Sow nutritional feed formula screening method based on optimization algorithm

Abstract

The invention belongs to the technical field of feed formula research, and particularly relates to a sow nutritional feed formula screening method based on an optimization algorithm. Population initialization distribution is analyzed, and a Bernoulli mapping sequence is introduced to initialize the population, so that population distribution is more uniform, and the initial search capability of the algorithm is improved; an improved self-adaptive nonlinear convergence factor is introduced, and the global search capability and the local development capability of the algorithm are balanced and enhanced; a neighborhood disturbance optimization mechanism is added, and a new neighborhood solution is generated in an iteration process, so that the algorithm is prevented from easily falling into a local optimal solution in iteration, especially in a later stage of iteration. Finally, the effectiveness of the BANChOA algorithm is further verified by comparing optimization of a sow feed formula before and after ChOA improvement. The analysis of the improved strategy of the invention shows that the improved algorithm needs more parameters, the optimization performance of the algorithm is greatly influenced by neighborhood disturbance optimization, and there is still an improvement space for population initialization and convergence factors.

Description

Technical Field

[0001] The invention belongs to the technical field of feed formula research, and more specifically, relates to a method for screening sow nutritional feed formula based on an optimization algorithm. Background Art

[0002] In pig farming, most large-scale farms adopt a self-breeding and self-feeding model. From the perspective of breeding costs, feed costs account for the highest proportion of pig breeding costs. Accurate feed formula not only meets the normal reproduction of sows and the future development of piglets, but also reduces the breeding costs of sows. Traditional feed formulation methods are difficult to handle complex multi-raw material optimization problems, which easily leads to increased feed costs. Therefore, the use of intelligent optimization algorithms to conduct accurate and reasonable sow feed formulation research is of great significance to today's large-scale self-breeding and self-feeding breeding model.

[0003] Scholars at home and abroad have widely applied intelligent optimization algorithms to feed formula optimization. Foreign scholars TOAliu et al. established a linear programming model to calculate the cost of the minimum feed formula while meeting nutritional needs, and optimized the feed for pigs at different stages such as pregnant sows and lactating sows. The experimental results showed that through the optimization of the linear programming model, the cost of pig feed was significantly reduced while ensuring nutritional quality. Dipti Singh et al. first combined linear programming technology (LP) with the self-organizing migration genetic algorithm (SOMGA) for feed formula optimization. This combination utilizes the fast preliminary solution capability of linear programming and the fine optimization capability of SOMGA. Compared with LP, SOMGA not only has significant improvements in computational efficiency and solution quality, but also finds lower-cost feed formulas. Adem AlpaslanAltun and Mehmet Akif The two proposed to use the particle swarm optimization (PSO) algorithm to solve the cost optimization problem of mixed feed, and compared the results of this method with those of linear programming and real-coded genetic algorithms, and found that the optimization effect of the PSO algorithm was better. Domestic scholars Li Hao and others used the particle swarm algorithm to optimize the pig feed formula, overcoming the difficulty of traditional methods in dealing with complex multi-constraint or multi-objective problems. The simulation results showed that the particle swarm algorithm was superior to the traditional LP method in solving the pig feed formula problem. Zhang Xuecong and others used the improved taboo search algorithm (ITS) to optimize the pig feed formula cost. By introducing new neighborhood movement rules and shortening the taboo table to optimize the traditional TS algorithm, compared with traditional algorithms such as LP and genetic algorithm (GA), it was proved that ITS is better at solving livestock feed formula problems. Yang Shasha and others used the improved genetic algorithm to optimize the pig feed formula, using real number coding, random generation of the initial population, and using the improved proportional selection strategy and elite preservation strategy to select excellent individuals to enter the next generation. The partial matching crossover method was used for adaptive crossover operations, and Gaussian mutation was used for adaptive mutation operations, which reduced the cost of pig breeding. Liu Qing and others applied the artificial fish swarm algorithm (AFSA) to feed formula optimization for the first time. In order to meet the convergence accuracy requirements of feed formula optimization, they adopted an artificial fish swarm algorithm operating framework based on a symbiotic system, which significantly improved the convergence accuracy and speed of the original algorithm. The calculation results were compared with some existing intelligent algorithms, and the results showed that they were superior to other existing intelligent algorithms.

