fuel cost, and operational time matching degree; use the particle swarm optimization (PSO) algorithm for global search to generate an initial solution group, and use simulated annealing algorithm for local fine-tuning to output the optimal navigation route that satisfies all constraints; specifically including the following steps:

[0149] S71) Assume the survey vessel departs from the dock, visits all stations in sequence, and finally returns to the dock;

[0150] S72) Based on the shortest travel distance, lowest fuel cost, and highest operation time matching degree, an optimization objective function is established, namely:

[0151]

[0152] ;

[0153]

[0154] In the formula, This is the straight-line distance from the dock to the first station; This is the straight-line distance from the last station to the dock; This represents the straight-line distance between two adjacent stations. ; ; This is the sea state drag coefficient; For the survey vessel's fuel consumption rate; For the first Time window matching coefficient for each site; The optimal number of survey sites; , , There are three objective functions;

[0155] S73), the three objective functions , , The results are integrated into the final comprehensive objective function. ,Right now:

[0156] ;

[0157] In the formula, , , The weights are the corresponding objective function weights. + + ; , The normalized objective function;

[0158] S74) The Particle Swarm Optimization (PSO) algorithm is used to perform a global search to generate an initial solution group, as detailed below:

[0159] Each particle corresponds to a station. Randomly generated A valid access sequence forms the initial solution group;

[0160] Calculate the comprehensive objective function value for each particle. ;

[0161] Speed ​​updated to: ;

[0162] Location update: ;

[0163] In the formula, Inertial weights; For the first The current particle in the next iteration The optimal position of an individual The optimal position for the entire population; A random number in the range [0,1]. , These are learning factors; Represents particles Current location; Indicates the current particle velocity;

[0164] When the maximum number of iterations is reached, output the global optimum position of the population. As the global optimal solution for the population.

[0165] S75) Local fine-grained optimization is performed using the simulated annealing algorithm (SA), as follows:

[0166] S751) The population global optimal solution output by the Particle Swarm Optimization (PSO) algorithm. As the initial current solution of SA Calculate its fitness value ;

[0167] S752), for the initial current solution Randomly perturb the order of the stations to generate neighborhood solutions. ;

[0168] S753), Calculate the neighborhood solution Sea state drag coefficient Channel detour coefficient Time window matching coefficient If all constraints are satisfied, then calculate its fitness value. Otherwise, discard the neighborhood solution and regenerate; the constraints are as follows:

[0169] ;

[0170]

[0171] ;

[0172] , ;

[0173] In the formula, , , These are ocean current speed, wind speed, and wave height, respectively. , , For the corresponding coefficients; Maximum sea state resistance; , These represent the actual distance traveled after bypassing the restricted area and the actual distance traveled while sailing in a straight line, respectively. This is the maximum detour coefficient for waterways; This refers to the actual speed. Rated speed; The maximum wave height that the ship can withstand;

[0174] S754), if fitness value , solve the neighborhood As the new current solution;

[0175] If fitness value Acceptable according to the Metropolis criterion, that is:

[0176] ;

[0177] like This represents the probability of receiving the signal. For temperature; in each iteration, the temperature... according to = attenuation, The attenuation coefficient;

[0178] like If the new solution is found, then accept the new solution; otherwise, retain the original solution.

[0179] S755), if continuous If no new solution is accepted in the next iteration, or the maximum number of iterations is reached, stop the SA iteration and output the current optimal solution. This refers to the optimal site access order.

[0180] Taking the autumn survey in the nearshore waters of the South China Sea as an example, the optimal number of stations is 50, and the CREE value of dynamic stratified sampling is 9.2, which is significantly lower than that of simple random sampling (17.5) and systematic sampling (15.8). After SA-PSO optimization, the total sailing distance of the route is 1025 nautical miles, the fuel cost is reduced by 23% compared with the traditional scheme, the operation time window matching degree is 98%, and all constraints are met.

[0181] Example 2

[0182] This embodiment provides a multi-objective nearshore fisheries resource monitoring station deployment and navigation route optimization system, including:

[0183] The optimization objective determination module is used to select species richness, total resource density, and species resource density as optimization objectives.

[0184] The quantification module is used to calculate the spatial and temporal variation coefficients of resources based on historical fishery resource data and environmental factor data. At the same time, it collects environmental factor data of the surveyed sea area and calculates the Pearson correlation coefficient between environmental factors and resource density to quantitatively characterize the spatiotemporal and environmental characteristics of resources.

[0185] The model reconstruction module is used to select GLM, GAM, Kriging, and sdmTMB models as candidate models, use spatial variation coefficient and temporal variation coefficient as spatiotemporal feature terms, and use environmental factors with Pearson correlation coefficient greater than or equal to a preset threshold as environmental driving terms, and integrate them into the input layer of the candidate model to construct an environment-spatiotemporal coupled candidate model.

