Marine ranch comprehensive evaluation method based on genetic algorithm

Abstract

The invention provides a marine ranch comprehensive evaluation method based on a genetic algorithm. Comprising the following steps of (1) evaluation target setting, (2) data acquisition and preprocessing, (3) target weight setting, (4) initial comprehensive score calculation, (5) genetic algorithm optimization, (6) non-dominated sorting and congestion degree sorting, and (7) sorting and recommendation. According to the method, multiple target functions are subjected to genetic optimization, non-dominated sorting and congestion degree calculation are combined, and multiple key indexes of the marine ranch are systematically and comprehensively evaluated. Through standardization processing and a weighted scoring method, comparability among targets and scientificity of evaluation results are ensured.

Description

Technical Field

[0001] The invention belongs to the field of marine ecological technology, and in particular relates to a marine ranch comprehensive evaluation method based on genetic algorithm. Background Art

[0002] As the global demand for marine resources continues to grow, marine ranching, as a sustainable way of utilizing marine resources, has gradually become an important part of modern aquaculture. Marine ranching sets up artificial facilities in specific sea areas to create a suitable ecological environment, allowing a variety of aquatic organisms to thrive and achieve efficient utilization of resources. However, with the expansion of the scale of marine ranching and the diversification of species, how to conduct a scientific and comprehensive evaluation of it has become an important research topic in related fields.

[0003] At present, the comprehensive evaluation methods of marine ranches mostly adopt single-objective optimization or multi-objective evaluation methods based on experience, which mainly have the following defects: (1) Limitations of single-objective optimization. Most evaluation methods focus on optimizing a single objective, such as optimizing ecological benefits, maximizing economic benefits, or maximizing productivity. This method only focuses on a single evaluation objective and ignores the effect of multi-objective comprehensive evaluation, resulting in one-sided evaluation results that cannot fully reflect the actual comprehensive benefits of marine ranches.

[0004] (2) Insufficient multi-objective trade-offs. Although some methods attempt to introduce the concept of multi-objective optimization, in practical applications, the trade-offs between multiple objectives often rely on the subjective experience of decision makers and lack a systematic optimization mechanism. This approach easily leads to insufficient scientificity and objectivity in the evaluation results when dealing with multiple conflicting objectives, making it difficult to provide the best guidance for decision makers.

[0005] (3) Lack of consideration of target diversity. Existing evaluation methods often ignore the diversity of solutions when optimizing, which easily leads to the evaluation target falling into the local optimal solution and missing the better global solution. This is particularly obvious in complex multi-objective optimization problems, and may ultimately lead to the evaluation results failing to achieve the global optimal effect.

[0006] The name of the invention is Method, Device and System for Evaluating Marine Ranches, and the application number is CN201910845630.1. It is recorded that the invention involves the evaluation of marine ranches. At present, when evaluating the effect of marine ranches, the effect of marine ranches is generally evaluated based solely on the quantitative changes of a certain indicator, which will lead to incomplete and inaccurate evaluation results. The invention provides a method, device and system for evaluating marine ranches based on multiple indicators. The method includes conducting biological resource surveys before and after the construction of marine ranches; constructing a food web matrix based on the species composition; counting data of multiple indicators of each trophic level; referring to the comparison area, counting the relative change rate of each indicator before and after the construction of the marine ranch; mathematically analyzing the relative change rate of each indicator to screen out key factors; and evaluating marine ranches based on key factors.

[0007] The disadvantages of the methods, devices and systems used to evaluate marine ranches are: (1) Lack of direct support for multi-objective optimization. This evaluation method mainly relies on multiple biological indicators for evaluation, and the results only reflect the ecological impact of marine ranches. However, when faced with multiple conflicting objectives, such as productivity, economic benefits, and environmental impact, the patent lacks a clear multi-objective optimization mechanism. Existing methods are more about analyzing historical data and fail to provide active optimization solutions, which may limit their application in actual decision-making.

