CN117011084B - Soybean planting optimization method and device based on black box constraint - Google Patents
Soybean planting optimization method and device based on black box constraint Download PDFInfo
- Publication number
- CN117011084B CN117011084B CN202310713438.3A CN202310713438A CN117011084B CN 117011084 B CN117011084 B CN 117011084B CN 202310713438 A CN202310713438 A CN 202310713438A CN 117011084 B CN117011084 B CN 117011084B
- Authority
- CN
- China
- Prior art keywords
- black box
- population
- individuals
- optimal
- hisoma
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 113
- 238000000034 method Methods 0.000 title claims abstract description 75
- 235000010469 Glycine max Nutrition 0.000 title claims abstract description 65
- 244000068988 Glycine max Species 0.000 title claims abstract description 65
- 230000005012 migration Effects 0.000 claims abstract description 37
- 238000013508 migration Methods 0.000 claims abstract description 37
- 230000035772 mutation Effects 0.000 claims abstract description 22
- 230000009286 beneficial effect Effects 0.000 claims abstract description 5
- 230000006872 improvement Effects 0.000 claims abstract description 4
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims description 30
- 239000003337 fertilizer Substances 0.000 claims description 21
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 claims description 16
- ZLMJMSJWJFRBEC-UHFFFAOYSA-N Potassium Chemical compound [K] ZLMJMSJWJFRBEC-UHFFFAOYSA-N 0.000 claims description 16
- 229910052698 phosphorus Inorganic materials 0.000 claims description 16
- 239000011574 phosphorus Substances 0.000 claims description 16
- 239000011591 potassium Substances 0.000 claims description 16
- 229910052700 potassium Inorganic materials 0.000 claims description 16
- 229910052757 nitrogen Inorganic materials 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 239000000618 nitrogen fertilizer Substances 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000000611 regression analysis Methods 0.000 claims description 5
- KWYUFKZDYYNOTN-UHFFFAOYSA-M Potassium hydroxide Chemical compound [OH-].[K+] KWYUFKZDYYNOTN-UHFFFAOYSA-M 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 238000011161 development Methods 0.000 claims description 3
- 238000013178 mathematical model Methods 0.000 claims description 3
- 239000002686 phosphate fertilizer Substances 0.000 claims description 3
- 229940072033 potash Drugs 0.000 claims description 3
- 235000015320 potassium carbonate Nutrition 0.000 claims description 3
- BWHMMNNQKKPAPP-UHFFFAOYSA-L potassium carbonate Substances [K+].[K+].[O-]C([O-])=O BWHMMNNQKKPAPP-UHFFFAOYSA-L 0.000 claims description 3
- 230000008901 benefit Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 21
- 241000196324 Embryophyta Species 0.000 description 6
- 230000008569 process Effects 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000029553 photosynthesis Effects 0.000 description 2
- 238000010672 photosynthesis Methods 0.000 description 2
- 230000008635 plant growth Effects 0.000 description 2
- 238000012271 agricultural production Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000012010 growth Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 235000015097 nutrients Nutrition 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000012887 quadratic function Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000017260 vegetative to reproductive phase transition of meristem Effects 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Marketing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Operations Research (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Computational Mathematics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Primary Health Care (AREA)
- Animal Husbandry (AREA)
- Quality & Reliability (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Physiology (AREA)
- Agronomy & Crop Science (AREA)
Abstract
A soybean planting optimization method and device based on black box constraint belong to the technical field of crop yield prediction optimization, and comprise an improvement method of a self-organizing migration algorithm and a black box optimization method; the improved method of the self-organizing migration algorithm comprises the following steps: the first step: initializing a population, and secondly: initializing parameters, and thirdly: entering a migration cycle, updating individual positions, and fourth step: fifth step of combined mutation operation: and (3) performing a similar removal operation, namely, a sixth step: judging whether an iteration termination condition is met; the black box optimization method comprises 10 steps. The beneficial effects of the invention are as follows: an improved self-organizing migration algorithm is provided, a combined updating strategy is provided, global and local searching capacity of the algorithm is effectively balanced, so that the optimal yield of soybean is obtained, and the optimization speed and accuracy are improved; in addition, reasonable planting density and fertilizing amount can not only improve soybean yield, but also save planting cost and improve economic benefit.