[0186] The model selection module utilizes the mean absolute error (MAE) and coefficient of determination. Percentage Bias (PBIAS) and Spatiotemporal Environment Fit (TEF) are used as evaluation indicators to select the optimal coupling model; the true distribution data of resource density is obtained through the optimal coupling model, and the comprehensive heterogeneity coefficient of resource spatiotemporal environment is output.

[0187] The optimization weight calculation module obtains expert subjective weights through the Delphi method, calculates objective weights of information entropy and coefficient of variation based on true resource density data, merges them according to a preset ratio to obtain the basic weights of the optimization target, and then corrects the basic weights of the optimization target.

[0188] The site determination module divides the entire survey sea area into non-overlapping micro-regions based on geographical zoning, environmental zoning, and heterogeneity zoning, and dynamically allocates the number of sampling sites in each micro-region; under the optimal sampling method for the entire area, the optimal number of survey sites is determined by the accuracy-cost elasticity coefficient.

[0189] The route optimization module is used to construct a comprehensive optimization objective function that minimizes the flight distance, fuel cost, and operation time matching. It uses the particle swarm optimization (PSO) algorithm for global search to generate an initial solution group, and then uses simulated annealing algorithm for local fine-tuning to output the optimal flight route that satisfies all constraints.

[0190] Example 3

[0191] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the method for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations as described in Embodiment 1.

[0192] In this embodiment, the memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. A processor, coupled to the memory, is used to execute computer programs stored in the memory.

[0193] The computer program includes computer program code, which may be in the form of source code, object code, executable file, or some intermediate form.

[0194] The embodiments and descriptions above are merely illustrative of the principles and preferred embodiments of the present invention. Various changes and modifications may be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed.

Claims

1. A method for optimizing the deployment and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives, characterized in that, Includes the following steps: S1) Select species richness, total resource density, and species resource density as optimization objectives; S2) Calculate the spatial variation coefficient and temporal variation coefficient of resources based on historical fishery resource data and environmental factor data. At the same time, collect environmental factor data of the surveyed sea area and calculate the Pearson correlation coefficient between environmental factors and resource density. S3) Select GLM, GAM, Kriging, and sdmTMB models as candidate models, use spatial variation coefficient and temporal variation coefficient as spatiotemporal feature terms, and use environmental factors with Pearson correlation coefficient greater than or equal to a preset threshold as environmental driving terms, and integrate them into the input layer of the candidate models to construct an environment-spatiotemporal coupled candidate model. S4) Using the mean absolute error (MAE) and coefficient of determination Percentage Bias (PBIAS) and Spatiotemporal Environment Fit (TEF) are used as evaluation indicators to select the optimal coupling model; the true distribution data of resource density is obtained through the optimal coupling model, and the comprehensive heterogeneity coefficient of resource spatiotemporal environment is output. S5) Obtain expert subjective weights through the Delphi method, calculate information entropy objective weights and coefficient of variation objective weights based on the true value data of resource density, fuse them according to a preset ratio to obtain the basic weights of the optimization target, and correct the basic weights of the optimization target. S6) The entire survey sea area is divided into non-overlapping micro-regions based on geographical zoning, environmental zoning, and heterogeneity zoning, and the number of sampling stations in each micro-region is dynamically allocated; under the optimal sampling method for the entire area, the optimal number of survey stations is determined by the accuracy-cost elasticity coefficient. S7) Construct a comprehensive optimization objective function that minimizes the navigation distance, fuel cost, and operation time matching degree; use the particle swarm optimization (PSO) algorithm for global search to generate an initial solution group, and use the simulated annealing algorithm for local fine optimization to output the optimal navigation route that satisfies all constraints.

2. The method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives as described in claim 1, characterized in that: In step S2), the spatial variation coefficient of resources during the survey period Represented as: In the formula, , These represent the mean and standard deviation of resource density during the survey period, respectively. The time variation coefficient Represented as: In the formula, , These represent the mean and standard deviation of resource density for the survey time series, respectively.

3. The method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives as described in claim 2, characterized in that: In step S2), the preprocessed standardized resource-environment fusion dataset is selected, and paired datasets of resource density and single environmental factors are constructed respectively, denoted as... ,in, To standardize the resource density dataset, To standardize a single environmental factor, The number of valid stations for synchronization; Calculate separately The mean, sample covariance, and sample standard deviation are used to obtain the Pearson correlation coefficient. for: In the formula, >0 indicates a positive correlation. <0 indicates a negative correlation; express The sample covariance; for The sample standard deviation; , The first Standardized resource density values ​​and standardized environmental factor values ​​for each station; , They are respectively The mean.