[0004] Although there are some papers that use intelligent algorithms to optimize sow feed formula, there are relatively few studies that apply new algorithms to this field. As a new intelligent algorithm, ChOA has two main features: first, it divides the population into independent individuals, which can effectively improve the algorithm's exploration ability; second, it introduces chaos factors, which helps to improve the convergence speed and accuracy of the development process. Summary of the invention

[0005] The present invention improves ChOA and proposes a method for screening sow nutritional feed formula based on optimization algorithm. In view of the shortcomings of ChOA, the population initialization of the chimpanzee optimization algorithm is improved, the Bernoulli mapping sequence is used to replace the random initialization, and an adaptive nonlinear convergence factor is introduced to better adapt to the actual situation of algorithm search. In addition, a neighborhood perturbation mechanism based on greedy selection is added. Through experimental comparison, the feasibility and superiority of the improved chimpanzee optimization algorithm in the cost optimization of multi-constrained and multi-raw material combinations of sow feed formula are verified.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for screening a sow nutritional feed formula based on an optimization algorithm, wherein m is the number of formula raw materials, and n is the number of nutritional components. The steps of the method for screening a sow nutritional feed formula are as follows:

[0008] Step 1: Initialize algorithm-related parameters, the dimension of the raw solution, the maximum number of iterations, the upper and lower bounds of the search space, and the objective function;

[0009] Step 2: Define the fitness function to evaluate the pros and cons of each ratio scheme. The fitness function has two evaluation indicators: total cost and whether the nutritional components of each raw material are within the constraint range;

[0010] Step 3: Use the Bernoulli mapping sequence to generate a population, where each population individual is a set of m raw material formula contents;

[0011] Step 4: Enter the main loop, calculate f and CR dynamic coefficients, calculate the fitness of each search agent position, and evaluate the quality of the recipe based on its nutritional content and cost;

[0012] The parameter f is the convergence factor. In the improved algorithm, the convergence factor of linear decay is changed to the convergence factor of nonlinear decay to balance the global search and local search capabilities of the algorithm. The neighborhood perturbation strategy is added to the algorithm improvement, and CR is a parameter used to determine whether to perform neighborhood perturbation. The later the algorithm enters, the greater the probability of neighborhood perturbation as CR increases.

[0013] Step 5: Update the position of the search agent according to the positions of the attacker, obstacle, chaser, and evictor to explore a better feed formula;

[0014] Step 6: Determine whether neighborhood operations can be performed. If so, use neighborhood operations to further optimize the position of each search agent. Otherwise, go directly to Step 7.

[0015] Step7: If the maximum number of iterations is reached, record the current optimal solution and output the result, otherwise return to Step4.

[0016] The technical solution is further optimized, and the objective function minF(X) model is as follows:

[0017]

[0018] In the formula, c i —The price of each raw material, x i —The proportion of different raw materials in 1kg feed formula.

[0019] The technical solution is further optimized, and the upper and lower limits of the amount of raw materials used in the sow feed and the constraints on the total nutritional content and raw material ratio are as follows:

[0020]

[0021] x 1 +x 2 +…+x m =1 (18)

[0022] Where: a ij —The percentage of the jth nutrient contained in the i-th raw material; d i —The minimum standards of various nutritional indicators that should be met in the formula; m—the number of raw materials in the formula; n—the number of nutritional ingredients.