Description
Technical Field
The invention relates to the technical field of crop yield prediction optimization, in particular to a soybean planting optimization method and device based on black box constraint.
Background
Under proper ecological conditions, the yield per unit area can be improved by increasing the planting density. This is because a higher planting density can increase the total amount of photosynthesis and the light energy utilization efficiency, and the growth of plants is made tighter, thereby increasing the leaf area index per unit area, the dry matter accumulation amount, the yield per plant, etc., and thus increasing the yield of soybeans. However, too high a density of plants can also result in increased competition between plants, affecting photosynthesis and ventilation and thus soybean yield. The nitrogen fertilizer is a main nutrient required for the growth of soybeans, and proper nitrogen fertilizer application can increase leaf area index, plant height and dry matter accumulation, so that the yield of the soybeans is improved. However, excessive nitrogen fertilizer can cause the plant to grow too luxuriantly, and the yield and quality are reduced. Phosphorus and potassium are important elements for promoting flowering and fruiting and stress resistance of soybeans, and proper application can improve yield and quality, but excessive application can affect soil environment and plant growth. Therefore, the amount of nitrogen, phosphorus and potassium fertilizer applied to the soybean and the planting density have a great influence on the soybean yield, but it is not clear what is specifically a relation between the soybean yield and the amount of nitrogen, phosphorus and potassium fertilizer applied and the planting density.
The black box optimization problem is to search for the input variable combination which enables the output to reach the optimum by observing the input and output data under the condition that the internal mechanism of the system cannot be directly observed. If the black box optimization problem is described in a mathematical language, the black box optimization problem refers to the inability to obtain a functional analytical expression, gradient information, or other relevant mathematical properties between the input and the output. The problem of optimizing black boxes in real life is also widely existed, for example, the problem of the relation between the soybean yield and the amount of applied nitrogen, phosphorus and potassium fertilizers and the planting density is a black box optimizing problem.
To solve the black box optimization problem, only the secondary orthogonal rotation regression test can be used to obtain the input and output data of the black box optimization problem. On the basis, two methods for solving the black box optimization problem are available: the first type is that a functional relation between the input and the output of the black box optimization problem is obtained through regression analysis, namely, an objective function of the black box optimization problem is obtained, and then the optimal solution and the optimal value of the black box optimization problem are obtained; the other is to fit a functional relation between the input and the output of the black box optimization problem by using a BP neural network, and then calculate the optimal solution and the optimal value of the black box optimization problem. When the objective function of the black box optimization problem is obtained, the conventional method for solving the optimal solution of the objective function of the black box optimization problem is a traditional optimization method, such as a gradient descent method, a gradient projection method and the like.
In summary, the existing optimization method for solving the black box optimization problem has the following problems: firstly, fitting a relation between the input and the output of the black box by using a multi-element quadratic function, wherein the fitting precision is not high; secondly, the self-organizing migration algorithm is easy to fall into local optimum, and population diversity is gradually lost in the later period of iteration.
Disclosure of Invention
The invention aims to solve the technical problem of providing the soybean planting optimization method and the soybean planting optimization device based on the black box constraint, which can give the optimal planting density of soybean, the application amount of nitrogen, phosphorus and potassium fertilizers and the optimal soybean yield value, improve the fitting precision and improve the solving quality of the objective function of the black box optimization problem.