4. The method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives as described in claim 3, characterized in that: In step S4), the optimal coupling model is selected based on the evaluation metrics, as follows: The reconstructed GLM, GAM, Kriging, and sdmTMB models were first screened using percentage bias PBIAS, and candidate models with percentage bias PBIAS in the range of [-25%, 25%] were retained. Then, the Spatiotemporal Environment Fit Factor (TEF) is used to perform a second screening of the candidate models selected in the first screening, and candidate models with a Spatiotemporal Environment Fit Factor (TEF) ≥ 0.55 are retained. Using the coefficient of determination The candidate models selected in the second round of screening were then screened a third time, retaining the coefficient of determination. Candidate models; The candidate models selected in the third screening were screened for the fourth time using the mean absolute error (MAE), and the candidate model with the smallest mean absolute error (MAE) was selected as the optimal coupling model. If only one candidate model remains after a certain screening, it is directly determined to be the optimal coupling model.

5. The method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives as described in claim 4, characterized in that: In step S6, the entire surveyed sea area is divided into non-overlapping micro-regions based on geographical zoning, environmental zoning, and heterogeneity zoning, and the number of sampling stations in each micro-region is dynamically allocated, as follows: S61) The entire survey sea area is divided according to geographical features into Each non-overlapping main area ; S62) Each main area is divided according to significant environmental factors. Divided into Sub-district ; S63) Extract each sub-region Resource spatiotemporal environment comprehensive heterogeneity coefficient of all grid nodes Calculate the average heterogeneity coefficient of the sub-region. And according to the preset heterogeneity classification threshold, the sub-regions are divided into... Divided into highly heterogeneous micro-regions Medium heterogeneous micro-regions Low heterogeneity micro-regions ;Right now: like This indicates a highly heterogeneous micro-region. ;like This is a micro-region of medium heterogeneity. ;like This indicates a low heterogeneity micro-region. ; Finally obtained Micro-area Number of grid nodes in each micro-region ; For each gradient of total number of stations Calculate each micro-region Site weight allocation Each gradient corresponds to a set of candidate total stations; that is: In the formula, For micro-regions The average heterogeneity coefficient; For micro-regions The accuracy requirement threshold; Therefore, the number of sampling stations in each micro-region for: ； In the formula, This indicates the rounding operation.

6. The method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives as described in claim 5, characterized in that: Step S6 involves determining the optimal number of survey sites, specifically including the following steps: S65), for each gradient of total number of stations K computer simulations were conducted using simple random sampling, systematic sampling, and dynamic stratified sampling respectively; and the relative estimation error REE and the absolute value of the relative deviation |RB| were calculated; as follows: The simple random sampling described above randomly selects nodes from the entire grid. Each site was sampled without replacement. The system sampling described above sorts the entire grid by latitude and longitude, randomly determines the starting point, and samples at fixed intervals. Each site; the dynamic stratified sampling in each micro-region Inside, from Randomly selected from each grid node Each site, after all micro-areas are sampled, is summarized as follows: One sampling site; For each optimization objective Based on the results of K resampling, calculate the relative estimation error REE and the absolute value of the relative deviation |RB|: ； In the formula, For the first Secondary sampling for the first Estimates of the optimization objective; For the first The optimization objective is the true value of the entire domain; For the first The relative estimation error REE of each optimization objective; In the formula, For the first The absolute value of the relative deviation of each optimization objective |RB|; S66) Combining the weights of the modified optimization objective Construct the comprehensive evaluation indices CREE and C|RB|, namely: For each gradient of total number of stations By comparing the CREE or C|RB| values ​​of the three sampling methods, the sampling scheme with the smallest value is selected as the optimal sampling under the corresponding gradient. The frequency of optimal sampling under all total station number gradients is counted, and the optimal sampling method with the highest frequency is selected as the optimal sampling method for the entire domain. S67) Calculate the elasticity coefficients of the gradient of the total number of stations. And filter the elasticity coefficients of all total station number gradients. The total number of stations gradient is ≥1, and the smallest total number of stations gradient is selected. As the optimal number of survey sites; In the formula, Gradient of total number of stations from Increase to The rate of change of accuracy over time; CREE was selected as the accuracy. ; Gradient of total number of stations from Increase to Cost change rate at time; Gradient of total number of stations The total cost of the survey.