[0023] The technical solution is further optimized. The fitness function is used to evaluate the pros and cons of each ratio scheme. The fitness function has two evaluation indicators: total cost and whether the nutritional components of each raw material are within the constraint range. The fitness function is as follows:

[0024] f(x)=minF(x)+pen(x) (19)

[0025] Where minF(x) is the objective function of formula (16); pen(x) is the penalty term used to penalize solutions that do not meet the constraints. The expression is as follows:

[0026] pen(x)=p 1 (x)+p 2 (x)+p 3 (x) (20)

[0027]

[0028] Where: p 1 (x) is the penalty for the ingredient ratio constraint, which is used when the raw material does not meet the upper and lower limits, p 2 (x) is the penalty for the total proportion constraint, which is used to not satisfy equation (18), p 3 (x) is the penalty for the nutritional requirement constraint, which is used for not satisfying equation (19); where in equation (21), li and ui are the lower and upper limits of the i-th raw material, respectively.

[0029] Different from the prior art, the above technical solution has the following beneficial effects:

[0030] By analyzing the population initialization distribution, the Bernoulli mapping sequence is introduced to initialize the population, so that the population distribution is more uniform, and the search ability of the algorithm in the early stage is improved; the improved adaptive nonlinear convergence factor is introduced to balance and enhance the global search ability and local development ability of the algorithm; the neighborhood perturbation optimization mechanism is added to generate new neighborhood solutions during the iteration process to prevent the algorithm from easily falling into the situation of local optimal solutions in the iteration, especially in the later stage of iteration. The superiority of the improved algorithm is verified by performing optimization tests on two basic test functions. Finally, by comparing the optimization of sow feed formula before and after the improvement of ChOA, the effectiveness of the BANChOA algorithm (which is an improvement of the basic chimpanzee optimization algorithm by introducing the Bernoulli mapping sequence to initialize the population; introducing an improved adaptive nonlinear convergence factor; adding the neighborhood perturbation optimization mechanism, named BANChOA) is further verified. Through the analysis of the improvement strategy of the present invention, it can be seen that the improved algorithm requires a large number of parameters and the optimization performance of the algorithm is greatly affected by the neighborhood perturbation optimization. There is still room for improvement for population initialization and convergence factor. In future work, we will continue to study improved optimization strategies to enhance the operability and applicability of ChOA and apply it to more complex optimization problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is the flow chart of BANChOA algorithm;

[0032] Figure 2 is a unimodal function F 1 Comparison chart;

[0033] Figure 3 is the nonlinear multimodal function F 2 Comparison chart;

[0034] Figure 4 This is a comparison chart of the price trend of ChOA and BANChOA for feed formula in the early pregnancy period (<90 days);

[0035] Figure 5 This is a comparison chart of the price trend of ChOA and BANChOA for feed formula in late pregnancy (>90 days);

[0036] Figure 6 This is a comparison chart of the price trends of ChOA and BANChOA in lactation feed formula. DETAILED DESCRIPTION

[0037] In order to explain the technical content, structural features, achieved objectives and effects of the technical solution in detail, the following is a detailed description in conjunction with specific embodiments and accompanying drawings.

[0038] The sow feed data and nutritional content required for the experiment of the present invention are mainly derived from the "China Feed Composition and Nutritional Value Table" (2019, 30th edition) and "Nutritional Needs of Pigs". The raw materials are proportioned according to the ratio of each kilogram of feed raw materials and the nutrients contained. The raw materials contain 11 raw materials such as corn, bran, fish meal, salt to supplement sodium, and additives containing mineral elements and vitamins, as well as 16 nutrients including digestible energy. Because sows have different stages in the breeding process, they are divided into pregnancy and lactation, and the pregnancy period is divided into the early pregnancy (<90 days) and the late pregnancy (>90 days), so different stages have corresponding raw material usage ranges and different nutritional requirements. The specific feed raw materials and nutritional composition data are shown in Table 1-1 below, and the raw material prices are shown in Table 1-2:

[0039] Table 1-1 Feed raw material data and nutritional ratio

[0040]

[0041]

[0042] Table 1-2 Raw material prices

[0043]

[0044] The dosage range of feed raw material formula required by sows at different stages is also different. The specific dosage range is shown in Table 1-3:

[0045] Table 1-3 Range of feed raw material usage for sows at different stages

[0046]

[0047] Table salt supplements the sodium content of sows, and additives supplement the vitamins and trace elements required by sows, and the usage amount is fixed.

