CN111241751B

CN111241751B - Wing profile optimization method and device based on agent assisted evolution algorithm

Info

Publication number: CN111241751B
Application number: CN202010041514.7A
Authority: CN
Inventors: 吴巽锋; 刘群锋; 林秋镇; 陈剑勇
Original assignee: Shenzhen University
Current assignee: Shenzhen University
Priority date: 2020-01-15
Filing date: 2020-01-15
Publication date: 2022-03-22
Anticipated expiration: 2040-01-15
Also published as: WO2021142916A1; CN111241751A

Abstract

The invention discloses a wing profile optimization method, a wing profile optimization device, computer equipment and a storage medium based on a proxy assisted evolution algorithm.

Description

Wing profile optimization method and device based on agent assisted evolution algorithm

Technical Field

The invention relates to the technical field of data processing, in particular to an airfoil profile optimization method and device based on a proxy assisted evolution algorithm, computer equipment and a storage medium.

Background

The evolutionary algorithm is also called meta-heuristic algorithm, and is widely applied to various engineering optimizations at present. However, if these evolutionary algorithms are applied to expensive optimization problems involving computationally expensive simulations, the computational cost will be quite large. The expensive optimization problem is very different from the general optimization problem, and the response function of the expensive optimization problem is usually a simulation model, such as finite element analysis (abbreviated as FEA) and computational fluid dynamics (abbreviated as CFD).

At present, the proxy assisted evolution algorithm combining proxy assistance and evolution algorithm gets more and more attention in processing expensive optimization problems, that is, adding points to sample data points lacking in expensive problems to increase the total number of sample data points needs to be considered in an important way.

Common dotting criteria are the PoI dotting criterion (i.e., the one based on the probability of improvement), the ExI dotting criterion (i.e., the one based on the expected improvement) and the LCB dotting criterion (i.e., the one based on the lower confidence bound), which all effectively enhance the accuracy of the proxy model in the application of the proxy assistance and evolution algorithm, thereby speeding up the final convergence of the algorithm. However, the above three kinds of point adding criteria are simply using a single criterion or combining a plurality of criteria into a scalar criterion, which results in a poor effect of adding sample points in a limited data sample and a less improvement of the precision of the proxy model by the added sample points.

Disclosure of Invention

The embodiment of the invention provides an airfoil optimization method, a device, computer equipment and a storage medium based on a proxy assisted evolution algorithm, aiming at solving the problems that in the prior art, when a point adding criterion based on an improved probability, a point adding criterion based on a desired improvement or a point adding criterion based on a lower confidence bound is adopted to process the point adding in an expensive optimization problem, only a single criterion is simply used or a plurality of criteria are combined into a scalar criterion, so that the effect of adding sample points in limited data samples is poor, and the accuracy of the added sample points to a proxy model is improved less.

In a first aspect, an embodiment of the present invention provides an airfoil optimization method based on a proxy assisted evolution algorithm, which includes:

judging whether a data point adding request sent by a client is received;

if a data point adding request sent by a client is received, acquiring an initial data point set in a sample base, and acquiring the maximum real evaluation times according to the total number of data points in the initial data point set and the preset total number of optimization points; each data point comprises a decision variable corresponding to the wing geometric shape control point and an evaluation value corresponding to the decision variable, and each decision variable is an n-dimensional row vector or an n-dimensional column vector;

judging whether the total number of data points in the initial data point set is less than the maximum real evaluation times or not;

if the total number of the data points in the initial data point set is smaller than the maximum real evaluation times, taking each data point in the initial data point set as a training sample of a first proxy model to be trained to obtain a corresponding first current proxy model, and searching in a first final population generated according to the first initial population genetic evolution according to the first current proxy model and a preset first population screening condition to obtain a first target individual and a second target individual;

adding the data point corresponding to the first target individual and the data point corresponding to the second target individual into the initial data point set to obtain a current data point set; the data point corresponding to the first target individual consists of the first target individual and a first real function value obtained by inputting the first target individual into a pre-stored target function for operation and correspondence; the data point corresponding to the second target individual consists of the second target individual and a second real function value obtained by inputting the second target individual to the target function;

acquiring data points in the current data point set, sorting the data points in the current data point set in an ascending order according to a real function value and sorting the data points before a preset ranking threshold value to form a target data point set, taking the data points in the target data point set as training samples of a second proxy model to be trained to obtain a corresponding second current proxy model, and searching in a second final population generated according to the second initial population genetic evolution according to the second current proxy model and a preset second individual screening condition to obtain a third target individual;

adding the data points corresponding to the third target individual into the current data point set to obtain a final data point set, taking the final data point set as an initial data point set, and returning to execute the step of judging whether the total number of the data points in the initial data point set is smaller than the maximum real evaluation time; the data point corresponding to the third target individual consists of the third target individual and a third real function value obtained by inputting the third target individual to the target function correspondingly;

and if the total number of the data points in the initial data point set is greater than or equal to the maximum real evaluation times, sending the initial data point set to the client.

In a second aspect, the embodiment of the present invention provides an airfoil optimization device based on a proxy assisted evolution algorithm, which includes a unit for executing the airfoil optimization method based on the proxy assisted evolution algorithm in the first aspect.

In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the processor implements the wing profile optimization method based on the agent-assisted evolution algorithm according to the first aspect.

In a fourth aspect, the embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, causes the processor to execute the method for optimizing an airfoil profile based on a proxy assisted evolution algorithm according to the first aspect.

The embodiment of the invention provides a wing profile optimization method, a wing profile optimization device, computer equipment and a storage medium based on a proxy assisted evolution algorithm.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a schematic view of an application scenario of an airfoil optimization method based on a proxy assisted evolution algorithm according to an embodiment of the present invention;

FIG. 2 is a schematic flowchart of an airfoil optimization method based on a proxy assisted evolution algorithm according to an embodiment of the present invention;

FIG. 3 is a schematic sub-flow diagram of an airfoil optimization method based on a proxy assisted evolution algorithm according to an embodiment of the present invention;

FIG. 4 is a schematic view of another sub-flow of an airfoil optimization method based on a proxy assisted evolution algorithm according to an embodiment of the present invention;

FIG. 5 is a schematic view of another sub-flow of an airfoil optimization method based on a proxy assisted evolution algorithm according to an embodiment of the present invention;

FIG. 6 is a schematic block diagram of an airfoil optimization device based on a proxy assisted evolution algorithm according to an embodiment of the present invention;

FIG. 7 is a schematic block diagram of a computer device provided by an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

Referring to fig. 1 and fig. 2, fig. 1 is a schematic view of an application scenario of an airfoil optimization method based on a proxy assisted evolution algorithm according to an embodiment of the present invention; fig. 2 is a schematic flow chart of an airfoil optimization method based on a proxy assisted evolution algorithm, which is provided in an embodiment of the present invention and is applied to a server, where the method is executed by application software installed in the server.

