CN113987806A - Atmospheric mode optimization method based on proxy model - Google Patents
Atmospheric mode optimization method based on proxy model Download PDFInfo
- Publication number
- CN113987806A CN113987806A CN202111273554.5A CN202111273554A CN113987806A CN 113987806 A CN113987806 A CN 113987806A CN 202111273554 A CN202111273554 A CN 202111273554A CN 113987806 A CN113987806 A CN 113987806A
- Authority
- CN
- China
- Prior art keywords
- model
- atmospheric
- mode
- proxy
- value
- 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.)
- Granted
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 62
- 238000000034 method Methods 0.000 title claims abstract description 41
- 239000002245 particle Substances 0.000 claims abstract description 47
- 238000005070 sampling Methods 0.000 claims abstract description 33
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 18
- 230000003044 adaptive effect Effects 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 8
- 238000004088 simulation Methods 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000013461 design Methods 0.000 abstract description 4
- 238000011156 evaluation Methods 0.000 abstract description 4
- 238000012795 verification Methods 0.000 abstract description 4
- 239000003795 chemical substances by application Substances 0.000 description 20
- 238000004364 calculation method Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000001556 precipitation Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 101100438241 Arabidopsis thaliana CAM5 gene Proteins 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012950 reanalysis Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Complex Calculations (AREA)
- Feedback Control In General (AREA)
Abstract
The invention is suitable for the field of proxy models, provides an atmospheric mode optimization method based on a proxy model, discloses an atmospheric mode parameter tuning method based on the proxy model, and belongs to the field of design parameter optimization, wherein the atmospheric mode parameter tuning method comprises the following steps: determining a parameter range and sampling, simulating results of each sample by using an atmospheric mode, fitting a functional relation between the samples and the mode results by constructing an agent model by taking Root Mean Square Error (RMSE) as an evaluation standard of the output results of the atmospheric mode, searching an optimal value of the agent model by using a Particle Swarm Optimization (PSO) algorithm and bringing the optimal value into the atmospheric mode for verification, and continuously and iteratively updating the agent model until the optimization requirement is met and a final parameter optimization result is output. The invention reduces the resource cost by applying the idea of the proxy model to the parameter tuning of the atmospheric mode; meanwhile, the time consumption of mode tuning is reduced, and the efficiency of atmospheric mode parameter tuning is improved.
Description
Technical Field
The invention belongs to the field of proxy models, and particularly relates to an atmospheric mode optimization method based on a proxy model.
Background
The atmospheric mode is one of the important components of the earth system mode, is an important tool for researching the climate rule and predicting the climate change, and simulates various physical and chemical changes of the earth through a nonlinear equation set; on a sub-grid scale, some physical processes are represented in the form of parameters, and therefore, the numerical determination of such parameters is crucial during the operation of the mode, and a small change may cause a large deviation of the result, so that the adjustment of these parameter values is one of the important conditions for stable and accurate operation of the mode.
The problem of selecting the parameters can be abstracted to the traditional parameter optimization problem, a large amount of research is carried out at home and abroad according to the characteristics and the parameter significance of the atmospheric mode, and various optimization algorithms are provided, but the problem is different from the traditional parameter optimization problem in that the cost required by the operation of the atmospheric mode is extremely high, and from the scientific point of view, the atmospheric mode needs 3-5 years of integral stabilization time from the starting state to the equilibrium state; from the application point of view of the optimization algorithm, it is common to reach a convergence condition that satisfies the requirements, requiring at least a dozen or dozen iterations. In such high-overhead scenarios, the time overhead of iterating through the optimization algorithm to obtain the optimal solution is often unacceptable.
The agent model is an analysis model which has small calculation amount, but the calculation result is similar to the calculation analysis result of the high-precision model. The method is widely applied to various engineering fields, on one hand, the time expenditure of parameter tuning can be greatly reduced by using the proxy model, on the other hand, more efficient optimization algorithms can be applied to parameter tuning of the atmospheric mode,
aiming at the problems in the related art, the research of applying the agent model to atmospheric mode parameter tuning is not developed at present.
Disclosure of Invention
The embodiment of the invention aims to provide an atmospheric mode optimization method based on a proxy model, and aims to solve the problem that the research of applying the proxy model to atmospheric mode parameter optimization is not carried out at present.