7. The method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives as described in claim 1, characterized in that: In step S7), the specific details are as follows: S71) Assume the survey vessel departs from the dock, visits all stations in sequence, and finally returns to the dock; S72) Based on the shortest travel distance, lowest fuel cost, and highest operation time matching degree, an optimization objective function is established, namely: ； In the formula, This is the straight-line distance from the dock to the first station; This is the straight-line distance from the last station to the dock; This represents the straight-line distance between two adjacent stations. ; ; This is the sea state drag coefficient; For the survey vessel's fuel consumption rate; For the first Time window matching coefficient for each site; The optimal number of survey sites; , , There are three objective functions; S73), the three objective functions , , The results are integrated into the final comprehensive objective function. ,Right now: ； In the formula, , , The weights are the corresponding objective function weights. + + ; , The normalized objective function; S74) Use the Particle Swarm Optimization (PSO) algorithm to perform a global search to generate an initial solution group. When the maximum number of iterations is reached, output the global optimal solution of the population. The details are as follows: Each particle corresponds to a station. Randomly generated A valid access sequence forms the initial solution group; Calculate the comprehensive objective function value for each particle. ; Adjust the site order using the PSO speed-location update formula, i.e.: Speed ​​updated to: ; Location update: ; In the formula, Inertial weights; For the first The current particle in the next iteration The optimal position of an individual The optimal position for the entire population; A random number in the range [0,1]. , These are learning factors; Represents particles Current location; Indicates the current particle velocity; When the maximum number of iterations is reached, output the global optimum position of the population. As the globally optimal solution for the population; S75) Local fine-tuning is performed using the simulated annealing algorithm (SA), and the current optimal solution is output. As the optimal site access order.

8. The method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations based on multiple objectives as described in claim 1, characterized in that: In step S75), local fine-tuning is performed using the simulated annealing algorithm (SA), as follows: S751) The population global optimal solution output by the Particle Swarm Optimization (PSO) algorithm. As the initial current solution of SA Calculate its fitness value ; S752), for the initial current solution Randomly perturb the order of the stations to generate neighborhood solutions. ; S753), Calculate the neighborhood solution Sea state drag coefficient Channel detour coefficient Time window matching coefficient If all constraints are satisfied, then calculate its fitness value. Otherwise, discard the neighborhood solution and regenerate; the constraints are as follows: ； ； ， ； In the formula, , , These are ocean current speed, wind speed, and wave height, respectively. , , For the corresponding coefficients; Maximum sea state resistance; , These represent the actual distance traveled after bypassing the restricted area and the actual distance traveled while sailing in a straight line, respectively. This is the maximum detour coefficient for waterways; This refers to the actual speed. Rated speed; The maximum wave height that the ship can withstand; S754), if fitness value , solve the neighborhood As the new current solution; If fitness value Acceptable according to the Metropolis criterion, that is: ； like This represents the probability of receiving the signal. For temperature; in each iteration, the temperature... according to = attenuation, The attenuation coefficient; like If the new solution is found, then accept the new solution; otherwise, retain the original solution. S755), if continuous If no new solution is accepted in the next iteration, or the maximum number of iterations is reached, stop the SA iteration and output the current optimal solution. This refers to the optimal site access order.

9. A multi-objective nearshore fisheries resource monitoring station deployment and navigation route optimization system, characterized in that, include: The optimization objective determination module is used to select species richness, total resource density, and species resource density as optimization objectives. The quantification module is used to calculate the spatial and temporal variation coefficients of resources based on historical fishery resource data and environmental factor data. At the same time, it collects environmental factor data of the surveyed sea area and calculates the Pearson correlation coefficient between environmental factors and resource density to quantitatively characterize the spatiotemporal and environmental characteristics of resources. The model reconstruction module is used to select GLM, GAM, Kriging, and sdmTMB models as candidate models, use spatial variation coefficient and temporal variation coefficient as spatiotemporal feature terms, and use environmental factors with Pearson correlation coefficient greater than or equal to a preset threshold as environmental driving terms, and integrate them into the input layer of the candidate model to construct an environment-spatiotemporal coupled candidate model. The model selection module utilizes the mean absolute error (MAE) and coefficient of determination. Percentage Bias (PBIAS) and Spatiotemporal Environment Fit (TEF) are used as evaluation indicators to select the optimal coupling model; the true distribution data of resource density is obtained through the optimal coupling model, and the comprehensive heterogeneity coefficient of resource spatiotemporal environment is output. The optimization weight calculation module obtains expert subjective weights through the Delphi method, calculates objective weights of information entropy and coefficient of variation based on true resource density data, merges them according to a preset ratio to obtain the basic weights of the optimization target, and then corrects the basic weights of the optimization target. The site determination module divides the entire survey sea area into non-overlapping micro-regions based on geographical zoning, environmental zoning, and heterogeneity zoning, and dynamically allocates the number of sampling sites in each micro-region; under the optimal sampling method for the entire area, the optimal number of survey sites is determined by the accuracy-cost elasticity coefficient. The route optimization module is used to construct a comprehensive optimization objective function that minimizes the flight distance, fuel cost, and operation time matching. It uses the particle swarm optimization (PSO) algorithm for global search to generate an initial solution group, and then uses simulated annealing algorithm for local fine-tuning to output the optimal flight route that satisfies all constraints.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for optimizing the layout and navigation routes of nearshore fishery resource monitoring stations as described in any one of claims 1-8.

Images