The technical problem to be solved by the invention is realized by the following technical scheme, and the invention is a soybean planting optimization method based on black box constraint, which is characterized in that: an improvement method comprising a self-organizing migration algorithm and a black box optimization method;
the improved method of the self-organizing migration algorithm comprises the following steps:
The first step: and initializing a population, wherein the initial population scale is n, and the dimension of the variable is D. Calculating objective function values of all individuals in the population, and keeping the optimal solutions and the optimal values of the individuals, wherein the optimal individuals in the population are leaders;
And a second step of: initializing parameters, namely taking values of a Step length and a disturbance parameter PRT;
and a third step of: entering a migration cycle, updating the positions of individuals, and calculating objective function values of all the individuals after the positions of all the individuals are updated;
fourth step: combining mutation operation, calculating fitness value of individuals after mutation, and reserving n better individuals from the population before mutation and after mutation;
Fifth step: the method comprises the steps of performing a dissimilarity operation, calculating an individual fitness value newly generated by the dissimilarity operation, and recording and updating a global optimal solution and an optimal value;
sixth step: judging whether the iteration termination condition is met, and if so, outputting a global optimal solution and an optimal value; otherwise, returning to the third step;
the black box optimization method comprises the following steps:
step1: initializing parameters including parameters of a Back Propagation Neural Network (BPNN) and parameters of a hybrid modified self-organizing migration algorithm (HISOMA-ASR) with adaptive dissimilarity operation, namely, the number of input layer neurons, the number of hidden layer neurons, the number of output layer neurons, weights of each layer and a threshold; HISOMA-ASR, a maximum value PATHLENGTH of migration step length, a perturbation parameter value lower limit PRT l and a maximum running time Maxruntime of an algorithm;
step2: selecting an activation function of each hidden layer and each output layer, and carrying out normalization processing on sample data;
Step3: randomly generating HISOMA-ASR initial population, namely optimizing the weight and the threshold of the BPNN;
step4: calculating the output of the BPNN;
Step5: taking the error of the BPNN as a HISOMA-ASR optimized objective function value, calculating the fitness value of the individuals in the HISOMA-ASR population, and determining a leader in the population;
step6: performing position updating on all individuals in HISOMA-ASR population;
Step7: judging whether HISOMA-ASR reaches the maximum running time, if so, calculating the fitness value of individuals in the population, determining a leader in HISOMA-ASR, wherein the leader is the optimal weight and threshold of the BPNN, and turning to Step7; otherwise, turning to Step6;
Step8: the method comprises the steps of obtaining a functional relation between the input and the output of a black box optimization problem, namely an objective function of the black box optimization problem, by taking the input of the black box optimization problem, namely the input of the BPNN as a decision variable, taking the output of the black box optimization problem as a variable, and taking the weight and the threshold of the BPNN as optimal weight and threshold obtained by optimization;
Step9: using HISOMA-ASR to find the optimal solution and optimal value of the black box optimization problem objective function;
Step10: and (3) performing inverse normalization processing on the optimal solution and the optimal value in Step9 to obtain the optimal solution and the optimal value of the black box optimization problem.
The technical problem to be solved by the invention is realized by the following technical scheme, and the Step length Step has the following calculation formula: step= (1.8-runtime/Maxruntime) × (0.15+0.1rand) where rand is a random number between 0 and 1, runtime is the current run time, maxruntime is the maximum run time.
The technical problem to be solved by the invention is realized by the following technical scheme, and the calculation formula of the disturbance parameter PRT is as follows: prt=prt l +0.5sin (0.5pi× runtime/Maxruntime), PRT l is the lower limit of the disturbance parameter, usually 0.1, runtime is the current running time, and Maxruntime is the maximum running time.
The technical problem to be solved by the invention is realized by the following technical scheme, and the steps of the dissimilarity operation are as follows: firstly, calculating the similarity S between any two individuals in the population,Wherein D is the dimension of the variable; secondly, setting zeta as a threshold value, and considering that the individual i and the individual j are similar when S is less than or equal to zeta; finally, the fitness values of the two individuals are compared, and the individual with the difference of the fitness values is initialized again.
The technical problem to be solved by the invention is realized by the following technical scheme, and the device of the soybean planting optimization method based on black box constraint is characterized by comprising the following components:
Parameter adjustment module: the method is used for adjusting the migration Step length Step and the disturbance parameter PRT;
and a position updating module: the rest individuals in the population migrate to the leader, and the leader does not update the position;
and a mixed migration module: the method is used for migrating individuals in the population, and the improved mixed migration strategy well balances the exploration and development capabilities of the algorithm, thereby being beneficial to improving the performance of the self-organizing migration algorithm;
a combined mutation module: performing mutation on individuals in the population;
The dissimilarity module: removing similar individuals in the population, and ensuring population diversity;
Fitting module: fitting the soybean planting density, the nitrogen fertilizer fertilizing amount, the phosphate fertilizer fertilizing amount, the potash fertilizer fertilizing amount and the soybean yield, and obtaining a functional relation between the soybean yield and the planting density and the nitrogen, phosphorus and potassium fertilizer fertilizing amount through regression analysis;
And an optimization module: using HISOMA-ASR to optimize a mathematical model of the black box optimization problem, wherein the obtained optimal solution and optimal value are the optimal solution of the black box optimization problem;
and an output module: outputting the optimal value and the optimal solution.