The embodiment of the invention is realized in such a way that an atmospheric mode optimization method based on a proxy model comprises the following steps:
the method comprises the following steps: determining the optimized parameters and the optimized target, and determining the upper limit and the lower limit of the optimized parameters according to the selected parameters;
step two: sampling within a parameter range by utilizing Latin hypercube sampling;
step three: sequentially bringing the parameter samples obtained in the step two into a parameter input list of the atmospheric mode, starting to execute the atmospheric mode, reading an output file of the atmospheric mode after the atmospheric mode corresponding to each sampling value is executed, obtaining the change of different samples to an optimization target, and judging the change level of each sample to the mode by using a Root Mean Square Error (RMSE);
step four: constructing a proxy model by using the data obtained in the step three;
step five: searching an optimal solution of the proxy model by utilizing a particle swarm algorithm;
step six: bringing the parameters represented by the solutions in the step five back to an atmospheric mode to obtain a new group of values, and adding the new group of values into a solution set for constructing the agent model to update the agent model;
step seven: judging whether the result in the step six meets the optimization requirement, if not, respectively adding the parameters in the step six and the RMSE for the atmospheric mode simulation into the sample set S and the RMSE set in the step four, updating the proxy model, and repeating the step four to the step six;
step eight: and when the RMSE obtained in the step six meets the optimization standard, ending the tuning process, and outputting the parameters obtained at the moment, wherein the parameters are the final tuning realization result.
According to the further technical scheme, according to the second step, the Latin hypercube sampling step is as follows:
step 201: firstly, determining the number N of samples, namely the number of samples to be extracted;
step 202: equally dividing the interval (0, 1) into N sections;
step 203: randomly drawing a value in each of the N segments;
step 204: mapping the extracted value into a standard normal distribution sample through an inverse function of the standard normal distribution;
step 205: and (5) disordering the sampling sequence to obtain a final sampling result.
According to a further technical scheme, according to the third step, the RMSE calculation method comprises the following steps:
wherein N represents the result of the nth group of sampling, the value range is between 1 and N, m represents the total lattice number of the optimization area, yiRepresenting the result of the simulation of the object at the ith lattice pattern, yo,iRepresenting the value of the observed data at the ith grid point.
According to the fourth step, the method of the agent model comprises the following steps:
401: the polynomial proxy model takes a second-order polynomial as an example, and the expression is as follows:
wherein, beta represents the coefficient to be estimated, and d is the number of parameters;
402: kriging's agent model, the expression is as follows:
y＝f(x)Tβl+zl(x)，l＝1，2，...，q
where β is a regression coefficient of a polynomial, the polynomial f (x) may be of any order, z (x) is a random process, the mean value E [ z (x) ] ═ 0;
403: the RBF agent model has the following expression:
where i-1 to n represent response values of the sample point i, ωiRepresents the ith sample point weight coefficient, riIs the Euclidean distance r between the point to be measured and the ith sample pointi＝∥y-yi∥，φ(ri) For the mirror function, the Guass function is commonly used: phi (r) ═ r2/c2Wherein c is a coefficient.
According to the fifth technical scheme, the particle swarm algorithm comprises the following steps:
step 5.1: randomly initializing each particle;
step 5.2: calculating the adaptive value of each particle, wherein the adaptive value is a proxy model
The output result of (1);
step 5.3: acquiring an individual optimal value of the particle, and updating the individual optimal value of the particle if the adaptive value obtained by the particle in the step 5.2 is superior to the original optimal value of the particle;
step 5.4: acquiring a global optimum value of the particles, and if the adaptive value of the particles obtained in the step 5.3 is superior to the global optimum value, updating the global optimum value;
step 5.5: the velocity and position of each is updated as follows:
vi＝vi+c1*rand*(pbesti-xi)+c2*rand*(gbest-xi)
xi＝xi+vi
wherein v isiDenotes the velocity of the ith particle, c1And c2Is a self-learning factor and a group learning factor, and is two constants, rand is a random number between 0 and 1, pbestiIs the individual optimum of the ith particle, gbest is the global optimum of the particle, xiRepresenting the current position of the ith particle;
step 5.6: judging whether a convergence condition is met, if not, returning to the step 5.2;
step 5.7: and obtaining an optimal solution.
According to the second step, the hierarchical sampling mode is random hierarchical sampling.
According to the further technical scheme, according to the fifth step, the solution of the proxy model is the optimal solution.