[0008] (2) Ignoring the diversity of solutions. The evaluation results obtained by existing methods are often a single optimal solution, that is, the optimal solution for fishery resources, while ignoring other suboptimal solutions or solutions with different advantages and disadvantages. This method is insufficient in terms of diversity protection and is prone to falling into the local optimal solution and missing the opportunity for the global optimal solution. Summary of the invention

[0009] The purpose of the present invention is to solve the defects of the above-mentioned prior art and provide a comprehensive evaluation method for marine ranches based on genetic algorithms, which can achieve an optimal balance among multiple objectives and generate a comprehensive score to help decision makers make scientific planning and management decisions.

[0010] The present invention simulates the natural selection and evolution process to conduct a scientific and comprehensive evaluation of marine ranches, thus making up for the shortcomings of traditional methods. By introducing genetic algorithms, it is possible to achieve a balanced optimization among multiple conflicting objectives, such as productivity, economic benefits, environmental impact and social benefits. Genetic algorithms effectively solve the trade-off problem between multiple objectives through mechanisms such as non-dominated sorting and crowding calculation, ensuring the comprehensiveness and scientificity of the evaluation results.

[0011] The present invention can dynamically adjust the structure according to actual needs and adapt to different environmental changes and market demands. Through the operations of selection, crossover and mutation, the present invention can continuously optimize the evaluation model, generate a more adaptable solution, and meet the requirements of modern marine ranch management for dynamic adjustment. The present invention effectively protects the diversity of the population during the optimization process through means such as crowding calculation and non-dominated sorting, and prevents the population from converging to the local optimal solution too early. This diversity protection mechanism enables the algorithm to explore a larger solution space, thereby increasing the probability of finding the global optimal solution and providing decision makers with a better guidance plan.

[0012] The present invention adopts the following technical solution: A comprehensive evaluation method for marine ranches based on genetic algorithms, comprising: Step 1. Set evaluation goals: Set multiple evaluation goals to fully reflect the comprehensive benefits of marine ranches. The evaluation goals include: (1) Productivity: the output of aquatic products per unit area, reflecting the production benefits of marine ranches, represented by Z1. (2) Economic benefits: the net income after deducting the cost from the sales revenue of aquatic products, reflecting the economic benefits of marine ranches, represented by Z2. (3) Environmental impact: including nitrogen and phosphorus emission indicators, reflecting the environmental friendliness of marine ranches, represented by Z3. (4) Social benefits: including social benefit indicators such as employment opportunities, represented by Z4.

[0013] Step 2. Data collection and preprocessing: (1) Data collection: Obtain relevant data on productivity, economic benefits, environmental impact, and social benefits of each area of ​​the marine ranch through research and review of relevant literature. (2) Data standardization: Standardize the data of different evaluation targets, namely productivity, economic benefits, environmental impact, and social benefits, eliminate dimensional differences, and ensure comparability between different evaluation targets. Among them, the standardization formula is:

[0014] in, is the standardized value of the i-th target, where i ranges from 1 to 4. is the original value of the i-th target, is the minimum value of the i-th target, Step 3. Set target weights: Decision makers set the weights of evaluation targets. When the evaluation targets are productivity, economic benefits, environmental impact, and social benefits, the weights of productivity, economic benefits, environmental impact, and social benefits are defined as: .

[0015] Step 4. Calculate the initial comprehensive score: When the evaluation objectives are productivity, economic benefits, environmental impact and social benefits, a weighted score is calculated for each pasture, and the comprehensive score formula is: in, For the The overall score of the marine ranch is is the standardized value of the first target in the jth marine ranch.