Compared with the prior art, the invention has the beneficial effects that:
(1) The optimization method for improving the self-organizing migration algorithm and designing the BP neural network based on the improved self-organizing migration algorithm is provided and is used for solving the black box optimization problem in the agricultural production practice so as to better solve the black box optimization problem in the various disciplinary fields such as natural science and engineering technology, save social cost and create larger economic value;
(2) Providing an improved self-organizing migration algorithm, respectively providing a self-adapting adjustment method of Step length and disturbance parameter PRT, providing a combined updating strategy, providing a disturbance strategy related to the current optimal solution, introducing a combined mutation operator and removing operations of similar individuals of a population, thereby effectively balancing the global and local searching capacities of the algorithm;
(3) The BP neural network structure is optimized by utilizing an improved self-organizing migration algorithm, so that the planting density and the fertilizing amount of soybeans are adjusted, the optimal yield of the soybeans is obtained, and the optimization speed and the optimization accuracy are improved; in addition, reasonable planting density and fertilizing amount can not only improve soybean yield, but also save planting cost and improve economic benefit.
Drawings
FIG. 1 is a technical roadmap of the invention;
FIG. 2 is a flow chart of an evolutionary strategy of HISOMA-ASR;
FIG. 3 is a flowchart of an optimization method for black box optimization problem based on HISOMA-ASR and BPNN;
Fig. 4 is a block diagram of a soybean planting optimization device based on black box constraint.
Detailed Description
Specific embodiments of the invention are described further below in order to facilitate a further understanding of the invention by those skilled in the art without limiting the scope of the claims thereto.
Referring to fig. 1 to 4, a soybean planting optimization method based on black box constraint is characterized in that: an improvement method comprising a self-organizing migration algorithm and a black box optimization method;
the improved method of the self-organizing migration algorithm comprises the following steps:
The first step: and initializing a population, wherein the initial population scale is n, and the dimension of the variable is D. Calculating objective function values of all individuals in the population, and keeping the optimal solutions and the optimal values of the individuals, wherein the optimal individuals in the population are leaders;
And a second step of: initializing parameters, namely taking values of a Step length and a disturbance parameter PRT;
And a third step of: entering a migration cycle, updating the positions of individuals, judging whether the individuals are leaders in the population, and if so, pressing the formula Performing position updating; otherwise, updating the position according to the formula; after all individuals finish the position updating, calculating objective function values of all individuals;
Fourth step: combining mutation operations, performing mutation according to formula x i'＝xi+ce(xbest-xi), formula x i'＝xi+Fe(xbest-xi)+Fe(xr1-xr2) or formula x i'＝xbest+Fe(xr1-xr2), calculating fitness values of individuals after mutation, and reserving n better individuals from the population before mutation and after mutation;
Fifth step: the method comprises the steps of performing a dissimilarity operation, calculating an individual fitness value newly generated by the dissimilarity operation, and recording and updating a global optimal solution and an optimal value;
sixth step: judging whether the iteration termination condition is met, and if so, outputting a global optimal solution and an optimal value; otherwise, returning to the third step;
the black box optimization method comprises the following steps:
step1: initializing parameters including parameters of a Back Propagation Neural Network (BPNN) and parameters of a hybrid modified self-organizing migration algorithm (HISOMA-ASR) with adaptive dissimilarity operation, namely, the number of input layer neurons, the number of hidden layer neurons, the number of output layer neurons, weights of each layer and a threshold; HISOMA-ASR, a maximum value PATHLENGTH of migration step length, a perturbation parameter value lower limit PRT l and a maximum running time Maxruntime of an algorithm;
step2: selecting an activation function of each hidden layer and each output layer, and carrying out normalization processing on sample data;
Step3: randomly generating HISOMA-ASR initial population, namely optimizing the weight and the threshold of the BPNN;
step4: calculating the output of the BPNN;
Step5: taking the error of the BPNN as a HISOMA-ASR optimized objective function value, calculating the fitness value of the individuals in the HISOMA-ASR population, and determining a leader in the population;
step6: performing position updating on all individuals in HISOMA-ASR population;
Step7: judging whether HISOMA-ASR reaches the maximum running time, if so, calculating the fitness value of individuals in the population, determining a leader in HISOMA-ASR, wherein the leader is the optimal weight and threshold of the BPNN, and turning to Step7; otherwise, turning to Step6;
Step8: the method comprises the steps of obtaining a functional relation between the input and the output of a black box optimization problem, namely an objective function of the black box optimization problem, by taking the input of the black box optimization problem, namely the input of the BPNN as a decision variable, taking the output of the black box optimization problem as a variable, and taking the weight and the threshold of the BPNN as optimal weight and threshold obtained by optimization;
Step9: using HISOMA-ASR to find the optimal solution and optimal value of the black box optimization problem objective function;
Step10: and (3) performing inverse normalization processing on the optimal solution and the optimal value in Step9 to obtain the optimal solution and the optimal value of the black box optimization problem.