The embodiment of the invention provides an atmospheric mode optimization method based on a proxy model, and discloses an atmospheric mode parameter tuning method based on a proxy model, which belongs to the field of design parameter optimization and comprises the following steps: determining a parameter range and sampling, simulating results of each sample by using an atmospheric mode, fitting a functional relation between the samples and the mode results by constructing an agent model by taking Root Mean Square Error (RMSE) as an evaluation standard of the output results of the atmospheric mode, searching an optimal value of the agent model by using a Particle Swarm Optimization (PSO) algorithm and bringing the optimal value into the atmospheric mode for verification, and continuously and iteratively updating the agent model until the optimization requirement is met and a final parameter optimization result is output. According to the invention, the idea of the proxy model is applied to parameter tuning of the atmospheric mode, so that the execution times of the atmospheric mode are effectively reduced, the resource overhead is reduced and the economic benefit is improved while the high-efficiency optimization algorithm is ensured; meanwhile, the time consumption of mode tuning is reduced, and the efficiency of atmospheric mode parameter tuning is improved.
Drawings
FIG. 1 is a schematic diagram of an optimization process according to an embodiment of the present invention;
fig. 2 is a schematic diagram of finding an optimal solution of a proxy model in the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Specific implementations of the present invention are described in detail below with reference to specific embodiments.
As shown in fig. 1 and 2, an atmospheric model optimization method based on a proxy model according to an embodiment of the present invention includes the following steps:
the method comprises the following steps: determining the optimized parameters and the optimized target, and determining the upper limit and the lower limit of the optimized parameters according to the selected parameters;
step two: sampling within a parameter range by utilizing Latin hypercube sampling;
step three: sequentially bringing the parameter samples obtained in the step two into a parameter input list of the atmospheric mode, starting to execute the atmospheric mode, reading an output file of the atmospheric mode after the atmospheric mode corresponding to each sampling value is executed, obtaining the change of different samples to an optimization target, and judging the change level of each sample to the mode by using a Root Mean Square Error (RMSE);
step four: constructing a proxy model by using the data obtained in the step three;
step five: searching an optimal solution of the proxy model by utilizing a particle swarm algorithm;
step six: bringing the parameters represented by the optimal solution in the step five back to an atmospheric mode to obtain a new group of values, and adding the new group of values into a solution set for constructing the agent model to update the agent model;
step seven: judging whether the result in the step six meets the optimization requirement, if not, respectively adding the parameters in the step six and the RMSE for the atmospheric mode simulation into the sample set S and the RMSE set in the step four, updating the proxy model, and repeating the step four to the step six;
step eight: and when the RMSE obtained in the step six meets the optimization standard, ending the tuning process, and outputting the parameters obtained at the moment, wherein the parameters are the final tuning realization result.
In the embodiment of the invention, the invention discloses an atmospheric mode parameter tuning method based on a proxy model, which belongs to the field of design parameter optimization and comprises the following steps: determining a parameter range and sampling, simulating results of each sample by using an atmospheric mode, fitting a functional relation between the samples and the mode results by constructing an agent model by taking Root Mean Square Error (RMSE) as an evaluation standard of the output results of the atmospheric mode, searching an optimal value of the agent model by using a Particle Swarm Optimization (PSO) algorithm and bringing the optimal value into the atmospheric mode for verification, and continuously and iteratively updating the agent model until the optimization requirement is met and a final parameter optimization result is output. According to the invention, the idea of the proxy model is applied to parameter tuning of the atmospheric mode, so that the execution times of the atmospheric mode are effectively reduced, the resource overhead is reduced and the economic benefit is improved while the high-efficiency optimization algorithm is ensured; meanwhile, the time consumption of mode tuning is reduced, and the efficiency of atmospheric mode parameter tuning is improved.
As shown in fig. 1 and 2, according to a preferred embodiment of the present invention, the step of latin hypercube sampling according to step two is as follows:
step 201: firstly, determining the number N of samples, namely the number of samples to be extracted;
step 202: equally dividing the interval (0, 1) into N sections;
step 203: randomly drawing a value in each of the N segments;
step 204: mapping the extracted value into a standard normal distribution sample through an inverse function of the standard normal distribution;
step 205: and (5) disordering the sampling sequence to obtain a final sampling result.
As shown in fig. 1 and 2, as a preferred embodiment of the present invention, according to step three, the RMSE calculation method is:
wherein N represents the result of the nth group of sampling, the value range is between 1 and N, m represents the total lattice number of the optimization area, yiRepresenting the result of the simulation of the object at the ith lattice pattern, yo,iRepresenting the value of the observed data at the ith grid point.
As shown in fig. 1 and 2, according to a preferred embodiment of the present invention, the method of the proxy model according to step four is as follows:
401: the polynomial proxy model takes a second-order polynomial as an example, and the expression is as follows:
wherein, beta represents the coefficient to be estimated, and d is the number of parameters;
402: kriging's agent model, the expression is as follows:
y＝f(x)Tβl+zl(x)，l＝1，2，...，q
where β is a regression coefficient of a polynomial, the polynomial f (x) may be of any order, z (x) is a random process, the mean value E [ z (x) ] ═ 0;
403: the RBF agent model has the following expression:
where i-1 to n represent response values of the sample point i, ωiRepresents the ith sample point weight coefficient, riIs the Euclidean distance r between the point to be measured and the ith sample pointi＝∥y-yi∥，φ(ri) For the mirror function, the Guass function is commonly used: phi (r) ═ r2/c2Wherein c is a coefficient.
As shown in fig. 1 and 2, as a preferred embodiment of the present invention, according to step five, the particle swarm algorithm comprises the following steps:
step 5.1: randomly initializing each particle;
step 5.2: calculating the adaptive value of each particle, wherein the adaptive value is a proxy model
The output result of (1);
step 5.3: acquiring an individual optimal value of the particle, and updating the individual optimal value of the particle if the adaptive value obtained by the particle in the step 5.2 is superior to the original optimal value of the particle;
step 5.4: acquiring a global optimum value of the particles, and if the adaptive value of the particles obtained in the step 5.3 is superior to the global optimum value, updating the global optimum value;
step 5.5: the velocity and position of each is updated as follows:
vi＝vi+c1*rand*(pbesti-xi)+c2*rand*(gbest-xi)
xi＝xi+vi
wherein v isiDenotes the velocity of the ith particle, c1And c2Is a reason for self-learningThe child and population learning factors are two constants, rand is a random number between 0 and 1, pbestiIs the individual optimum of the ith particle, gbest is the global optimum of the particle, xiRepresenting the current position of the ith particle;
step 5.6: judging whether a convergence condition is met, if not, returning to the step 5.2;
step 5.7: and obtaining an optimal solution.
As shown in fig. 1 and 2, according to a preferred embodiment of the present invention, according to step two, the sampling mode is random sampling.
As shown in fig. 1 and 2, according to step five, the solution of the proxy model is the optimal solution as a preferred embodiment of the present invention.
In the embodiment of the present invention, according to the embodiment of the present invention, there is provided an atmospheric model parameter tuning method based on a proxy model, using a Kriging-based proxy model, where the atmospheric model uses an atmospheric model CAM5.3 of CESM1.3 as an example, and takes global total precipitation as an optimization target, and fig. 1 describes the whole optimization process, and each optimization step is as follows:
the method comprises the following steps: determining a global optimization target as total precipitation, selecting the following six parameters, and determining an upper limit and a lower limit according to the parameter meanings:
the parameter lists and ranges are as follows:
| parameter name | Default value | Lower limit of | Upper limit of |
| cldfrc_rhminl | 0.8975 | 0.80 | 0.99 |
| micro_mg_dcs | 5*10^-6 | 1*10^-6 | 5*10^-6 |
| zmconv_dmpdz | -1.0*10^-3 | -2.0*10^-3 | -0.2*10^-3 |
| zmconv_tau | 3600 | 1800 | 28800 |
| zmconv_c0_ocn | 0.03 | 0.001 | 0.1 |
| micro_mg_ai | 700 | 350 | 1400 |
TABLE 1 parameter List and ranges
Step two: in the embodiment, 6 parameters are used, 60 groups of samples with 10 times of the number of the parameters are selected as the sampling number, each sample can be regarded as a 6-dimensional vector, and each parameter in each sample is a sample value randomly obtained according to a sampling rule.
Step three: the 60 samples were individually brought into the CAM mode, which selected F2000_ CAM5, ne30 resolution, and the root mean square error RMSE was calculated for the 60 samples individually using the ERA5 reanalysis dataset.
Step four: and constructing a Kriging proxy model F (x ^) by using the sample set S and the root mean square error set RMSE to obtain a fitting estimation relation of the parameters and the root mean square error.