[0048] The nutritional requirements of sows vary at different stages. The protein demand is moderate in the early pregnancy to support protein synthesis in the sow and early embryonic development; the protein demand increases in the late pregnancy to support the formation of fetal muscle and tissue, and calcium and phosphorus are needed to support fetal bone development and sow delivery; the protein demand is the highest during lactation, which is used for protein synthesis in milk, and a large amount of calcium and phosphorus are needed to support milk production and sow health. The specific data are shown in Table 1-4:

[0049] Table 1-4 Minimum nutritional requirements for sows at different stages

[0050]

[0051] The proposed multi-strategy neighborhood perturbation chimpanzee algorithm (BANChOA) introduces the above three improved strategies based on ChOA. The input of BANChOA algorithm is the Bernoulli mapping sequence to initialize the population and related parameters, and the output is the position of the optimal chimpanzee individual. The specific implementation steps and processes are as follows: Figure 1 shown.

[0052] Step 1: Initialize the population size, the maximum number of iterations Max_ter, the search range lb, ub, and set the corresponding parameters, where lb is the lower bound of the search range and ub is the upper bound of the search range.

[0053] Step 2: Initialize the population X using the Bernoulli mapping sequence i , i=1,2,...,N.

[0054] Step 3: Calculate the fitness value of each individual, find the top four optimal values ​​and record them as four chimpanzee individuals, denoted as X Attacker , X Barrier , X Chaser and X Driver .

[0055] Step 4: Update the coefficients a, c and the value of the convergence factor f according to the formula.

[0056] Step 5: Use the probability factor CR to determine whether to further update the locations of the four types of chimpanzees and the locations of other chimpanzees.

[0057] Step 6: Determine whether the target satisfaction value is reached. If so, output the optimal value, otherwise go to Step 7.

[0058] Step 7: Determine whether the maximum number of iterations has been reached. If not, jump to Step 3. Otherwise, end the algorithm and output the position of the optimal chimpanzee.

[0059] Experimental environment and parameter settings

[0060] The hardware environment of the simulation experiment of the present invention is based on Intel (R) Core (TM) i5-4590 CPU, 3.30GHz main frequency, 16.0GB memory and Windows 10 (64-bit) operating system to simulate the proposed algorithm. The programming software is Matlab R2022a. The benchmark test function is selected to perform performance testing on the improved algorithm. Table 2-1 is the single peak function F 1 , nonlinear multimodal function F 2 Table 2-2 shows the parameter settings for each algorithm.

[0061] Table 2-1 Benchmark functions

[0062]

[0063]

[0064] Table 2-2 Algorithm parameter settings

[0065]

[0066] Analysis and evaluation of algorithm model simulation results

[0067] Comparison results of Bernoulli mapping sequence initialization population

[0068] The ChOA using the Bernoulli mapping sequence is called BChOA (Chimp Optimization Algorithm for Bernoulli). Bernoulli (mapping sequence base amount is λ = 0.7, initial z k is a random value between (0,1), the population size N = 50, the maximum number of iterations Max_ter = 500, ChOA and BChOA are in the unimodal function F 1 , nonlinear multimodal function F 2 The optimization comparison is shown in Table 2-3.

[0069] Table 2-3 Comparison of optimization results between ChOA and BChOA

[0070]

[0071] From Table 2-3, we can see that for the single-peak function F 1 , nonlinear multimodal function F 2 In the optimization test, BChOA's ability to find the optimal value is better than ChOA's. From the standard deviation, we can see that BChOA has a smaller standard deviation in the two test functions, so BChOA has better stability and robustness. In summary, the improvement of the Bernoulli mapping sequence initialization population has an impact on the performance of the chimpanzee optimization algorithm.