[0016] Step 5. Genetic algorithm optimization: S501. (1) Initialize the population: Generate the initial population, where each individual represents a marine ranch. (2) Select the parents: According to the comprehensive score in step 4, select the two individuals with the highest scores as the parents to generate the next generation. (3) Crossover: Generate new offspring individuals by combining the characteristics of two parent individuals. The crossover operation simulates the gene exchange process in biological genetics. Common methods include single-point crossover, multi-point crossover and uniform crossover. During the crossover process, part of the characteristics of the parent individuals are exchanged to produce new offspring individuals. If the genes of parent individuals A and B are represented as [x1, x2, x3] and [y1, y2, y3], two new offspring [x1, y2, x3] and [y1, x2, y3] may be generated through crossover.

[0017] S502. Mutation: Introduce random changes to increase the diversity of the population and prevent the algorithm from falling into a local optimal solution. Mutation simulates the mutation phenomenon in biological evolution. Usually, small random changes are introduced to certain gene sites of an individual. For example, change a gene value from xi to xi'. If the gene of an individual is represented as [x1, x2, x3], during the mutation process, the value of x2 may be randomly changed to generate a new individual [x1, x2', x3].

[0018] S503. (1) Generate a new population: The initial population generates a new set of individuals by selecting parents, crossover and mutation operations. These individuals constitute the new generation of the population. (2) Merge populations: Merge the generated new population with the initial population. The merging process ensures that all possible solutions are considered and provides more solutions for the next step of sorting.

[0019] Step 6. Sort the individuals in the population merged in step 5 by non-dominated sorting and crowding degree: S601. Non-dominated sorting: The purpose of non-dominated sorting is to sort the individuals in the population according to multiple evaluation objectives, that is, to sort the configuration schemes of each marine ranch, and to identify which individuals are Pareto optimal solutions. A Pareto optimal solution refers to a solution that is not dominated by other individuals in all objectives. Among them, the dominance relationship is, for example: combination A dominates combination B if and only if the following conditions are met: Ⅰ. A is not worse than B in all objectives, that is, each objective value of A is better than or equal to B. Ⅱ. A is better than B in at least one objective, that is, one of the objective values ​​of A is better than B.

[0020] Non-dominated sorting includes: (1) Non-dominated sorting initialization: For each individual x in the merged population, calculate the number of times n it is dominated by other individuals x , and the set of other individuals it dominates S x ; (2) Identification of the Pareto front F1: All individuals whose number of dominations is zero constitute the first Pareto front F1; (3) Identification of subsequent Pareto fronts: For individual y in the first Pareto front F1, traverse the set S dominated by individual y y , for each individual z dominated by individual y, the number of times it is dominated is n z Decrease by 1; if n z becomes zero, and individual z is placed in the next Pareto frontier F2. If n z If it is not zero, it means that individual z is still dominated by other individuals, so it cannot be placed in the next Pareto frontier. In this case, individual z remains in its current position and continues to wait for other individuals that dominate it to be classified into the frontier, and then rechecks its dominance status and repeats the above process until all individuals are assigned to the corresponding Pareto frontier. Eventually, each individual will be assigned to a certain Pareto frontier, such as F2, F3, etc.

[0021] S602. Crowding ranking: Among the individuals in the same Pareto front, the crowding of each individual is further calculated. The smaller the crowding, the closer the individual is to other individuals in the target space. The larger the crowding, the farther the individual is from other individuals in the target space, and thus the individual is retained first.

[0022] Crowding sorting includes: (1) Initializing the crowding: For each individual in the Pareto front, set its initial crowding to zero. (2) Target value sorting: For each target, sort the individuals in the Pareto front from small to large according to the target value. For each target sorting, mark the crowding of the minimum and maximum individuals as infinite to protect the boundary individuals. (3) Calculating the crowding of internal individuals: For each non-boundary individual i, calculate its crowding. Crowding d i Calculated by the following formula:

[0023] in, is the target value of the previous individual of individual i on target k, is the target value of the next individual after individual i on target k. is the maximum value of target k, is the minimum value of target k.

[0024] Step 7. Sort the scores and provide optimization suggestions: According to the two rankings of Pareto optimal solution and crowding target value in step 6, the configuration schemes of each marine ranch are ranked, the scheme with the highest score is recommended, and detailed optimization suggestions are provided for the configuration schemes of other marine ranches.