The technical problem to be solved by the invention is realized by the following technical scheme, and the Step length Step has the following calculation formula:
Step= (1.8-runtime/Maxruntime) × (0.15+0.1rand) where rand is a random number between 0 and 1, runtime is the current run time, maxruntime is the maximum run time.
The technical problem to be solved by the invention is realized by the following technical scheme, and the calculation formula of the disturbance parameter PRT is as follows: prt=prt l +0.5sin (0.5pi× runtime/Maxruntime), PRT l is the lower limit of the disturbance parameter, usually 0.1, runtime is the current running time, and Maxruntime is the maximum running time.
The technical problem to be solved by the invention is realized by the following technical scheme, and the steps of the dissimilarity operation are as follows: firstly, calculating the similarity S between any two individuals in the population,Wherein D is the dimension of the variable; secondly, setting zeta as a threshold value, and considering that the individual i and the individual j are similar when S is less than or equal to zeta; finally, the fitness values of the two individuals are compared, and the individual with the difference of the fitness values is initialized again.
The technical problem to be solved by the invention is realized by the following technical scheme, and the device of the soybean planting optimization method based on black box constraint is characterized by comprising the following components:
Parameter adjustment module 10: the method is used for adjusting the migration Step length Step and the disturbance parameter PRT, and the improved self-adaptive parameter adjustment mode balances the global and local searching capacity of the algorithm well, so that the performance of the algorithm is improved;
the location update module 20: setting the optimal individual in the population as x leader,xleader as a leader in the t-th iteration, and migrating the rest individuals except the leader in the population to x leader in the migration cycle process of each iteration, wherein the leader does not update the position; because the leader is the optimal individual in the population, if the leader is positioned near the local extreme point, the algorithm is easy to sink into the local optimal;
Hybrid migration module 30: the method is used for migrating individuals in the population, and the improved mixed migration strategy well balances the exploration and development capabilities of the algorithm, thereby being beneficial to improving the performance of the self-organizing migration algorithm;
The combination mutation module 40: the individual in the population is mutated, mutation operators with strong global searching capability are selected in the initial stage of iteration, and mutation operators with strong local searching capability are selected in the later stage of iteration;
The dissimilarity module 50: the optimizing process of the algorithm is a process of continuously converging towards the optimal solution in the population, more and more individuals are gathered near a leader in the population along with the increase of iteration times, the diversity of the population is poor, two or more similar individuals possibly appear in the population, and the algorithm is easy to fall into local optimal when solving the multi-extremum optimizing problem; therefore, in the iterative process of the algorithm, similar individuals in the population should be removed so as to improve the global searching capability of the algorithm and reduce the possibility of the algorithm falling into local optimum;
fitting module 60: fitting the soybean planting density, the nitrogen fertilizer fertilizing amount, the phosphate fertilizer fertilizing amount, the potash fertilizer fertilizing amount and the soybean yield, and obtaining a functional relation between the soybean yield and the planting density and the nitrogen, phosphorus and potassium fertilizer fertilizing amount through regression analysis;
The optimization module 70: the method is characterized in that when the planting density of the soybeans and the application amount of the nitrogen, phosphorus and potassium fertilizers are the same, the yield of the soybeans is highest; using HISOMA-ASR to optimize a mathematical model of the black box optimization problem, wherein the obtained optimal solution and optimal value are the optimal solution of the black box optimization problem;
Output module 80: outputting the optimal value and the optimal solution.