Step five: the PSO particle swarm optimization algorithm is used for finding the optimal solution of the proxy model F (x ^) and the process is shown in figure 2 to obtain a group of optimal solutions which comprise a group of vectors with 6 dimensions, parameter values representing the optimal solution obtained by 6 parameters and RMSE estimated by the physical model F (x ^) when the parameters obtain the optimal solution.
And step six, the parameter values of the optimal solution obtained in the step five are re-introduced into the CAM5.3 to obtain a real simulation result, and the group of results and the root mean square error RMSE of the analysis data are calculated.
Step seven: and judging whether the result in the step six meets the optimization requirement, if not, respectively adding the parameters in the step six and the RMSE simulated by the CAM5.3 into the sample set S and the root-mean-square error set RMSE in the step four, updating the proxy model, and repeating the step four to the step six.
Step eight: and when the RMSE obtained in the step six meets the optimization standard, ending the tuning process, and outputting the parameters obtained at the moment, wherein the parameters are the final tuning realization result.
The embodiment of the invention provides an atmospheric mode optimization method based on a proxy model, and the invention discloses an atmospheric mode parameter tuning method based on the proxy model, which belongs to the field of design parameter optimization and comprises the following steps: determining a parameter range and sampling, simulating results of each sample by using an atmospheric mode, fitting a functional relation between the samples and the mode results by constructing an agent model by taking Root Mean Square Error (RMSE) as an evaluation standard of the output results of the atmospheric mode, searching an optimal value of the agent model by using a Particle Swarm Optimization (PSO) algorithm and bringing the optimal value into the atmospheric mode for verification, and continuously and iteratively updating the agent model until the optimization requirement is met and a final parameter optimization result is output. According to the invention, the idea of the proxy model is applied to parameter tuning of the atmospheric mode, so that the execution times of the atmospheric mode are effectively reduced, the resource overhead is reduced and the economic benefit is improved while the high-efficiency optimization algorithm is ensured; meanwhile, the time consumption of mode tuning is reduced, and the efficiency of atmospheric mode parameter tuning is improved.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (7)
1. An atmospheric mode optimization method based on a proxy model is characterized by comprising the following steps:
the method comprises the following steps: determining the optimized parameters and the optimized target, and determining the upper limit and the lower limit of the optimized parameters according to the selected parameters;
step two: sampling within a parameter range by utilizing Latin hypercube sampling;
step three: sequentially bringing the parameter samples obtained in the step two into a parameter input list of the atmospheric mode, starting to execute the atmospheric mode, reading an output file of the atmospheric mode after the atmospheric mode corresponding to each sampling value is executed, obtaining the change of different samples to an optimization target, and judging the change level of each sample to the mode by using a Root Mean Square Error (RMSE);
step four: constructing a proxy model by using the data obtained in the step three;
step five: searching an optimal solution of the proxy model by utilizing a particle swarm algorithm;
step six: bringing the parameters represented by the solutions in the step five back to an atmospheric mode to obtain a new group of values, and adding the new group of values into a solution set for constructing the agent model to update the agent model;
step seven: judging whether the result in the step six meets the optimization requirement, if not, respectively adding the parameters in the step six and the RMSE for the atmospheric mode simulation into the sample set S and the RMSE set in the step four, updating the proxy model, and repeating the step four to the step six;
step eight: and when the RMSE obtained in the step six meets the optimization standard, ending the tuning process, and outputting the parameters obtained at the moment, wherein the parameters are the final tuning realization result.
2. The proxy model-based atmospheric model optimization method of claim 1, wherein according to step two, the Latin hypercube sampling steps are as follows:
step 201: firstly, determining the number N of samples, namely the number of samples to be extracted;
step 202: equally dividing the interval (0, 1) into N sections;
step 203: randomly drawing a value in each of the N segments;
step 204: mapping the extracted value into a standard normal distribution sample through an inverse function of the standard normal distribution;
step 205: and (5) disordering the sampling sequence to obtain a final sampling result.
3. The proxy-model-based atmospheric model optimization method of claim 1, wherein according to step three, the RMSE is calculated by:
wherein N represents the result of the nth group of sampling, the value range is between 1 and N, m represents the total lattice number of the optimization area, yiRepresenting the result of the simulation of the object at the ith lattice pattern, yo,iRepresenting the value of the observed data at the ith grid point.
4. The proxy model-based atmospheric model optimization method of claim 1, wherein according to step four, the proxy model method is as follows:
401: the polynomial proxy model takes a second-order polynomial as an example, and the expression is as follows:
wherein, beta represents the coefficient to be estimated, and d is the number of parameters;
402: kriging's agent model, the expression is as follows:
y＝f(x)Tβl+zl(x)，l＝1，2，...，q
where β is a regression coefficient of a polynomial, the polynomial f (x) may be of any order, z (x) is a random process, the mean value E [ z (x) ] ═ 0;
403: the RBF agent model has the following expression:
where i-1 to n represent response values of the sample point i, ωiRepresents the ith sample point weight coefficient, riIs the Euclidean distance r between the point to be measured and the ith sample pointi＝∥y-yi∥，φ(ri) For the mirror function, the Guass function is commonly used: phi (r) ═ r2/c2Wherein c is a coefficient.
5. The proxy model-based atmospheric model optimization method of claim 1, wherein according to step five, the particle swarm algorithm comprises the following steps:
step 5.1: randomly initializing each particle;
step 5.2: calculating the adaptive value of each particle, wherein the adaptive value is a proxy model
The output result of (1);
step 5.3: acquiring an individual optimal value of the particle, and updating the individual optimal value of the particle if the adaptive value obtained by the particle in the step 5.2 is superior to the original optimal value of the particle;
step 5.4: acquiring a global optimum value of the particles, and if the adaptive value of the particles obtained in the step 5.3 is superior to the global optimum value, updating the global optimum value;
step 5.5: the velocity and position of each is updated as follows:
vi＝vi+c1*rand*(pbesti-xi)+c2*rand*(gbest-xi)
xi＝xi+vi
wherein v isiDenotes the velocity of the ith particle, c1And c2Is a self-learning factor and a group learning factor, and is two constants, rand is a random number between 0 and 1, pbestiIs the individual optimum of the ith particle, gbest is the global optimum of the particle, xiRepresenting the current position of the ith particle;
step 5.6: judging whether a convergence condition is met, if not, returning to the step 5.2;
step 5.7: and obtaining an optimal solution.
6. The proxy-model-based atmospheric-mode optimization method of claim 1, wherein according to step two, the hierarchical sampling is random hierarchical sampling.
7. The proxy-model-based atmospheric-pattern optimization method of claim 1, wherein the solution of the proxy model is an optimal solution according to step five.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111273554.5A CN113987806B (en) | 2021-10-29 | 2021-10-29 | Atmosphere mode optimization method based on proxy model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111273554.5A CN113987806B (en) | 2021-10-29 | 2021-10-29 | Atmosphere mode optimization method based on proxy model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113987806A true CN113987806A (en) | 2022-01-28 |
| CN113987806B CN113987806B (en) | 2024-04-26 |
Family
ID=79744539
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111273554.5A Active CN113987806B (en) | 2021-10-29 | 2021-10-29 | Atmosphere mode optimization method based on proxy model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113987806B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106886657A (en) * | 2017-03-15 | 2017-06-23 | 武汉理工大学 | A kind of FEM model method for building up based on kriging functions |
| WO2018072351A1 (en) * | 2016-10-20 | 2018-04-26 | 北京工业大学 | Method for optimizing support vector machine on basis of particle swarm optimization algorithm |
| CN108268728A (en) * | 2018-01-22 | 2018-07-10 | 上海交通大学 | Automobile tail gate structural optimization method based on two-step Modified particle swarm optimization algorithm |
| CN109783918A (en) * | 2019-01-04 | 2019-05-21 | 上海交通大学 | Based on the Gear Reducer Optimal Design implementation method for switching sequential sampling configuration |
| CN110008499A (en) * | 2019-01-21 | 2019-07-12 | 华南理工大学 | A kind of method of optimizing its structure based on Bayesian kriging model |
| AU2020103709A4 (en) * | 2020-11-26 | 2021-02-11 | Daqing Oilfield Design Institute Co., Ltd | A modified particle swarm intelligent optimization method for solving high-dimensional optimization problems of large oil and gas production systems |
| CN112784361A (en) * | 2021-01-25 | 2021-05-11 | 武汉理工大学 | Method for optimizing structure of automobile engine compartment heat dissipation system based on proxy model |
| CN113011076A (en) * | 2021-03-29 | 2021-06-22 | 西安理工大学 | Efficient particle swarm optimization method based on RBF proxy model |
| CN113139334A (en) * | 2021-04-07 | 2021-07-20 | 柳培忠 | Simulation optimization method based on bee colony |
-
2021
- 2021-10-29 CN CN202111273554.5A patent/CN113987806B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018072351A1 (en) * | 2016-10-20 | 2018-04-26 | 北京工业大学 | Method for optimizing support vector machine on basis of particle swarm optimization algorithm |
| CN106886657A (en) * | 2017-03-15 | 2017-06-23 | 武汉理工大学 | A kind of FEM model method for building up based on kriging functions |
| CN108268728A (en) * | 2018-01-22 | 2018-07-10 | 上海交通大学 | Automobile tail gate structural optimization method based on two-step Modified particle swarm optimization algorithm |
| CN109783918A (en) * | 2019-01-04 | 2019-05-21 | 上海交通大学 | Based on the Gear Reducer Optimal Design implementation method for switching sequential sampling configuration |
| CN110008499A (en) * | 2019-01-21 | 2019-07-12 | 华南理工大学 | A kind of method of optimizing its structure based on Bayesian kriging model |
| AU2020103709A4 (en) * | 2020-11-26 | 2021-02-11 | Daqing Oilfield Design Institute Co., Ltd | A modified particle swarm intelligent optimization method for solving high-dimensional optimization problems of large oil and gas production systems |
| CN112784361A (en) * | 2021-01-25 | 2021-05-11 | 武汉理工大学 | Method for optimizing structure of automobile engine compartment heat dissipation system based on proxy model |
| CN113011076A (en) * | 2021-03-29 | 2021-06-22 | 西安理工大学 | Efficient particle swarm optimization method based on RBF proxy model |
| CN113139334A (en) * | 2021-04-07 | 2021-07-20 | 柳培忠 | Simulation optimization method based on bee colony |
Non-Patent Citations (1)
| Title |
|---|
| 夏露;王丹;: "基于Kriging自适应代理模型的气动优化方法", 航空计算技术, no. 01, 15 January 2013 (2013-01-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113987806B (en) | 2024-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106022521B (en) | Short-term load prediction method of distributed BP neural network based on Hadoop architecture | |
| CN111814956B (en) | Multi-task learning air quality prediction method based on multi-dimensional secondary feature extraction | |
| CN110705029B (en) | Flow field prediction method of oscillating flapping wing energy acquisition system based on transfer learning | |
| CN111983927B (en) | Ellipsoid collective filtering method for maximum covariance MCC (MCC) criterion | |
| CN112446110B (en) | Application method of agent model based on EOASM algorithm in robot palletizer driving arm seat | |
| CN112733435A (en) | Whole vehicle size matching deviation prediction method based on multi-model fusion | |
| CN110442911B (en) | High-dimensional complex system uncertainty analysis method based on statistical machine learning | |
| CN112884236B (en) | Short-term load prediction method and system based on VDM decomposition and LSTM improvement | |
| CN111415010A (en) | Bayesian neural network-based wind turbine generator parameter identification method | |
| CN113777931B (en) | Icing wing type pneumatic model construction method, device, equipment and medium | |
| CN114564787A (en) | Bayesian optimization method, device and storage medium for target-related airfoil design | |
| CN114861304A (en) | Nonlinear aerodynamic force data rapid modeling method and system and storage medium | |
| CN112580855A (en) | Cable group steady-state temperature rise prediction method based on self-adaptive variation PSO-BP neural network | |
| CN113609763A (en) | Uncertainty-based satellite component layout temperature field prediction method | |
| CN111639388B (en) | Method and system for simulating parameters of elastic element of automobile | |
| CN113987806A (en) | Atmospheric mode optimization method based on proxy model | |
| CN116933386A (en) | Aircraft pneumatic data fusion method based on MCOK proxy model | |
| CN108362307B (en) | Method for determining principal component factors of star sensor on-orbit attitude measurement accuracy | |
| CN113722853B (en) | Group intelligent evolutionary engineering design constraint optimization method for intelligent computation | |
| CN116415177A (en) | Classifier parameter identification method based on extreme learning machine | |
| CN115391745A (en) | Rainfall forecast correction method and system based on probability matching average method | |
| CN114329320A (en) | Partial differential equation numerical solution method based on heuristic training data sampling | |
| CN109033678A (en) | A kind of aircraft near-optimal design method generated based on virtual sample | |
| CN117010260A (en) | Automatic history fit model prediction method, system and equipment for fractured reservoir | |
| CN107480381A (en) | The method of response surface model is built based on simulated annealing and applies its system |
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 |