As shown in fig. 2, the method includes steps S110 to S180.

S110, judging whether a data point adding request sent by the client side is received.

In order to more clearly understand the technical solution of the present application, the following describes the related terminal. The technical scheme is described in the perspective of a server.

The first is a client, which can be understood as a user terminal, the user terminal can be an electronic device with a communication function, such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, a wearable device, and the like, and the user terminal sends a data point adding request to a server.

And secondly, the server receives a data point adding request sent by the client, and selectively adds sample points to the initial data point set by combining the proxy model and the genetic evolution algorithm on the basis of the initial data point set stored in the server to obtain the data point set. And the server obtains a data point set and sends the data point set to the client.

In this embodiment, whether a data point adding request sent by a client is received or not is detected by a server, when the server receives a data point adding request sent by the client, the subsequent step S120 is executed, and when the server does not receive a path planning request sent by the client, the step S110 is executed again after waiting for a preset delay time.

S120, if a data point adding request sent by a client is received, acquiring an initial data point set in a sample base, and acquiring the maximum real evaluation times according to the total number of data points in the initial data point set and the preset total number of optimization points; each data point comprises a decision variable corresponding to the wing geometric shape control point and an evaluation value corresponding to the decision variable, and each decision variable is an n-dimensional row vector or an n-dimensional column vector.

In this embodiment, the initial total number of the initial data points in the initial data point set in the sample library is known (for example, 70 initial data points), each data point in the initial data point set includes a decision variable corresponding to the airfoil geometry control point, for example, 14 control points composed of a B-spline curve are used as the airfoil geometry control point, and the positions of the 14 control points constitute the decision variable corresponding to one data point in the initial data point set. And the evaluation value of each decision variable can be estimated by calculating as the following formula (1):

wherein, in formula (1), f_{Airfoil_j}Evaluation value, D, corresponding to decision variable representing jth data point of initial data point set_1jResistance, D, obtained by CFD simulation of decision variables representing the jth data point under preset fluid dynamics design condition 1_2jRepresents the resistance, L, obtained by CFD simulation of decision variables representing the jth data point under preset fluid dynamic design condition 2_1jA lift coefficient, L, obtained by CFD simulation of decision variables representing the jth data point under preset fluid dynamics design conditions 1_2jRepresenting a lift coefficient obtained by CFD simulation of decision variables representing the jth data point under preset fluid dynamics design conditions 2;

the decision variables representing the jth data point were CFD simulated under the preset hydrodynamic design condition 1 to yield the resistance of the baseline design,

the decision variables representing the jth data point were CFD simulated under the preset hydrodynamic design condition 2 to obtain the resistance of the baseline design,

CFD simulation is carried out on decision variables representing the jth data point under preset fluid dynamic design conditions 1 to obtain a lift coefficient of a baseline design,

and (4) carrying out CFD simulation on decision variables representing the jth data point under preset fluid dynamic design conditions 2 to obtain a lift coefficient of the baseline design.

And because the total number of the optimization points is preset (for example, the total number of the optimization points is 84), at this time, a maximum true evaluation time (for example, the maximum true evaluation time is 70+84 + 154) is obtained from the sum of the total number of the data points in the initial data point set and the preset total number of the optimization points, where the maximum true evaluation time is obtained in order to determine the iteration times of the subsequent steps S130-S170, so that a plurality of target data points are added to the initial data point set in the sample library, and the defect that the number of the data points is insufficient in the expensive optimization problem of the airfoil design of the airfoil is improved.

S130, judging whether the total number of the data points in the initial data point set is smaller than the maximum real evaluation times.

In this embodiment, since the initial total number of data points in the initial data point set is less than the maximum true evaluation time, the subsequent steps S130 to S170 need to be iterated multiple times until the initial data point set is sent to the client if the total number of data points in the initial data point set is greater than or equal to the maximum true evaluation time. And continuously adding data points with better performance to the initial data point set through an iterative process, thereby realizing data point addition of expensive optimization problems.

And S140, if the total number of the data points in the initial data point set is smaller than the maximum real evaluation times, taking each data point in the initial data point set as a training sample of a first to-be-trained proxy model to obtain a corresponding first current proxy model, and searching in a first final population generated according to the first initial population genetic evolution according to the first current proxy model and a preset first individual screening condition to obtain a first target individual and a second target individual.

In this embodiment, when the total number of the data points in the initial data point set is smaller than the maximum true evaluation number, a plurality of target data points generated by gradual genetic evolution based on the data points in the initial data point set are continuously added to the initial data point set.

In the wing-shaped expensive optimization problem of the wing, there must be more superior solution individuals in several sub-areas of the whole feasible region, so exploring the promising search area in the feasible region is the main objective of the global search stage (i.e. the main objective of step S140). In the global search phase, a first current proxy model is used as the proxy model in the global search phase. After the agent model is established, in the process that the evolution algorithm starts to search in the whole feasible domain, the agent model is used for obtaining the predicted value of the target individual, and meanwhile, the uncertainty value corresponding to each target individual is also obtained.

In this embodiment, the first to-be-trained proxy model includes a first to-be-trained kriging model, a first to-be-trained radial basis function model, and a first to-be-trained polynomial response surface model; the first current proxy model comprises a first Criss golden model, a first radial basis function model and a first polynomial response surface model.

The first Kriging model is a first Kriging model, the first radial basis function model to be trained is a first RBF model, and the first polynomial response surface model is a first PR model, and the three proxy models are existing proxy models, and the model expressions are not described herein again.

Since the first current proxy model includes the 3 proxy models, in order to improve the accuracy of the proxy model, the final predicted value input to the first current proxy model by a certain data point is a weighted sum of the predicted values respectively corresponding to the three models input by the data point.