[0072] Comparison results of adding adaptive nonlinear convergence factor

[0073] BChOA with an adaptive nonlinear convergence factor f is called BAChOA (Chimp Optimization Algorithm for Bernoulli and Adaptive nonlinear convergence factor). 1 , nonlinear multimodal function F2 The optimization comparison is shown in Table 2-4.

[0074] Table 2-4 Comparison of optimization results between BChOA and BAChOA

[0075]

[0076] From Table 2-4, we can see that for the single-peak function F 1 , nonlinear multimodal function F 2 In the optimization test, BAChOA's ability to find the optimal value is better than BChOA's. From the comparison of standard deviations, we can see that BAChOA is lower, so BAChOA has better stability and robustness. In summary, adding an adaptive nonlinear convergence factor f has an impact on the performance of the Chimpanzee optimization algorithm.

[0077] Added comparison results of neighborhood perturbation optimization

[0078] The BAChOA with the added neighborhood perturbation function is called BANChOA. 1 , nonlinear multimodal function F 2 The optimization comparison is shown in Table 2-5.

[0079] Table 2-5 Comparison of optimization results between BAChOA and BANChOA

[0080]

[0081] From Table 2-5, we can see that for the single-peak function F 1 , nonlinear multimodal function F 2 In the optimization test, BANChOA's ability to find the optimal value is better than BAChOA's, and BANChOA is better at finding the optimal value for the test function F 2 The theoretical value is reached during optimization. BANChOA has a lower standard deviation in the optimization test of the two functions, so BANChOA has better stability and robustness. In summary, the functional improvement of adding neighborhood perturbation optimization has an impact on the performance of the Chimpanzee optimization algorithm.

[0082] Comprehensive comparison results of different improvement strategies

[0083] Combined with the above ablation experiment comparison, ChOA, BChOA, BAChOA, BANChOA are used to compare the single peak function F 1 , nonlinear multimodal function F 2 The average convergence curves of Figure 2 and Figure 3As shown in the figure, it is concluded that each addition of an improvement strategy improves the algorithm optimization. Compared with ChOA, the average convergence performance of the final improved BANChOA is significantly stronger, achieving the expected effect.

[0084] Comparative experiments with other algorithms

[0085] In order to further observe the effectiveness of the improvements to the Chimpanzee optimization algorithm, BANChOA and ChOA are compared with the classic particle swarm optimization algorithm (PSO) and the Marine Predators Algorithm (MPA) proposed by Afshin Faramarzi et al. in 2020. The same two benchmark functions are used, the unimodal function F 1 , nonlinear multimodal function F 2 , assuming the population size is 50 and the maximum number of iterations is 500. The comparison is shown in Table 2-6.

[0086] Table 2-6 Comparison of results of different optimization algorithms

[0087]

[0088]

[0089] It can be seen from the above table that BANChOA is better than other intelligent optimization algorithms in solving the optimal value, and its standard deviation is also lower and more stable than other algorithms, which further shows that the improved chimpanzee optimization algorithm BANChOA has stronger optimization performance and robustness.

[0090] Optimization of sow feed formula based on BANChOA

[0091] Constructing fitness function

[0092] BANChOA is used to optimize the cost of sow feed formula. The optimization problem of sow feed formula is a multi-objective and multi-constrained combinatorial optimization problem.

[0093] The experimental design takes the lowest economic cost as the objective function, and the upper and lower limits of the raw material usage at different stages and the nutritional indicators of sows at different weights and stages as constraints. The objective function minF(X) model is as follows:

[0094]

[0095] In the formula, c i —The price of each raw material, x i —The proportion of different raw materials in 1kg feed formula.