[0025] Beneficial effects of the present invention: (1) Traditional marine ranch evaluation methods often focus on a single objective or simple multi-objective analysis, and cannot effectively deal with conflicts between multiple objectives. The present invention can find a balance between multiple objectives through multi-objective optimization in genetic algorithms, making the evaluation results not only more comprehensive, but also ensuring the coordinated optimization of each objective. In this way, decision makers can formulate more scientific management and development strategies while meeting the needs of multiple parties.

[0026] (2) The environment and conditions of marine ranches often change dynamically. Traditional methods lack flexibility and are difficult to cope with such changes. The present invention uses the adaptive mechanism of genetic algorithms to dynamically adjust the population structure and timely generate solutions that adapt to new environmental conditions. This flexibility ensures the real-time and accuracy of the evaluation and optimization process, allowing the management of marine ranches to keep pace with the times and better adapt to the complex and changing natural environment and market demand.

[0027] (3) Traditional optimization methods often fall into local optimal solutions when solving multi-objective problems, thus missing out on better global solutions. The present invention introduces crowding calculation into the genetic algorithm, which not only prevents the population from converging prematurely, but also increases the chance of finding the global optimal solution by protecting the diversity of solutions. This opportunity enables the comprehensive evaluation of marine ranches to better reflect the actual situation and provide managers with more decision-making options, thereby effectively achieving sustainable utilization of resources.

[0028] (4) Traditional evaluation methods that rely on experience are easily affected by human factors, which may lead to subjectivity and inaccuracy in the evaluation results. The present invention uses an intelligent genetic algorithm to automatically perform multi-objective optimization, reduce human intervention, and improve the objectivity and scientificity of the evaluation. Decision makers can make more scientific marine ranch management decisions based on more accurate and comprehensive evaluation results, thereby effectively improving the comprehensive benefits of marine ranches.

[0029] (5) Due to the versatility and flexibility of the genetic algorithm itself, the present invention is not only applicable to the current comprehensive evaluation of marine ranches, but can also be adjusted and expanded according to different application scenarios. This feature of easy implementation and promotion allows the present invention to be applied in a wider range, providing scientific evaluation and optimization tools for more marine ranches, and promoting the sustainable development of marine resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a flow chart of the steps of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Example like Figure 1 As shown in step 1, it is assumed that the original data of five marine ranches are collected, and productivity, economic benefits, environmental impact and social benefits are set as evaluation targets, as shown in Table 1 below: Table 1 Original data of five marine ranches Step 2. Data preprocessing: In order to eliminate the dimensional differences between different targets, the data is standardized. The standardized data is shown in Table 2: Table 2 Standardized data of five marine ranches Step 3. Set target weights: The target weights set by decision makers are 0.3 for productivity, 0.3 for economic benefit, 0.2 for environmental impact, and 0.2 for social benefit.

[0033] Step 4. Calculate the initial comprehensive score: The comprehensive score of each ranch is calculated, and the comprehensive score of marine ranch A is: The overall score of B Ocean Ranch is: . C Ocean Ranch's overall rating: D Ocean Ranch's overall rating: . E Ocean Ranch's overall rating: =0.4984.

[0034] Step 5. Genetic algorithm optimization: Select parents: Select the two combinations with the highest fitness as parents according to the comprehensive score to generate the next generation of individuals. Select the two combinations with the highest comprehensive scores, D and B, as parents. Crossover: Generate new individuals from the selected parents through crossover operations. Use single-point crossover: The genes of parent D are [1.00, 1.00, 0.0, 0.00]. The genes of parent B are [0.625, 0.583, 0.4, 0.33]. Choose to perform single-point crossover after the third target, namely the environmental influence: New individual 1: Inherits the first two target genes of D and the last two target genes of B, and obtains [1.00, 1.00, 0.4, 0.33]. New individual 2: Inherits the first two target genes of B and the last two target genes of D, and obtains [0.625, 0.583, 0.00, 0.00].