In this embodiment, in step S140, taking each data point in the initial data point set as a training sample of the first to-be-trained proxy model to obtain a corresponding first current proxy model, including:

taking the decision variables of the data points in the initial data point set as the input of the first to-be-trained kriging model, taking the evaluation values corresponding to the decision variables as the output of the first to-be-trained kriging model, and training the first to-be-trained kriging model to obtain a first kriging model;

taking a decision variable of each data point in the initial data point set as an input of the first radial basis function model to be trained, taking an evaluation value corresponding to each decision variable as an output of the first radial basis function model to be trained, and training the first radial basis function model to be trained to obtain a first radial basis function model;

and taking the decision variables of the data points in the initial data point set as the input of the first polynomial response surface model to be trained, taking the evaluation values corresponding to the decision variables as the output of the first polynomial response surface model to be trained, and training the first polynomial response surface model to be trained to obtain the first polynomial response surface model.

In this embodiment, all data points included in the initial data point set are used as training samples of the first to-be-trained kriging model, the first to-be-trained radial basis function model, and the first to-be-trained polynomial response surface model, and the first kriging model, the first radial basis function model, and the first polynomial response surface model are obtained through training. Through the above process, the acquisition of the first current agent model is realized.

In this embodiment, as shown in fig. 3, the step S140 of searching in the first final population generated according to the first initial population genetic evolution according to the first current agent model and the preset first individual screening condition to obtain the first target individual and the second target individual includes:

s1401, randomly generating Ng variable solutions by a Latin hypercube design according to the vector characteristic dimension of the decision variables in the initial data point set to form a first-class initial population; each variable solution is an individual in the first-class initial population, and the feature dimension of each variable solution is the same as that of the decision variable;

s1402, obtaining a first type of current iteration algebra, and judging whether the first type of current iteration algebra reaches a preset maximum iteration algebra; wherein, the initial value of the first type current iteration algebra is 1;

s1403, if the current iteration algebra of the first type does not reach the maximum iteration algebra, performing simulated binary intersection and polynomial variation on the initial population of the first type to obtain a sub-population of the first type with the same total number of individuals as the initial population of the first type;

s1404, combining the first-class initial population and the first-class sub-population to obtain a first-class mixed population;

s1405, inputting each individual in the first-class mixed population into the first current agent model to obtain a predicted value corresponding to each individual in the first-class mixed population, and acquiring an uncertainty value corresponding to each individual according to the predicted value corresponding to each individual in the first-class mixed population;

s1406, sequencing each individual in the first-class mixed population in an ascending order according to the corresponding uncertainty value to obtain a first-class sequenced mixed population;

s1407, according to a preset grouping quantity Q, averagely dividing the mixed population after the first-class sorting to obtain Q groups of first-class sub-mixed populations; wherein Q ═ Ng;

s1408, respectively obtaining individuals with the minimum uncertainty value in each group of sub-mixed populations in the Q groups of first-class sub-mixed populations to form a first-class current population, and taking the first-class current population as a first-class initial population;

s1409, adding one to the first type current iteration algebra to be used as the first type current iteration algebra, and returning to execute the step S1402;

and S1410, if the current iteration algebra of the first type reaches the maximum iteration algebra, taking the initial population of the first type as the final population of the first type, obtaining the individual with the smallest predicted value in the final population of the first type as a first target individual, and obtaining the individual with the largest uncertainty value in the final population of the first type as a second target individual.

In this embodiment, since the vector feature dimension of the decision variable in the initial data point set is known, in order to generate a plurality of individuals with the same feature dimension as the vector feature dimension of the decision variable, Ng variable solutions (where Ng is a positive integer) can be randomly generated by a latin hypercube design with reference to the vector feature dimension of the decision variable.

The Latin hypercube design is Latin hypercube sampling (i.e. m samples are extracted in an n-dimensional vector space), and a plurality of variable solutions can be randomly generated by the method so as to form a first-class initial population. Wherein each variable solution is an individual in the first type of initial population, and the feature dimension of each variable solution is the same as the feature dimension of the decision variable. The multiple variable solutions are randomly generated by initialization, so that the target individuals with better performance are generated by evolution.

After the first-class initial population is obtained, the steps S1402-S1409 may be iteratively executed for a plurality of times until the first-class current iteration algebra reaches the maximum iteration algebra, the first-class initial population after the plurality of iterations is taken as the first-class final population, an individual with the smallest predicted value in the first-class final population is obtained as a first target individual, and an individual with the largest uncertainty value in the first-class final population is obtained as a second target individual.

In the population evolution process, simulated binary crossing and polynomial variation are adopted, the first-class sub-population is generated by randomly selecting two individuals from the first-class current population each time to perform simulated binary crossing until Ng first-class new individuals are obtained through crossing, then performing variation on the Ng first-class new individuals according to variation probability and polynomial variation to obtain Ng first-class new individuals after the polynomial variation, and the Ng first-class new individuals after the polynomial variation form the first-class sub-population. Here, the process of randomly selecting two individuals from the first-class current population for binary crossing a plurality of times is also similar to an iterative process, and the process of binary crossing a plurality of times is not stopped until the number of new individuals reaches the population size Ng corresponding to the first-class initial population. In addition, binary interleaving and polynomial mutation are conventional processes, and are not described herein again.

And after a first-class sub-population is obtained according to the first-class initial population and is mixed to obtain a first-class mixed population (the total number of individuals in the first-class mixed population is 2Ng), inputting each individual in the first-class mixed population into the first current agent model at the moment to obtain a predicted value corresponding to each individual in the first-class mixed population, and obtaining an uncertainty value corresponding to each individual according to the predicted value corresponding to each individual in the first-class mixed population. And acquiring a predicted value corresponding to each individual in the first-class mixed population and an uncertainty value corresponding to each individual, and taking the two values as reference parameter values for selecting a target data point (namely a target individual).

And then, sequencing each individual in the first-class mixed population in an ascending order according to the corresponding uncertainty value to obtain a first-class sequenced mixed population, and at the moment, dividing the first-class sequenced mixed population according to a preset grouping quantity Q, wherein the total number of the individuals in the first-class sequenced mixed population is 2Ng, so that the number of the individuals included in each group of first-class mixed populations in the divided Q groups of first-class mixed populations is 2 Ng/Q. Preferably, the value of Q is set to Ng, the number of individuals included in each group of the first-type sub-mixed population is 2, and at this time, an individual having the smallest uncertainty value is selected from the 2 individuals included in each group of the first-type sub-mixed population, so as to reconstruct the first-type current population including Ng individuals. And after one iteration, if the current iteration algebra of the first type does not reach the maximum iteration algebra after adding one, repeating the steps S1403-S1409 for a plurality of times until the current iteration algebra of the first type reaches the maximum iteration algebra, and taking the final initial population of the first type as the final population of the first type. At this time, the individual with the smallest predicted value in the first-class final population is obtained to serve as a first target individual, and the individual with the largest uncertainty value in the first-class final population is obtained to serve as a second target individual.