[0096] The upper and lower limits of raw material usage for sow feed are shown in Table 1-3, and the constraints on the total nutritional content and raw material ratio are as follows:

[0097]

[0098] x 1 +x 2 +…+x m =1 (18)

[0099] Where: a ij —The percentage of the jth nutrient contained in the i-th raw material; d i —The minimum standard of various nutritional indicators that should be met in the formula; m—the number of raw materials in the formula; n—the number of nutrients. As shown in Table 1-1, there are 11 raw materials and 16 nutrients in sow feed, so the value of m is 11 and the value of n is 16.

[0100] The fitness function is defined to evaluate the pros and cons of each ratio scheme. The fitness function has two evaluation indicators: total cost and whether the nutritional components of each raw material are within the constraint range. The fitness function is as follows:

[0101] f(x)=minF(x)+pen(x) (19)

[0102] Where minF(x) is the objective function of formula (16); pen(x) is the penalty term used to penalize solutions that do not meet the constraints. The expression is as follows:

[0103] pen(x)=p1(x)+p2(x)+p3(x) (20)

[0104]

[0105] Where: p1(x) is the penalty for the ingredient ratio constraint, which is used when the raw material does not meet the upper and lower limits, p 2 (x) is the penalty for the total proportion constraint, which is used for not satisfying equation (18), and p3(x) is the penalty for the nutritional demand constraint, which is used for not satisfying equation (19); in equation (21), li and ui are the lower limit and upper limit of the i-th raw material, respectively.

[0106] Optimization process

[0107] As shown in Table 1-1, there are 11 raw materials and 16 nutrients in sow feed, so m is 11 and n is 16. The specific implementation steps of BANChOA in the research of sow feed formula are as follows.

[0108] Step 1: Initialize algorithm-related parameters, such as the dimension of the raw material solution, the maximum number of iterations, the upper and lower bounds of the search space, the objective function, etc.

[0109] Step 2: Define the fitness function to evaluate the quality of each ratio scheme. The fitness function has two evaluation indicators: total cost and whether the nutritional content of each raw material is within the constraint range.

[0110] Step 3: Use the Bernoulli mapping sequence to generate a population, where each population individual is a set of 11 raw material formula contents.

[0111] Step 4: Enter the main loop, calculate f and CR dynamic coefficients, calculate the fitness of each search agent position, and evaluate the quality of the recipe based on its nutritional content and cost. Parameter f is the convergence factor. In the improved algorithm, the convergence factor of linear decay is changed to the convergence factor of nonlinear decay. The improvement of the algorithm adds a neighborhood perturbation strategy, and CR is a parameter used to determine whether to perform neighborhood perturbation. The later the algorithm enters, the greater the probability of neighborhood perturbation as CR increases.

[0112] Step 5: Update the position of the search agent according to the positions of attackers, obstaclers, pursuers, and evictors to explore a better feed formula.

[0113] Step 6: Determine whether neighborhood operations can be performed. If so, use neighborhood operations to further optimize the position of each search agent. Otherwise, go directly to Step 7.

[0114] Step7: If the maximum number of iterations is reached, record the current optimal solution and output the result, otherwise return to Step4.

[0115] Analysis of recipe optimization results

[0116] The above experimental design was used to compare BANChOA and ChOA on sow feed formula to verify the effectiveness of the improved algorithm in solving the actual sow feed formula. The following is a comparison of sow feed formulas during early pregnancy (<90 days), late pregnancy (>90 days) and lactation.

[0117] Table 2-7 Comparison of ChOA and BANChOA in feed formula during early pregnancy (<90 days)

[0118]

[0119] Table 2-8 Comparison of ChOA and BANChOA in feed formula for late pregnancy (>90 days)

[0120]

[0121]

[0122] Table 2-9 Comparison of ChOA and BANChOA in lactation feed formula

[0123]

[0124]

[0125] Tables 2-7, 2-8 and 2-9 show the proportion and cost of different raw materials in the feed formulas for sows at different stages obtained by ChOA and BANChOA. Figure 4 , Figure 5 and Figure 6 The following are the corresponding total cost trend comparison charts. The experimental results show that, under the condition of meeting the nutritional content required by sows, the improved ChOA reduces the cost per kilogram by 8.5% in the early pregnancy, 14.4% in the late pregnancy, and 18.2% in the lactation period compared with the standard ChOA. The price of BANChOA-optimized sow feed has been significantly reduced.

[0126] It should be noted that, in the present invention, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "include..." or "comprise..." do not exclude the existence of other elements in the process, method, article or terminal device including the elements. In addition, in the present invention, "greater than", "less than", "exceed" and the like are understood to exclude the number itself; "above", "below", "within" and the like are understood to include the number itself.

[0127] Although the above embodiments have been described, once those skilled in the art know the basic creative concepts, they can make additional changes and modifications to these embodiments. Therefore, the above description is only an embodiment of the present invention and does not limit the patent protection scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the specification and drawings of the present invention, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for screening a nutritional feed formula for sows based on an optimization algorithm, characterized in that: Sow feed m is the number of formula raw materials, n is the number of nutrients, and the steps of the sow nutritional feed formula screening method are as follows: Step 1: Initialize algorithm-related parameters, the dimension of the raw solution, the maximum number of iterations, the upper and lower bounds of the search space, and the objective function; Step 2: Define the fitness function to evaluate the pros and cons of each ratio scheme. The fitness function has two evaluation indicators: total cost and whether the nutritional components of each raw material are within the constraint range; Step 3: Use the Bernoulli mapping sequence to generate a population, where each population individual is a set of m raw material formula contents; Step 4: Enter the main loop, calculate the convergence factor f and the neighborhood disturbance CR dynamic coefficient, calculate the fitness of each search agent position, and evaluate the quality of the recipe based on its nutritional content and cost; Step 5: Update the position of the search agent according to the positions of the attacker, obstacle, chaser, and evictor to explore a better feed formula; Step 6: Determine whether neighborhood operations can be performed. If so, use neighborhood operations to further optimize the position of each search agent. Otherwise, go directly to Step 7. Step7: If the maximum number of iterations is reached, record the current optimal solution and output the result, otherwise return to Step4.

2. The method for screening sow nutritional feed formula based on optimization algorithm according to claim 1, characterized in that: The objective function minF(X) model is as follows: In the formula, c i —The price of each raw material, x i —The proportion of different raw materials in 1kg feed formula.

3. The method for screening sow nutritional feed formula based on optimization algorithm as claimed in claim 2, characterized in that The upper and lower limits of the amount of raw materials used in the sow feed, as well as the constraints on the nutritional composition and the total proportion of raw materials are as follows: x1+x2+…+x m =1 (18) Where: a ij —The percentage of the jth nutrient contained in the i-th raw material; d i —The minimum standards of various nutritional indicators that should be met in the formula; m—the number of raw materials in the formula; n—the number of nutritional ingredients.

4. The method for screening sow nutritional feed formula based on optimization algorithm according to claim 3, characterized in that: The fitness function is used to evaluate the pros and cons of each ratio scheme. The fitness function has two evaluation indicators: total cost and whether the nutritional components of each raw material are within the constraint range. The fitness function is as follows: f(x)=minF(x)+pen(x) (19) Where minF(x) is the objective function of formula (16); pen(x) is the penalty term used to penalize solutions that do not meet the constraints. The expression is as follows: pen(x)=p 1 (x)+p 2 (x)+p 3 (x) (20) Where: p 1 (x) is the penalty for the ingredient ratio constraint, which is used when the raw material does not meet the upper and lower limits, p 2 (x) is the penalty for the total proportion constraint, which is used to not satisfy equation (18), p 3 (x) is the penalty for the nutritional requirement constraint, which is used for not satisfying equation (19); where in equation (21), li and ui are the lower and upper limits of the i-th raw material, respectively.

Images