[0035] Mutation: Introduce random mutations in the generated new individuals to increase the diversity of the population. Suppose the following mutations are performed: mutate new individual 1 [1.00, 1.00, 0.4, 0.33] to change the productivity Z1 to 0.95. New individual 1 becomes [0.95, 1.00, 0.4, 0.33]. Mutate new individual 2 [0.625, 0.583, 0.0, 0.00] to change the economic benefit Z2 to 0.60. New individual 2 becomes [0.625, 0.60, 0.0, 0.00].

[0036] Step 6. Sort the individuals in the population by non-dominated sorting and crowding degree: (1). Non-dominated sorting: The initial population and the generated new population are shown in Table 3 below: Table 3 Initialized population and newly generated population Pareto frontier F1: Combination C is not dominated by other combinations, so it constitutes the optimal first Pareto frontier F1. Pareto frontier F2: Combination B, combination E and new individual 1 are dominated by combination C, but not dominated by other combinations, so they constitute the second Pareto frontier F2. Pareto frontier F3: Combination A, combination D and new individual 2 are dominated by the combination in F1 and F2, so they constitute the third Pareto frontier F3.

[0037] (2) Crowding calculation: The purpose of crowding calculation is to evaluate the distribution of individuals within the same Pareto front to ensure the diversity of the population. Take the Pareto front F2 (including combination B, combination E and new individual 1) as an example for calculation: Productivity Z1 sorting: Sorting results: E (0.375), B (0.625), new individual 1 (0.95). The crowding of boundary individuals E and new individual 1 is set to infinity. Crowding calculation of intermediate individual B: = (0.95-0.375) / (1.00-0.00) = 0.575. Economic benefit Z2 sorting: sorting results E (0.333), B (0.583), new individual 1 (1.00) The crowding degree of boundary individual E and new individual 1 is set to infinity.

[0038] Calculation of the crowding degree of the intermediate individual B: (1.00-0.333) / (1.00-0.00)=0.667 Environmental impact Z3 sorting: sorting results B (0.4), E (0.6), new individual 1 (0.4) Since the environmental impact of new individual 1 is the same as that of the middle individual B, their crowding increases. The crowding of boundary individual B and new individual 1 is set to infinity. Social benefit Z4 sorting: sorting results: new individual 1 (0.33), B (0.33), E (0.83), the crowding of new individual 1 and boundary individual E is set to infinity.

[0039] The total crowding degree of the middle individual B is the sum of the crowding degrees of each dimension: dB = 0.575 + 0.667 + ∞ + ∞ = ∞. Since the distance between the middle individual B and other individuals in the two dimensions of environmental impact and social benefit is ∞, it means that the middle individual B is at the boundary in these dimensions, so its total crowding degree is ∞. In the selection process of the genetic algorithm, such individuals are usually retained first to ensure the diversity of the population.