In an embodiment, as shown in fig. 4, in step S1405, inputting each individual in the first mixed population into the first current agent model, and obtaining a predicted value corresponding to each individual in the first mixed population includes:

s14051, inputting each individual in the first mixed population into the first Kreiskin model to obtain a first subtype predicted value corresponding to each individual in the first mixed population;

s14052, inputting each individual in the first mixed population into the first radial basis function model to obtain a second subtype predicted value corresponding to each individual in the first mixed population;

s14053, inputting each individual in the first mixed population into the first polynomial response surface model to obtain a third subtype predicted value corresponding to each individual in the first mixed population;

s14054, obtaining a first weight value corresponding to the first kriging model, obtaining a second weight value corresponding to the first radial basis function model, and obtaining a third weight value corresponding to the first polynomial response surface model;

s14055, calling a pre-stored predicted value weight summation model to obtain a predicted value corresponding to each individual in the first-class mixed population; the predicted value weight summation model is as follows:

wherein,

representing an individual x in said first mixed population_iCorresponding predicted values;

e₁is the root mean square error of the first Kreisen model, e₂Is the root mean square error of the first radial basis function model, e₃Is the root mean square error of the first polynomial response surface model,

representing an individual x_iIn response to the first sub-type predictor,

representing an individual x_iIn response to the second sub-type predictor,

representing an individual x_iCorresponding to the third subtype predictor.

In this embodiment, when calculating the predicted value corresponding to each individual in the first-class mixed population, the first sub-type predicted value, the second sub-type predicted value, and the third sub-type predicted value corresponding to the individual are obtained after the individual is respectively input to the first kriging model, the first radial basis function model, and the polynomial response surface model. At this time, the first subtype predicted value, the second subtype predicted value and the third subtype predicted value corresponding to the individual are weighted and summed to obtain the predicted value corresponding to the individual, and the specific calculation process refers to step S14055. And one mode of weighted summation is adopted, so that the accuracy of the predicted value obtained by inputting the individual into the first current agent model is higher, and the individual with better performance is conveniently screened out as a target individual by taking the predicted value as a standard.

In an embodiment, the obtaining an uncertainty value corresponding to each individual according to the predicted value corresponding to each individual in the first-class mixed population in step S1405 includes:

repeatedly executing the acquisition of the individual x in the first mixed population_iCorresponding first subtype prediction value

Second subtype prediction value

And third subtype prediction value

The maximum difference between every two is taken as an individual x in the first mixed population_iCorresponding uncertainty value U_ens(x_i) Until obtaining the uncertainty value corresponding to each individual in the first type mixed population; wherein the value range of i is [1,2Ng]。

In this embodiment, when obtaining the uncertainty value corresponding to each individual in the first-class mixed population, one of the individuals x is obtained₁For example, individual x is obtained first₁Corresponding first subtype prediction value

Second subtype prediction value

And third subtype prediction value

Then obtain again

Maximum value in between, as the individual x₁Corresponding uncertainty values (uncertainty is defined as the maximum difference between two predicted values), the process of obtaining uncertainty values for other individuals in the first mixed population being referred to the above exemplified individual x₁The uncertainty value of (2). And acquiring uncertainty values corresponding to all the individuals in the first-class mixed population, so that individuals with better performance can be screened out as target individuals by taking the uncertainty values as standards.

S150, adding the data point corresponding to the first target individual and the data point corresponding to the second target individual into the initial data point set to obtain a current data point set; the data point corresponding to the first target individual consists of the first target individual and a first real function value obtained by inputting the first target individual into a pre-stored target function for operation and correspondence; and the data point corresponding to the second target individual consists of the second target individual and a second real function value obtained by inputting the second target individual into the target function correspondingly.

In this embodiment, each data point in the initial data point set includes a decision variable and an evaluation value corresponding to the decision variable, so after the first target individual and the second target individual are obtained, a first true function value obtained by inputting the first target individual to a pre-stored target function for operation and corresponding thereto and a second true function value obtained by inputting the second target individual to the target function are calculated, the data point corresponding to the first target individual is composed of the first target individual and the first true function value, and the data point corresponding to the second target individual is composed of the second target individual and the second true function value.

After the data points corresponding to the first target individual and the data points corresponding to the second target individual are added to the initial data point set, if the total number of the data points is still less than the maximum real evaluation time, step S160 and the following steps are executed until the total number of the data points in the initial data point set is greater than or equal to the maximum real evaluation time, and the initial data point set is sent to the client. Through the steps S140-S150, the first target individual and the second target individual are obtained by fast screening the first-class final population, and the data points corresponding to the two individuals can be used as two selected data points after the iteration is finished and added to the initial data point set to obtain the current data point set.

And S160, acquiring data points in the current data point set, sorting the data points according to the real function values in an ascending order, sorting the data points before a preset ranking threshold value to form a target data point set, taking the data points in the target data point set as training samples of a second proxy model to be trained to obtain a corresponding second current proxy model, and searching in a second final population generated according to the second initial population genetic evolution according to the second current proxy model and a preset second individual screening condition to obtain a third target individual.

In this embodiment, after the first round of data point adding process of the initial data point set is completed through global search, the second round of data point adding process of the current data point set is completed through local search, so as to obtain a final data point set after the round of iterative process is completed.

In one embodiment, as shown in fig. 5, step S160 includes:

s1601, calling a prestored second radial basis function model to be trained to serve as the second proxy model to be trained;

s1602, taking the decision variables corresponding to the data points in the target data point set as the input of the second radial basis function model to be trained, taking the evaluation values corresponding to the decision variables as the output of the second radial basis function model to be trained, training the second radial basis function model to be trained to obtain a second radial basis function model, and taking the second radial basis function model as the second current proxy model;

s1603, obtaining decision variables corresponding to each data point in the target data point set to form a second type of initial population; wherein the decision variable corresponding to each data point in the target data point set corresponds to an individual in the second type of initial population;

s1604, obtaining a second type of current iteration algebra, and judging whether the second type of current iteration algebra reaches a preset maximum iteration algebra; wherein the initial value of the second type of current iteration algebra is 1;

s1605, if the second type current iteration algebra does not reach the maximum iteration algebra, carrying out simulated binary intersection and polynomial variation on the second type initial population to obtain a second type sub-population with the same total number of individuals as the second type initial population;

s1606, merging the second type initial population and the second type sub population to obtain a second type mixed population;

s1607, inputting each individual in the second type of mixed population into the second radial basis function model to obtain a predicted value corresponding to each individual in the second type of mixed population;

s1608, sorting each individual in the second mixed population in an ascending order according to the corresponding predicted value to obtain a sorted second mixed population;

s1609, obtaining the individuals in the sorted second mixed population before the ranking threshold value to form a second current population, and taking the second current population as a second initial population;

s1610, adding one to the second type current iteration algebra to serve as the second type current iteration algebra, and returning to execute the step S1604;

s1611, if the second type current iteration algebra reaches the maximum iteration algebra, taking the second type initial population as the second type final population, and obtaining an individual with the smallest predicted value in the second type final population as a third target individual.

In this embodiment, a third target individual satisfying the second individual screening condition is searched in the current data point set by means of local search.

In the specific process of obtaining the third target individual, the data points in the current data point set are firstly obtained and sorted in an ascending order according to the real function value, the data points are sorted before a preset ranking threshold value to form a target data point set, and then the decision variables corresponding to the data points in the target data point set are used for forming a second type initial population.

After the second-class initial population is obtained, steps S1604 to S1611 may be iteratively executed for multiple times until the second-class current iteration algebra reaches the maximum iteration algebra, the second-class initial population after multiple iterations is used as the second-class final population, and an individual with the smallest predicted value in the second-class final population is obtained as a third target individual.

In one embodiment, step S1605 includes:

randomly selecting two individuals from the second initial population to perform binary crossing in sequence until M new cross-processed second individuals are generated, performing polynomial variation on the M new cross-processed second new individuals, and forming a second sub-population by the second new cross-processed second new individuals; wherein M is the ranking threshold-1.

In the population evolution process, simulated binary crossing and polynomial variation are adopted, the generation of the second-class sub-population is to randomly select two individuals from the second-class current population each time to perform simulated binary crossing until M new individuals of the second class (wherein the value of M is a positive integer) after cross processing are obtained through crossing, then the M new individuals of the second class after cross processing are varied according to variation probability and polynomial variation to obtain M new individuals of the second class after polynomial variation, and the M new individuals of the second class after polynomial variation form the second-class sub-population. Here, the process of randomly selecting two individuals from the second type of current population for binary crossing a plurality of times is also similar to an iterative process, and the process of binary crossing a plurality of times is not stopped until the number of new individuals reaches the size M of the second population corresponding to the second type of initial population.

And after a second-class sub-population is obtained according to the second-class initial population and is mixed to obtain a second-class mixed population (the total number of individuals in the second-class mixed population is 2M), inputting each individual in the second-class mixed population into the second current agent model at the moment, and obtaining a predicted value corresponding to each individual in the second-class mixed population. And acquiring a predicted value corresponding to each individual in the second-class mixed population, wherein the predicted value is also convenient to be used as a reference parameter value for selecting a target data point (namely a target individual).

And then, sequencing each individual in the second-class mixed population in an ascending order according to the corresponding predicted value to obtain a second-class sequenced mixed population, and at the moment, obtaining the individual sequenced before the ranking threshold value from the second-class sequenced mixed population to form a second-class current population so as to form a second-class current population including M individuals again. After one iteration is performed on the second-type initial population, if the second-type current iteration algebra is added by one and does not reach the maximum iteration algebra, the steps S1604 to S1611 are repeatedly executed for many times until the second-type current iteration algebra reaches the maximum iteration algebra, and the final second-type initial population is used as the second-type final population. At this time, the individual with the smallest predicted value in the second-class final population is obtained as a third target individual.

S170, adding the data points corresponding to the third target individual into the current data point set to obtain a final data point set, taking the final data point set as an initial data point set, and returning to execute the step S130; and the data point corresponding to the third target individual consists of the third target individual and a third real function value obtained by inputting the third target individual to the target function correspondingly.

In this embodiment, after the data points corresponding to the third target individual are added to the current data point set, if the total number of the data points is still less than the maximum true evaluation time, the step is executed and the step S120 and the subsequent steps are executed again until the total number of the data points in the initial data point set is greater than or equal to the maximum true evaluation time, and the initial data point set is sent to the client. Through steps S160-S170, a third target individual is obtained by quickly screening the second-class final population, a data point corresponding to the third target individual may be added to the current data point set as a selected data point after the iteration is completed, a final data point set is obtained, the final data point set is updated after the iteration is completed as a new initial data point set, and the step S120 is executed again.

And S180, if the total number of the data points in the initial data point set is greater than or equal to the maximum real evaluation times, sending the initial data point set to the client.

In this embodiment, after the initial data point set is obtained in the server, the initial data point set can be sent to the client. The client can further perform wing-shaped optimization on the wing according to more data points in the initial data point set of the final state after multiple iterations.

The method realizes the mode of combining the agent assistance and the evolution algorithm and simultaneously considering the predicted value and the uncertainty of the agent model, quickly increases data points in a limited data sample, and improves the precision of the agent model by the increased sample points.

The embodiment of the invention also provides an airfoil optimization device based on the agent assisted evolution algorithm, which is used for executing any embodiment of the airfoil optimization method based on the agent assisted evolution algorithm. Specifically, referring to fig. 6, fig. 6 is a schematic block diagram of an airfoil optimization device based on a proxy assisted evolution algorithm according to an embodiment of the present invention. The wing profile optimization device 100 based on the agent assisted evolution algorithm can be configured in a server.

As shown in fig. 6, the airfoil profile optimization device 100 based on the agent-aided evolution algorithm includes a dotting request detection unit 110, an initial data point set acquisition unit 120, a total number of data points determination unit 130, a global search unit 140, a first round of dotting unit 150, a local search unit 160, a second round of dotting unit 170, and a post-dotting set sending unit 180.

The adding point request detecting unit 110 is configured to determine whether a data point adding point request sent by a client is received.

An initial data point set obtaining unit 120, configured to obtain an initial data point set in the sample base if a data point adding request sent by the client is received, and obtain the maximum true evaluation time according to the total number of data points in the initial data point set and the preset total number of optimized points; each data point comprises a decision variable corresponding to the wing geometric shape control point and an evaluation value corresponding to the decision variable, and each decision variable is an n-dimensional row vector or an n-dimensional column vector.

A total data point number determining unit 130, configured to determine whether the total number of data points in the initial data point set is smaller than the maximum true evaluation time.

And the global search unit 140 is configured to, if the total number of data points in the initial data point set is smaller than the maximum true evaluation times, obtain a corresponding first current agent model by using each data point in the initial data point set as a training sample of a first agent model to be trained, and search a first type of final population generated according to the first type of initial population genetic evolution according to the first current agent model and a preset first individual screening condition to obtain a first target individual and a second target individual.

In this embodiment, the global search unit 140 includes:

a first proxy model training unit, configured to use a decision variable of each data point in the initial data point set as an input of the first kriging model to be trained, use an evaluation value corresponding to each decision variable as an output of the first kriging model to be trained, and train the first kriging model to be trained to obtain a first kriging model;

a second proxy model training unit, configured to use a decision variable of each data point in the initial data point set as an input of the first radial basis function model to be trained, use an evaluation value corresponding to each decision variable as an output of the first radial basis function model to be trained, and train the first radial basis function model to be trained to obtain a first radial basis function model;

and the third rational model training unit is used for taking the decision variables of the data points in the initial data point set as the input of the first polynomial response surface model to be trained, taking the evaluation values corresponding to the decision variables as the output of the first polynomial response surface model to be trained, and training the first polynomial response surface model to be trained to obtain the first polynomial response surface model.

In this embodiment, the global search unit 140 further includes:

a first-class initial population generating unit, configured to randomly generate Ng variable solutions by a latin hypercube design according to the vector feature dimension of the decision variable in the initial data point set, so as to form a first-class initial population; each variable solution is an individual in the first-class initial population, and the feature dimension of each variable solution is the same as that of the decision variable;

the first-class current iteration algebra judging unit is used for acquiring a first-class current iteration algebra and judging whether the first-class current iteration algebra reaches a preset maximum iteration algebra; wherein, the initial value of the first type current iteration algebra is 1;

the first-class population cross variation unit is used for carrying out analog binary cross and polynomial variation on the first-class initial population to obtain a first-class sub-population with the same total number of individuals as the first-class initial population if the first-class current iteration algebra does not reach the maximum iteration algebra;

a first-class mixed population obtaining unit, configured to combine the first-class initial population and the first-class sub-population to obtain a first-class mixed population;

a first-class parameter obtaining unit, configured to input each individual in the first-class mixed population into the first current proxy model, obtain a predicted value corresponding to each individual in the first-class mixed population, and obtain an uncertainty value corresponding to each individual according to the predicted value corresponding to each individual in the first-class mixed population;

the first-class sorting unit is used for sorting each individual in the first-class mixed population in an ascending order according to the corresponding uncertainty value to obtain a first-class sorted mixed population;

the first-class population dividing unit is used for averagely dividing the mixed population after the first class is sorted according to a preset grouping quantity Q to obtain Q groups of first-class sub-mixed populations; wherein Q ═ Ng;

the first-class population screening unit is used for respectively obtaining individuals with the minimum uncertainty value in each group of sub-mixed populations in Q groups of first-class sub-mixed populations to form a first-class current population, and using the first-class current population as a first-class initial population;

the first-class current iteration algebra secondary judgment unit is used for adding one to the first-class current iteration algebra to serve as a first-class current iteration algebra, returning to execute the step of obtaining the first-class current iteration algebra and judging whether the first-class current iteration algebra reaches a preset maximum iteration algebra;

and the first-class target individual obtaining unit is used for taking the first-class initial population as the first-class final population if the first-class current iteration algebra reaches the maximum iteration algebra, obtaining an individual with the smallest predicted value in the first-class final population as a first target individual, and obtaining an individual with the largest uncertainty value in the first-class final population as a second target individual.

In an embodiment, the first-type parameter obtaining unit includes:

a first subtype prediction value obtaining unit, configured to input each individual in the first type mixed population into the first kriging model, so as to obtain a first subtype prediction value corresponding to each individual in the first type mixed population;

a second subtype prediction value obtaining unit, configured to input each individual in the first-class mixed population into the first radial basis function model, so as to obtain a second subtype prediction value corresponding to each individual in the first-class mixed population;

a third sub-type prediction value obtaining unit, configured to input each individual in the first type mixed population into the first polynomial response surface model, so as to obtain a third sub-type prediction value corresponding to each individual in the first type mixed population;

the weight value obtaining unit is used for obtaining a first weight value corresponding to the first kriging model, obtaining a second weight value corresponding to the first radial basis function model, and obtaining a third weight value corresponding to the first polynomial response surface model;

the weight summation unit is used for calling a pre-stored predicted value weight summation model to obtain a predicted value corresponding to each individual in the first-class mixed population; the predicted value weight summation model is as follows:

wherein,

representing an individual x_iIn response to the first sub-type predictor,

representing an individual x_iIn response to the second sub-type predictor,

representing an individual x_iCorresponding to the third subtype predictor.

In an embodiment, the first population cross mutation unit is further configured to:

Second subtype prediction value

And third subtype prediction value

A first round adding unit 150, configured to add the data point corresponding to the first target individual and the data point corresponding to the second target individual to the initial data point set, so as to obtain a current data point set; the data point corresponding to the first target individual consists of the first target individual and a first real function value obtained by inputting the first target individual into a pre-stored target function for operation and correspondence; and the data point corresponding to the second target individual consists of the second target individual and a second real function value obtained by inputting the second target individual into the target function correspondingly.

And the local search unit 160 is configured to acquire data points in the current data point set, sort the data points in the current data point set in an ascending order according to the real function value and before a preset ranking threshold to form a target data point set, obtain a corresponding second current proxy model by using each data point in the target data point set as a training sample of a second proxy model to be trained, and search a second final population generated according to the second initial population genetic evolution according to the second current proxy model and a preset second individual screening condition to obtain a third target individual.

In one embodiment, the local search unit 160 includes:

the fourth agent model obtaining unit is used for calling a prestored second radial basis function model to be trained to serve as the second agent model to be trained;

a fourth agent model training unit, configured to use a decision variable corresponding to each data point in the target data point set as an input of the second radial basis function model to be trained, use an evaluation value corresponding to each decision variable as an output of the second radial basis function model to be trained, train the second radial basis function model to be trained, obtain a second radial basis function model, and use the second radial basis function model as the second current agent model;

a second-class initial population generating unit, configured to obtain decision variables corresponding to data points in the target data point set to form a second-class initial population; wherein the decision variable corresponding to each data point in the target data point set corresponds to an individual in the second type of initial population;

the second type current iteration algebra judging unit is used for acquiring a second type current iteration algebra and judging whether the second type current iteration algebra reaches a preset maximum iteration algebra; wherein the initial value of the second type of current iteration algebra is 1;

a second type population cross variation unit, configured to perform simulated binary cross and polynomial variation on the second type initial population if the second type current iteration algebra does not reach the maximum iteration algebra, to obtain a second type sub-population having the same total number of individuals as the second type initial population;

a second-type mixed population obtaining unit, configured to combine the second-type initial population and the second-type sub-population to obtain a second-type mixed population;

a second-class parameter obtaining unit, configured to input each individual in the second-class mixed population into the second radial basis function model, so as to obtain a predicted value corresponding to each individual in the second-class mixed population;

the second-class sorting unit is used for sorting each individual in the second-class mixed population in an ascending order according to the corresponding predicted value to obtain a sorted second-class mixed population;

a second-class population screening unit, configured to obtain individuals sorted before the ranking threshold in the sorted second-class mixed population to form a second-class current population, and use the second-class current population as a second-class initial population;

a second-class current iteration algebra secondary judgment unit, configured to add one to the second-class current iteration algebra to serve as a second-class current iteration algebra, return to the step of executing the step of obtaining the second-class current iteration algebra, and judge whether the second-class current iteration algebra reaches a preset maximum iteration algebra;

and the second-class target individual obtaining unit is used for taking the second-class initial population as the second-class final population and obtaining an individual with the minimum predicted value in the second-class final population as a third target individual if the second-class current iteration algebra reaches the maximum iteration algebra.

A second round adding unit 170, configured to add the data point corresponding to the third target individual to the current data point set to obtain a final data point set, use the final data point set as an initial data point set, and return to perform the step of determining whether the total number of data points in the initial data point set is less than the maximum true evaluation number; and the data point corresponding to the third target individual consists of the third target individual and a third real function value obtained by inputting the third target individual to the target function correspondingly.

And a post-dotting set sending unit 180, configured to send the initial data point set to the client if the total number of data points in the initial data point set is greater than or equal to the maximum true evaluation number.

The device realizes the mode of combining the agent assistance and the evolution algorithm and simultaneously considering the predicted value and the uncertainty of the agent model, quickly increases data points in a limited data sample, and improves the precision of the agent model by the increased sample points.

The wing profile optimization device based on the agent-assisted evolution algorithm can be implemented in the form of a computer program which can be run on a computer device as shown in fig. 7.

Referring to fig. 7, fig. 7 is a schematic block diagram of a computer device according to an embodiment of the present invention. The computer device 500 is a server, and the server may be an independent server or a server cluster composed of a plurality of servers.

Referring to fig. 7, the computer device 500 includes a processor 502, memory, and a network interface 505 connected by a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.

The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032, when executed, may cause the processor 502 to perform a method of profile optimization based on a proxy assisted evolution algorithm.

The processor 502 is used to provide computing and control capabilities that support the operation of the overall computer device 500.

The internal memory 504 provides an environment for running the computer program 5032 in the non-volatile storage medium 503, and when the computer program 5032 is executed by the processor 502, the processor 502 may be caused to perform a method for profile optimization based on a proxy assisted evolution algorithm.

The network interface 505 is used for network communication, such as providing transmission of data information. Those skilled in the art will appreciate that the configuration shown in fig. 7 is a block diagram of only a portion of the configuration associated with aspects of the present invention and is not intended to limit the computing device 500 to which aspects of the present invention may be applied, and that a particular computing device 500 may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

The processor 502 is configured to run a computer program 5032 stored in the memory to implement the method for optimizing the profile based on the agent assisted evolution algorithm disclosed in the embodiment of the present invention.

Those skilled in the art will appreciate that the embodiment of a computer device illustrated in fig. 7 does not constitute a limitation on the specific construction of the computer device, and that in other embodiments a computer device may include more or fewer components than those illustrated, or some components may be combined, or a different arrangement of components. For example, in some embodiments, the computer device may only include a memory and a processor, and in such embodiments, the structures and functions of the memory and the processor are consistent with those of the embodiment shown in fig. 7, and are not described herein again.

It should be understood that, in the embodiment of the present invention, the Processor 502 may be a Central Processing Unit (CPU), and the Processor 502 may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components, and the like. Wherein a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

In another embodiment of the invention, a computer-readable storage medium is provided. The computer readable storage medium may be a non-volatile computer readable storage medium. The computer readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the method for profile optimization based on the agent assisted evolution algorithm disclosed in the embodiments of the present invention.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses, devices and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described in a functional general in the foregoing description for the purpose of illustrating clearly the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

In the embodiments provided by the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only a logical division, and there may be other divisions when the actual implementation is performed, or units having the same function may be grouped into one unit, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a storage medium. Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a magnetic disk, or an optical disk.

While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A wing profile optimization method based on a proxy assisted evolution algorithm is characterized by comprising the following steps:

judging whether a data point adding request sent by a client is received;

adding the data points corresponding to the third target individual into the current data point set to obtain a final data point set, taking the final data point set as an initial data point set, and returning to execute the step of judging whether the total number of the data points in the initial data point set is smaller than the maximum real evaluation time; the data point corresponding to the third target individual consists of the third target individual and a third real function value obtained by inputting the third target individual to the target function correspondingly; and

if the total number of data points in the initial data point set is greater than or equal to the maximum real evaluation times, sending the initial data point set to the client;

the first to-be-trained agent model comprises a first to-be-trained Kriging model, a first to-be-trained radial basis function model and a first to-be-trained polynomial response surface model; the first current proxy model comprises a first Crisi gold model, a first radial basis function model and a first polynomial response surface model;

the obtaining of the corresponding first current agent model by using each data point in the initial data point set as a training sample of the first agent model to be trained includes:

taking the decision variables of the data points in the initial data point set as the input of the first polynomial response surface model to be trained, taking the evaluation values corresponding to the decision variables as the output of the first polynomial response surface model to be trained, and training the first polynomial response surface model to be trained to obtain a first polynomial response surface model;

searching in a first type final population generated according to the first type initial population genetic evolution according to the first current agent model and a preset first individual screening condition to obtain a first target individual and a second target individual, wherein the searching comprises the following steps:

according to the vector characteristic dimension of the decision variable in the initial data point set, designing and randomly generating Ng variable solutions by using a Latin hypercube so as to form a first-class initial population; each variable solution is an individual in the first-class initial population, and the feature dimension of each variable solution is the same as that of the decision variable;

acquiring a first type of current iteration algebra, and judging whether the first type of current iteration algebra reaches a preset maximum iteration algebra; wherein, the initial value of the first type current iteration algebra is 1;

if the current iteration algebra of the first type does not reach the maximum iteration algebra, performing simulated binary intersection and polynomial variation on the initial population of the first type to obtain a sub-population of the first type with the same total number of individuals as the initial population of the first type;

combining the first type initial population and the first type sub population to obtain a first type mixed population;

inputting each individual in the first-class mixed population into the first current agent model to obtain a predicted value corresponding to each individual in the first-class mixed population, and acquiring an uncertainty value corresponding to each individual according to the predicted value corresponding to each individual in the first-class mixed population;

sequencing each individual in the first-class mixed population in an ascending order according to the corresponding uncertainty value to obtain a first-class sequenced mixed population;

according to a preset grouping quantity Q, averagely dividing the mixed population after the first-class sorting to obtain Q groups of first-class sub-mixed populations; wherein Q ═ Ng;

respectively obtaining individuals with the uncertainty value as the minimum value in each group of sub-mixed populations in Q groups of first-class sub-mixed populations to form a first-class current population, and taking the first-class current population as a first-class initial population;

adding one to the first type current iteration algebra to serve as a first type current iteration algebra, and returning to the step of judging whether the first type current iteration algebra reaches a preset maximum iteration algebra;

if the first type current iteration algebra reaches the maximum iteration algebra, the first type initial population is used as the first type final population, the individual with the smallest predicted value in the first type final population is obtained to be used as a first target individual, and the individual with the largest uncertainty value in the first type final population is obtained to be used as a second target individual.

2. The method for wing profile optimization based on the agent assisted evolution algorithm according to claim 1, wherein the inputting each individual in the first mixed population into the first current agent model to obtain a predicted value corresponding to each individual in the first mixed population comprises:

inputting each individual in the first mixed population into the first kriging model to obtain a first subtype predicted value corresponding to each individual in the first mixed population;

inputting each individual in the first type mixed population into the first radial basis function model to obtain a second subtype predicted value corresponding to each individual in the first type mixed population;

inputting each individual in the first type mixed population into the first polynomial response surface model to obtain a third subtype predicted value corresponding to each individual in the first type mixed population;

acquiring a first weight value corresponding to the first kriging model, acquiring a second weight value corresponding to the first radial basis function model, and acquiring a third weight value corresponding to the first polynomial response surface model;

calling a pre-stored predicted value weight summation model to obtain a predicted value corresponding to each individual in the first-class mixed population; the predicted value weight summation model is as follows:

wherein,

representing an individual x_iIn response to the first sub-type predictor,

representing an individual x_iIn response to the second sub-type predictor,

representing an individual x_iCorresponding to the third subtype predictor.

3. The method for optimizing wing profiles based on the agent assisted evolution algorithm according to claim 2, wherein the obtaining of the uncertainty value corresponding to each individual according to the predicted value corresponding to each individual in the first mixed population comprises:

Second subtype prediction value

And third subtype prediction value

4. The wing profile optimization method based on the agent-aided evolution algorithm according to claim 1, wherein the step of obtaining a corresponding second current agent model by using each data point in the target data point set as a training sample of a second agent model to be trained, and the step of searching in a second final population generated according to the genetic evolution of the second initial population according to the second current agent model and a preset second individual screening condition to obtain a third target individual comprises the steps of:

calling a prestored second radial basis function model to be trained to serve as the second proxy model to be trained;

taking a decision variable corresponding to each data point in the target data point set as an input of the second radial basis function model to be trained, taking an evaluation value corresponding to each decision variable as an output of the second radial basis function model to be trained, training the second radial basis function model to be trained to obtain a second radial basis function model, and taking the second radial basis function model as the second current agent model;

obtaining decision variables corresponding to the data points in the target data point set to form a second type of initial population; wherein the decision variable corresponding to each data point in the target data point set corresponds to an individual in the second type of initial population;

acquiring a second type of current iteration algebra, and judging whether the second type of current iteration algebra reaches a preset maximum iteration algebra; wherein the initial value of the second type of current iteration algebra is 1;

if the second type of current iteration algebra does not reach the maximum iteration algebra, performing simulated binary intersection and polynomial variation on the second type of initial population to obtain a second type of sub-population with the same total number of individuals as the second type of initial population;

merging the second type initial population and the second type sub population to obtain a second type mixed population;

inputting each individual in the second mixed population into the second radial basis function model to obtain a predicted value corresponding to each individual in the second mixed population;

sequencing each individual in the second mixed population in an ascending order according to the corresponding predicted value to obtain a sequenced second mixed population;

obtaining individuals sorted before the ranking threshold value in the sorted second mixed population to form a second current population, and taking the second current population as a second initial population;

adding one to the second type current iteration algebra to serve as the second type current iteration algebra, and returning to the step of judging whether the second type current iteration algebra reaches the preset maximum iteration algebra or not;

and if the second type current iteration algebra reaches the maximum iteration algebra, taking the second type initial population as the second type final population, and acquiring an individual with the minimum predicted value in the second type final population as a third target individual.

5. The method for optimizing wing profiles based on the agent assisted evolution algorithm according to claim 4, wherein the performing simulated binary cross and polynomial variation on the second type initial population to obtain a second type sub-population having the same total number of individuals as the second type initial population comprises:

6. An airfoil optimization device based on a proxy assisted evolution algorithm, characterized by comprising a unit for executing the airfoil optimization method based on the proxy assisted evolution algorithm according to any one of claims 1 to 5.

7. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method for wing profile optimization based on a proxy assisted evolution algorithm according to any of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a processor, causes the processor to carry out the method of wing profile optimization based on a proxy assisted evolution algorithm according to any one of claims 1 to 5.