Embodiment one: in order to determine the functional relation between the soybean yield and the application amount and the planting density of the nitrogen, phosphorus and potassium fertilizers, the application amount and the planting density of the nitrogen, phosphorus and potassium fertilizers are used as the input of a black box, the soybean yield is used as the output of the black box, and the data of the soybean yield, the application amount and the planting density of the nitrogen, phosphorus and potassium fertilizers are obtained through a secondary orthogonal rotation combination experiment. On the basis, the optimization method based on HISOMA-ASR and BPNN black box optimization problem is used for solving the soybean planting density and the nitrogen, phosphorus and potassium fertilizer application amount which enable the soybean yield to reach the highest.
The soybean planting parameter optimization refers to the highest soybean yield when the planting density of the soybean and the application amount of the nitrogen, phosphorus and potassium fertilizer are the same. If the soybean planting density and the nitrogen, phosphorus and potassium fertilizer application amount are used as the input of the BPNN, namely the input of the black box optimization problem; the soybean yield is used as the output of the BPNN, namely the output of the black box optimization problem; the function between the input and the output of the BPNN is the target function of the black box optimization problem if the function between the input and the output of the BPNN is fit to the function between the input and the output of the black box in the black box optimization problem. In addition, the reasonable value range of the soybean planting density and the nitrogen, phosphorus and potassium fertilizer application amount is used as the constraint condition of the black box optimization problem.
In order to solve for the optimal planting parameters of soybeans, the soybean planting parameters were optimized using the black box optimization method (HISOMA-ASR and BPNN based black box optimization method) presented herein. In addition, the soybean yield is fitted by using a multi-element quadratic regression model obtained by a regression analysis method, and the multi-element quadratic regression model is optimized by using HISOMA-ASR, so that the obtained optimal solution is the optimal planting parameter of the soybean.
Soybean planting parameters were optimized with the black box optimization method presented herein (HISOMA-ASR and BPNN based black box optimization method), respectively, with the existing BPNN based linear constraint optimization method (optimization with BP neural network fitting and gradient projection (gradient projection method, GPM) first).
Setting the population scale of HISOMA-ASR as 40, the dimension of the variable as 4, processing the constraint optimization problem by using a penalty function method, changing the constraint optimization problem into an unconstrained optimization problem, wherein the maximum running time of the algorithm is 60s, and the optimization results of 3 optimization methods are shown in table 1:
Table 13 optimization results of optimization methods
If the soybean variety is Heihe', the soybean yield is 3839.57kg/hm 2 at maximum when the planting density is 51.34 ×10 4 plants/hm 2, the N fertilizer application amount is 67.11kg/hm 2、P2O5, the fertilizer application amount is 68.42kg/hm 2、K2 O fertilizer application amount is 27.41kg/hm 2. Compared with the black box optimization method provided by the invention, the first black box optimization method is different in fitting method, the fitting accuracy of the multiple quadratic regression model is low, the soybean yield obtained by the two black box optimization methods is different, and the higher the fitting accuracy is, the higher the soybean yield is; compared with the black box optimization method, the gradient projection method is used in the optimization of the second black box optimization method, HISOMA-ASR is used in the optimization of the black box optimization method, the fitting method and the fitting precision of the two black box optimization methods are the same, but the soybean yields are different, and the soybean yield of the black box optimization method is the highest. Thus, the higher the fitting accuracy, the higher the large yield resulting from the optimization.