[0040] Step 7. Sort the scores and provide optimization suggestions: Finally, according to the Pareto optimal solution and the target value of crowding, the combination with the best performance and diversity is selected: The first Pareto frontier F1: contains combination C, which has been identified as not dominated by any combination and is therefore optimal. The second Pareto frontier F2: The crowding of combination B and new individual 1 is infinite, but in terms of productivity and economic benefits, the score of combination B is closer to other solutions, so combination B and combination E are retained as diversity representatives. The third Pareto frontier. F3: Select new individual 2 and combination A because they have higher crowding and represent different solutions.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A comprehensive evaluation method for marine ranches based on genetic algorithm, characterized in that: include: Step 1. Set evaluation targets: Set four evaluation targets, including productivity Z1, economic benefit Z2, environmental impact Z3 and social benefit Z4, to fully reflect the comprehensive benefits of marine ranching; Step 2. Data collection and preprocessing: Through research and literature review, obtain the evaluation targets set in step 1 for each area of ​​the marine ranch; standardize the data of different evaluation targets to eliminate dimensional differences. The standardization formula is: in, is the standardized value of the i-th target, i ranges from 1 to 4, is the original value of the i-th target, is the minimum value of the i-th target, ; Step 3. Set target weights: Decision makers set the weights of evaluation targets. Decision makers set the weights of productivity, economic benefits, environmental impact and social benefits respectively. ; Step 4. Calculate the initial comprehensive score: Calculate the weighted score for each ranch. The comprehensive score formula is: in, For the The overall score of the marine ranch is is the first target standardized value in the jth marine ranch; Step 5. Genetic algorithm optimization: (1). Initialize the population: Generate the initial population, where each individual represents a marine ranch; (2). Select parents: According to the comprehensive score in step 4, select the two individuals with the highest scores as parents to generate the next generation; (3). Crossover: Generate new offspring individuals by combining the characteristics of two parent individuals; (4). Mutation: Introduce random changes to increase the diversity of the population; (5). Generate a new population: The initialized population is constructed by selecting parents, crossover and mutation operations to form a new generation of population; (6). Merge populations: Merge the generated new population with the initialized population; Step 6. Sort the individuals in the population merged in step 5 by non-dominated sorting and crowding degree: S601. Non-dominated sorting: (1) Non-dominated sorting initialization: For each individual x in the merged population, calculate the number of times n it is dominated by other individuals x , and the set of other individuals it dominates S x ; (2) Identification of the Pareto front F1: All individuals whose number of dominations is zero constitute the first Pareto front F1; (3) Identification of subsequent Pareto fronts: For individual y in the first Pareto front F1, traverse the set S dominated by individual y y , for each individual z dominated by individual y, the number of times it is dominated is n z Decrease by 1 if n z becomes zero, and individual z is placed in the next Pareto frontier F2; S602. Crowding sorting: (1). Initializing the crowding: For each individual in the Pareto front, set its initial crowding to zero; (2). Target value sorting: First, for each target, sort the individuals in the Pareto front from small to large according to the target value. For each target sorting, mark the crowding of the minimum and maximum individuals as infinite; (3). Calculating the crowding of internal individuals: For each non-boundary individual i, calculate its crowding, crowding d i Calculated by the following formula: in: is the target value of the previous individual of individual i on target k, is the target value of the next individual after individual i on target k, is the maximum value of target k, is the minimum value of target k; Step 7. Sort the scores and provide optimization suggestions: The configuration plans of each marine ranch are ranked according to the two ranking methods of Pareto optimality and crowding target value in step 6, and the plan with the highest score is recommended, and detailed optimization suggestions are provided for the configuration plans of other marine ranches.

2. The method according to claim 1, characterized in that Productivity: the output of aquatic products per unit area, reflecting the production efficiency of the marine ranch, represented by Z1; Economic benefits: the net income after deducting the cost from the sales revenue of aquatic products, reflecting the economic benefits of the marine ranch, represented by Z2; Environmental impact: including nitrogen and phosphorus emission indicators, reflecting the environmental friendliness of the marine ranch, represented by Z3; Social benefits: including social benefit indicators of employment opportunities, represented by Z4.

3. The method according to claim 1, characterized in that The crossover methods in step 5 include single-point crossover, multi-point crossover and uniform crossover.

4. The method according to claim 1, characterized in that: S601 Domination relationship: Combination A dominates combination B if and only if the following conditions are met: (1) A is not worse than B in all objectives; (2) A is better than B in at least one objective.

5. The method according to claim 1, characterized in that S601 further includes: if n z is not zero, individual z is still dominated by other individuals, and individual z remains in its current position, waiting for other individuals that dominate it to be classified to the frontier, and then rechecks its dominance status, repeating the above process until all individuals are assigned to the corresponding Pareto frontier, and eventually each individual will be assigned to a certain Pareto frontier.

Images