Claims (5)
1. A soybean planting optimization method based on black box constraint is characterized in that: an improvement method comprising a self-organizing migration algorithm and a black box optimization method;
the improved method of the self-organizing migration algorithm comprises the following steps:
the first step: initializing a population, wherein the initial population scale is n, the dimension of a variable is D, calculating objective function values of all individuals in the population, and keeping the optimal solution and the optimal value of the objective function values, wherein the optimal individuals in the population are leaders;
And a second step of: initializing parameters, namely taking values of a Step length and a disturbance parameter PRT;
and a third step of: entering a migration cycle, updating the positions of individuals, and calculating objective function values of all the individuals after the positions of all the individuals are updated;
fourth step: combining mutation operation, calculating fitness value of individuals after mutation, and reserving n better individuals from the population before mutation and after mutation;
Fifth step: the method comprises the steps of performing a dissimilarity operation, calculating an individual fitness value newly generated by the dissimilarity operation, and recording and updating a global optimal solution and an optimal value;
sixth step: judging whether the iteration termination condition is met, and if so, outputting a global optimal solution and an optimal value; otherwise, returning to the third step;
the black box optimization method comprises the following steps:
step1: initializing parameters including parameters of a Back Propagation Neural Network (BPNN) and parameters of a hybrid modified self-organizing migration algorithm (HISOMA-ASR) with adaptive dissimilarity operation, namely, the number of input layer neurons, the number of hidden layer neurons, the number of output layer neurons, weights of each layer and a threshold; HISOMA-ASR, a maximum value PATHLENGTH of migration step length, a perturbation parameter value lower limit PRT l and a maximum running time Maxruntime of an algorithm;
step2: selecting an activation function of each hidden layer and each output layer, and carrying out normalization processing on sample data;
Step3: randomly generating HISOMA-ASR initial population, namely optimizing the weight and the threshold of the BPNN;
step4: calculating the output of the BPNN;
Step5: taking the error of the BPNN as a HISOMA-ASR optimized objective function value, calculating the fitness value of the individuals in the HISOMA-ASR population, and determining a leader in the population;
step6: performing position updating on all individuals in HISOMA-ASR population;
Step7: judging whether HISOMA-ASR reaches the maximum running time, if so, calculating the fitness value of individuals in the population, determining a leader in HISOMA-ASR, wherein the leader is the optimal weight and threshold of the BPNN, and turning to Step7; otherwise, turning to Step6;
Step8: the method comprises the steps of obtaining a functional relation between the input and the output of a black box optimization problem, namely an objective function of the black box optimization problem, by taking the input of the black box optimization problem, namely the input of the BPNN as a decision variable, taking the output of the black box optimization problem as a variable, and taking the weight and the threshold of the BPNN as optimal weight and threshold obtained by optimization;
Step9: using HISOMA-ASR to find the optimal solution and optimal value of the black box optimization problem objective function;
Step10: and (3) performing inverse normalization processing on the optimal solution and the optimal value in Step9 to obtain the optimal solution and the optimal value of the black box optimization problem.
2. The black box constraint-based soybean planting optimization method as claimed in claim 1, wherein: the Step has the following formula: step= (1.8-runtime/Maxruntime) × (0.15+0.1rand) where rand is a random number between 0 and 1, runtime is the current run time, maxruntime is the maximum run time.
3. The black box constraint-based soybean planting optimization method as claimed in claim 1, wherein: the calculation formula of the disturbance parameter PRT is as follows: prt=prt l +0.5sin (0.5pi× runtime/Maxruntime), PRT l is the lower limit of the disturbance parameter, usually 0.1, runtime is the current running time, and Maxruntime is the maximum running time.
4. The black box constraint-based soybean planting optimization method as claimed in claim 1, wherein: the step of the dissimilarity operation is as follows: firstly, calculating the similarity S between any two individuals in the population,Wherein D is the dimension of the variable; secondly, setting zeta as a threshold value, and considering that the individual i and the individual j are similar when S is less than or equal to zeta; finally, the fitness values of the two individuals are compared, and the individual with the difference of the fitness values is initialized again.
5. An apparatus for applying the black box constraint-based soybean planting optimization method of claim 1, comprising:
Parameter adjustment module: the method is used for adjusting the migration Step length Step and the disturbance parameter PRT;
and a position updating module: the rest individuals in the population migrate to the leader, and the leader does not update the position;
and a mixed migration module: the method is used for migrating individuals in the population, and the improved mixed migration strategy well balances the exploration and development capabilities of the algorithm, thereby being beneficial to improving the performance of the self-organizing migration algorithm;
a combined mutation module: performing mutation on individuals in the population;
The dissimilarity module: removing similar individuals in the population, and ensuring population diversity;
Fitting module: fitting the soybean planting density, the nitrogen fertilizer fertilizing amount, the phosphate fertilizer fertilizing amount, the potash fertilizer fertilizing amount and the soybean yield, and obtaining a functional relation between the soybean yield and the planting density and the nitrogen, phosphorus and potassium fertilizer fertilizing amount through regression analysis;
And an optimization module: using HISOMA-ASR to optimize a mathematical model of the black box optimization problem, wherein the obtained optimal solution and optimal value are the optimal solution of the black box optimization problem;
and an output module: outputting the optimal value and the optimal solution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310713438.3A CN117011084B (en) | 2023-06-16 | 2023-06-16 | Soybean planting optimization method and device based on black box constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310713438.3A CN117011084B (en) | 2023-06-16 | 2023-06-16 | Soybean planting optimization method and device based on black box constraint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117011084A CN117011084A (en) | 2023-11-07 |
| CN117011084B true CN117011084B (en) | 2024-05-03 |
Family
ID=88575283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310713438.3A Active CN117011084B (en) | 2023-06-16 | 2023-06-16 | Soybean planting optimization method and device based on black box constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117011084B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106373022A (en) * | 2016-09-13 | 2017-02-01 | 重庆科技学院 | BP-GA-based greenhouse crop plantation efficiency condition optimization method and system |
| CN112001112A (en) * | 2020-06-23 | 2020-11-27 | 辽宁科技学院 | Optimization method of linear constraint black box optimization problem based on BP neural network |
| CN115392426A (en) * | 2022-08-17 | 2022-11-25 | 支付宝(杭州)信息技术有限公司 | Self-organizing migration service processing method, device and equipment |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230083892A1 (en) * | 2020-02-07 | 2023-03-16 | Google Llc | Population-based black-box optimization |
-
2023
- 2023-06-16 CN CN202310713438.3A patent/CN117011084B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106373022A (en) * | 2016-09-13 | 2017-02-01 | 重庆科技学院 | BP-GA-based greenhouse crop plantation efficiency condition optimization method and system |
| CN112001112A (en) * | 2020-06-23 | 2020-11-27 | 辽宁科技学院 | Optimization method of linear constraint black box optimization problem based on BP neural network |
| CN115392426A (en) * | 2022-08-17 | 2022-11-25 | 支付宝(杭州)信息技术有限公司 | Self-organizing migration service processing method, device and equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117011084A (en) | 2023-11-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110782658B (en) | Traffic prediction method based on LightGBM algorithm | |
| CN112906298B (en) | Blueberry yield prediction method based on machine learning | |
| CN109214579B (en) | BP neural network-based saline-alkali soil stability prediction method and system | |
| CN107145941B (en) | Method for dynamically acquiring light demand quantity in real time based on optimal light quality and photon flux density | |
| CN112330487B (en) | Photovoltaic power generation short-term power prediction method | |
| CN115063252A (en) | Crop fertilizer accurate application method and system based on neural network | |
| CN111915062A (en) | Greenhouse crop water demand regulation and control method with water utilization rate and photosynthetic rate coordinated | |
| CN105184400A (en) | Tobacco field soil moisture prediction method | |
| CN116258060A (en) | Soil testing formula fertilization method based on machine learning | |
| CN117688404B (en) | Intelligent fertilization formula decision method, intelligent fertilization formula decision device, intelligent fertilization formula decision equipment, intelligent fertilization formula decision medium and program product | |
| CN105913326B (en) | Constrain the Crop growing stage model Cultivar parameter optimization method of sex knowledge and the tactful Genetic Algorithm Fusion of elite individual | |
| Cheng et al. | Precision agriculture management based on a surrogate model assisted multiobjective algorithmic framework | |
| Saleh et al. | A prediction of grain yield based on hybrid intelligent algorithm | |
| CN117011084B (en) | Soybean planting optimization method and device based on black box constraint | |
| An et al. | A simulator-based planning framework for optimizing autonomous greenhouse control strategy | |
| CN116595333B (en) | Soil-climate intelligent rice target yield and nitrogen fertilizer consumption determination method | |
| Acharjya et al. | An integrated fuzzy rough set and real coded genetic algorithm approach for crop identification in smart agriculture | |
| CN109934400B (en) | Rain collecting, regulating and deficiency crop water demand prediction method based on improved neural network | |
| CN111126827A (en) | Input-output accounting model construction method based on BP artificial neural network | |
| Sakthipriya et al. | Precision agriculture based on convolutional neural network in rice production nutrient management using machine learning genetic algorithm | |
| Ujjainia et al. | Crop Yield Prediction using Regression Model | |
| Jiao et al. | Prediction model of rice seedling growth and rhizosphere fertility based on the improved Elman neural network | |
| Yuan et al. | Photosynthetic rate prediction model based on PSO-LSSVM for optimization and control of greenhouse environment | |
| Li et al. | An Efficient Competitive Swarm Optimizer for Solving Large-Scale Multi-objective Optimization Problems | |
| Roshan et al. | Intelligent Chemical Fertilizer Recommendation System for Rice Fields |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |