CN117973215A - Intelligent optimization method for double-wall cooling structure based on reinforcement learning - Google Patents
Intelligent optimization method for double-wall cooling structure based on reinforcement learning Download PDFInfo
- Publication number
- CN117973215A CN117973215A CN202410178806.3A CN202410178806A CN117973215A CN 117973215 A CN117973215 A CN 117973215A CN 202410178806 A CN202410178806 A CN 202410178806A CN 117973215 A CN117973215 A CN 117973215A
- Authority
- CN
- China
- Prior art keywords
- double
- cooling structure
- wall cooling
- action
- optimization
- 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.)
- Pending
Links
- 238000001816 cooling Methods 0.000 title claims abstract description 112
- 238000005457 optimization Methods 0.000 title claims abstract description 83
- 230000002787 reinforcement Effects 0.000 title claims abstract description 37
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000009471 action Effects 0.000 claims abstract description 111
- 239000000112 cooling gas Substances 0.000 claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000011156 evaluation Methods 0.000 claims description 19
- 230000008859 change Effects 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 8
- 238000004088 simulation Methods 0.000 claims description 7
- 230000000694 effects Effects 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000000547 structure data Methods 0.000 claims description 5
- 230000001186 cumulative effect Effects 0.000 claims description 4
- 238000002474 experimental method Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000013461 design Methods 0.000 abstract description 17
- 238000011160 research Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 239000003795 chemical substances by application Substances 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000007789 gas Substances 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007664 blowing Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Landscapes
- Turbine Rotor Nozzle Sealing (AREA)
Abstract
The invention relates to the technical field of aviation cooling system design, in particular to an intelligent optimization method of a double-wall cooling structure based on reinforcement learning, which solves the technical problems of how to quickly obtain geometric parameters of an optimized structure under the initial condition of a given double-wall cooling structure and realize a given optimization target in the prior art, and specifically comprises the following steps: parameterizing a double-wall cooling structure; defining reinforcement learning elements of the parameterized double-wall cooling structure under an optimized background; generating a plurality of data sets of optimization experience with respect to the double-wall cooling structure; building and training an intelligent model for optimizing decision-making; using the action decision network to assist the optimization of the double-wall cooling structure; according to the technical scheme, under the conditions of given initial geometric parameters of the double-wall cooling structure, working conditions corresponding to the initial structure, cooling gas usage amount and temperature field corresponding to the initial structure and optimization targets, the optimized geometric parameters meeting the optimization targets are rapidly given within a few milliseconds.
Description
Technical Field
The invention relates to the technical field of aviation cooling system design, in particular to an intelligent optimization method of a double-wall cooling structure based on reinforcement learning.
Background
Aero-engines are important in the modern aviation industry, and increasing the temperature before the turbine of an aero-engine is an effective means for increasing the thrust and improving the thermal efficiency of the aero-engine;
Under the current research trend, the temperature level before the turbine of the advanced aeroengine is far higher than the temperature resistant level of the manufacturing materials, and the gap is still increasing, so that the development of efficient blade cooling technology is particularly important; in the development process of cooling technology, the design of the double-wall turbine blade cooling structure depends on the characteristics of high-efficiency heat exchange property, high reliability, easiness in processing and manufacturing and the like, and is widely focused and studied.
The evaluation standard of the double-wall cooling structure studied at present is generally divided into two aspects of cooling gas usage amount and blade temperature field;
On the one hand, the cooling of the turbine blades is achieved by extracting the low temperature gases from the fan or compressor, which are heat exchanged through cooling structures on the blades; the more cooling gas used, the less gas used to generate thrust, the greater the thrust loss caused;
On the other hand, the temperature and uniformity of the turbine blade directly affect the blade life; the double-wall cooling structure is optimally designed, and an optimal set of design parameter combination is required to be searched in a parameter change space, so that the high-efficiency cooling target of more uniform and lower surface temperature level is achieved under the condition of using the least amount of cooling gas; this has an important role in ensuring blade life and reducing engine thrust losses due to bleed air.
In the aspect of the optimal design of the double-wall cooling structure, a great deal of research is carried out in the prior literature;
The paper name "study of the effect of changing cooling arrangement and wall thickness on improving full-coverage cooling efficiency of double-wall structures" authors Li Weigong, lu Xunfeng, li Xueying, ren Jing, jiang Hongde; journal of papers is journal of heat transfer, volume 2019,141, stage 4: 042201]; in the paper, the cooling performance of the double-wall cooling structure is compared by experimental means when the length-diameter ratio of the air film holes is 1.0 and 2.5, and the research shows that when the length-diameter ratio of the air film holes is 1.0, the researched structure has higher cooling efficiency under the same blowing ratio;
Paper name "irregular turbulent column shape optimizing based on deep learning method" [ authors Yang Li, wang Qi, rao Yu; the journal of paper is International journal of thermal science, volume 2021,161: 106746. In the paper, a Pix2Pix generation countermeasure network is used for establishing a predictive model of an external temperature field and a middle section pressure field of a double-wall internal turbulent flow column channel structure, and a genetic algorithm is adopted to perform structural optimization on the basis of a predictive agent model for training convergence, so that the optimized structure has better cooling performance.
Most of the existing cooling structure optimization design methods adopt the idea of combining a proxy model with an optimizing algorithm; however, such a method has repeated optimization work when the design conditions are changed, and the optimization efficiency is low. At present, a more advanced optimization method capable of establishing design conditions to optimize a forward calculation path of design parameters exists. The method learns decision information in historical data through reinforcement learning, and the trained model has the capability of quickly optimizing decision when any initial design condition is given. In particular to a paper of a wind blade torsion angle distribution optimization searching method based on reinforcement learning [ authors Gu Liangyue, hao Jia, john Hall and the like; the journal is an energy source, 2021,215 volumes; in the article, a fully-connected neural network for predicting aerodynamic performance parameters of the wind turbine blade is established by utilizing numerical simulation data and is used as an online learning environment for reinforcement learning. And then, establishing an intelligent model for quickly searching the torsion angle distribution of the optimal wind turbine blade under different wind speed conditions by adopting a reinforcement learning algorithm. Through verification, the intelligent model can quickly obtain the optimal torsion angle distribution in 0.1 seconds when the wind speed changes.
As shown by the research results of the prior literature, the reinforcement learning method has advantages in realizing high-efficiency structural optimization; however, reinforcement learning has not been applied to optimization of double-wall cooling structures for aircraft turbine blades; in the existing research of structural optimization by reinforcement learning, the cold air flow and physical field calculation results obtained in the initial design round are not fully utilized; meanwhile, the existing research has not solved the problem of how to obtain the optimization result rapidly when the optimization target changes.
Disclosure of Invention
The invention provides an intelligent optimization method of a double-wall cooling structure based on reinforcement learning, which aims to solve the technical problems of how to quickly obtain the geometric parameters of the optimized structure under the initial condition of the given double-wall cooling structure and realize the given optimization target in the prior art.
In order to solve the technical problems, the technical scheme of the invention is as follows:
an intelligent optimization method of a double-wall cooling structure based on reinforcement learning comprises the following steps:
step S1, parameterizing a double-wall cooling structure;
Parameterizing a double-wall cooling structure to be optimized to obtain geometric parameters, working condition parameters, temperature and cooling gas usage evaluation parameters and optimization target description parameters describing the double-wall cooling structure so as to obtain the parameterized double-wall cooling structure;
step S2, defining reinforcement learning elements of the parameterized double-wall cooling structure under an optimized background;
The definition of the reinforcement learning element includes: status, actions, smart models, environments, rewards, and policies;
the state is defined as a geometric parameter describing a double-wall cooling structure, a working condition parameter, a temperature and cooling gas usage evaluation parameter and an optimization target description parameter;
the motion is the geometric parameter change amount of the double-wall cooling structure;
The intelligent model is an established action decision network for predicting the optimal action by the state and an action scoring network for predicting the maximum possible accumulated rewards by the state and the action;
The environment is an offline numerical simulation or experiment environment;
The rewards are gains of cooling effects before and after executing actions;
the strategy is to select the preferable action corresponding to the highest accumulated rewards;
step S3, generating a plurality of data sets about decision experience of the double-wall cooling structure;
calling a double-wall cooling structure data set generation module on the basis of parameterization of the double-wall cooling structure in the step S1 and clear definition of reinforcement learning state, action, intelligent model, environment, rewards and strategy in the step S2;
the double-wall cooling structure data set generation module is used for obtaining a large number of data sets about decision experience of the double-wall cooling structure;
The data set is in the form of a current state, an executed action, a later state reached by the executed action and an instant reward obtained by the executed action;
s4, building and training an intelligent model for optimizing decision-making;
in step S3, on the basis of the data set generation, an optimization decision intelligent model building and training module is called to obtain a training convergence action decision network for predicting the optimal action;
s5, using the action decision network to assist in optimizing the double-wall cooling structure;
Placing the action decision network obtained in the step S4 in a double-layer wall cooling structure optimization module;
And when the initial geometric parameters, the working condition parameters, the temperature and the cooling gas consumption evaluation parameters and the optimization target description parameters of the double-wall cooling structure are known, invoking the double-wall cooling structure optimization module to obtain the optimized geometric parameter values of the double-wall cooling structure.
Specifically, in step S3, there is a sampling process, specifically: changing the geometric parameters of the initial structure, the working condition parameters, the executed actions, optimizing the target description parameters, and calculating even rewards according to the temperature before and after the execution of the actions and the evaluation parameter values of the cooling gas usage;
And repeating the sampling process, and processing the result into a data set in the form of the current state, the executed action, the last state reached by the executed action and the instant rewards obtained by the executed action.
Specifically, in step S4, an action decision network and an action scoring network are established as intelligent models;
the input of the action decision network is a parameter representing the state, and the output is the optimal action predicted by the network;
the inputs of the action scoring network are states and actions, and the output is a scoring value of the maximum cumulative rewards available for the actions.
Specifically, in step S4, training the intelligent model with the data set in the experience playback buffer based on the offline reinforcement learning rationale;
after training convergence, the action decision network is saved so as to be utilized in the optimization of the double-wall cooling structure;
the input of step S4 is a data set in the form of a current state, an executed action, a later state reached by the executed action, and an instant prize obtained by the executed action, and the output is an action decision network for predicting an optimal action, which is converged by training.
The invention has the following beneficial effects:
(1) According to the technical scheme, the optimized geometric parameters meeting the optimization target can be rapidly and accurately given under the conditions of the given initial geometric parameters of the double-wall cooling structure, the working conditions corresponding to the initial structure, the cooling gas usage amount and the temperature field corresponding to the initial structure and the optimization target.
(2) In the traditional optimization method based on the agent model and the optimizing algorithm, the agent model for evaluating the cooling effect is required to be established for realizing a given optimization target by the double-layer wall cooling structure, and then the optimized parameters are obtained by reversely searching by utilizing the optimizing algorithm. However, such a method has repeated optimization works when the design conditions are changed, and the optimization efficiency is low. The current more advanced method is a structure optimization method based on reinforcement learning, and the method can establish a forward calculation path from a design condition to an optimized design parameter, and has higher optimization efficiency.
(3) Reinforcement learning has not been applied to optimization of a double-wall cooling structure of a turbine blade at present; in the existing research of structural optimization by reinforcement learning, the cold air flow and physical field calculation results obtained in the initial design round are not fully utilized; meanwhile, the existing research has not solved the problem of how to obtain the optimization result rapidly when the optimization target changes.
(4) The technical scheme provided by the invention is used for constructing and training the decision-making intelligent model through the reinforcement learning algorithm, so that the problems of fully utilizing the cold air flow and the physical field calculation result obtained in the initial design round and rapidly obtaining the optimization result when the optimization target changes are solved; the method provided by the invention can rapidly and accurately give the optimized geometric parameters meeting the optimization target within a few milliseconds under the conditions of giving the initial geometric parameters of the double-wall cooling structure, the working conditions corresponding to the initial structure, the cooling gas usage amount and the temperature field corresponding to the initial structure and the optimization target.
Drawings
The invention is described in further detail below with reference to the drawings and the detailed description.
FIG. 1 is a schematic diagram of a double-wall cooling structure;
FIG. 2 is a schematic diagram of the definition of reinforcement learning elements in the context of the optimization of the double-wall cooling structure of the present invention;
FIG. 3 is a schematic diagram of the construction and training of the action decision network and action scoring network of the present invention;
FIG. 4 is an optimized schematic diagram of a dual wall cooling structure embodying the present invention;
Fig. 5 is a flow chart of the method of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention mainly solves the problems that: under the initial conditions of a given double-wall cooling structure, such as initial geometric parameters, working conditions corresponding to the initial structure, cooling gas usage amount and temperature field corresponding to the initial structure, and optimization targets, how to quickly obtain the geometric parameters of the optimized structure, and realize the given optimization targets.
In order to solve the problems, the invention provides an intelligent optimization method for a double-wall cooling structure of a turbine blade based on reinforcement learning.
Firstly, parameterizing a double-wall cooling structure to be optimized to obtain geometric parameters, working condition parameters, temperature and cooling gas usage evaluation parameters and optimization target description parameters for describing the double-wall cooling structure.
Next, in the context of double-wall cooling structure optimization, explicitly defined states, actions, smart models, environments, rewards, and policies are given to reinforcement learning: the state comprises geometric parameters describing a double-wall cooling structure, working condition parameters, temperature and cooling gas usage evaluation parameters and optimization target description parameters;
the optimization target description parameters comprise description parameters such as optimization weights between the cold air flow and the temperature, average temperature expected values and the like;
Action is a geometric parameter change amount;
The intelligent model is an action decision network which is defined and built by a designer and predicts the optimal action (the change amount of the optimal geometric parameters) by the state and an action scoring network which predicts the most possible accumulated rewards by the state and the action;
The environment is an offline numerical simulation or experiment environment;
rewards are gains of cooling effects before and after performing actions;
The policy is to select the action corresponding to the highest cumulative prize as the optimal action.
And then, giving geometric parameters and working condition parameters of the initial structure, and carrying out numerical simulation calculation on the initial design to obtain evaluation parameter values of the temperature and the cooling gas consumption.
Giving the action (geometric parameter change) to be executed, obtaining the structural geometric parameter and working condition parameter after the action is executed, and carrying out numerical simulation calculation on the structure to obtain the temperature and cooling gas usage evaluation parameter values.
Changing the geometric parameters of the initial structure, the working condition parameters, the actions to be executed and the optimization target description parameters, and calculating even rewards according to the temperature before and after executing the actions and the cooling gas usage evaluation parameter values. And repeating the sampling process, and processing the result into a data set in the form of the current state, the executed action, the last state reached by the executed action and the instant rewards obtained by the executed action.
The data sets are stored in a designer's experience playback buffer allocated by the computer for experience playback during subsequent training.
And then, training the built intelligent model through a data set based on an offline reinforcement learning basic principle to obtain the intelligent model capable of accurately outputting the optimal geometric parameter change amount through inputting an initial state.
The trained intelligent model can be used for the rapid optimization of the double-layer wall cooling structure of the turbine blade.
According to the invention, under the conditions of given initial geometric parameters of the double-wall cooling structure, working conditions corresponding to the initial structure, cooling gas usage amount and temperature field corresponding to the initial structure and optimization targets, the optimized geometric parameters meeting the optimization targets can be rapidly and accurately given within a few milliseconds.
Specifically, referring to fig. 1-5, the method comprises the following main processes and contents:
Step S1, parameterizing a double-wall cooling structure;
Parameterizing the double-wall cooling structure to be optimized to obtain geometric parameters describing the double-wall cooling structure, working condition parameters, temperature and cooling gas usage evaluation parameters and optimization target description parameters. The input is a double-wall cooling structure to be optimized, and the output is an evaluation parameter for describing the geometric parameter, the working condition parameter, the temperature and the cooling gas consumption of the double-wall cooling structure and an optimization target description parameter.
(II) definition of reinforcement learning elements under the optimization background of the double-wall cooling structure, namely, step S2;
On the basis of parameterization of the double-wall cooling structure in the step S1, the state, action, intelligent model, environment, rewards and strategy of reinforcement learning under the background of double-wall cooling structure optimization are given clear definition. The state is defined as a geometric parameter describing a double-wall cooling structure, a working condition parameter, a temperature and cooling gas usage evaluation parameter and an optimization target description parameter; action is a geometric parameter change amount; the intelligent model is an action decision network which is defined and built by a designer and predicts the optimal action (the change amount of the optimal geometric parameters) by the state and an action scoring network which predicts the most possible accumulated rewards by the state and the action; the environment is an offline numerical simulation or experiment environment; rewards are gains of cooling effects before and after performing actions; the policy is to select the action corresponding to the highest cumulative prize as the optimal action.
(III) generating a plurality of data sets about decision experience of the double-wall cooling structure, namely, step S3;
And (2) on the basis of parameterizing the double-wall cooling structure in the step S1 and giving clear definition to the reinforcement learning state, action, intelligent model, environment, rewards and strategies in the step 2, calling a double-wall cooling structure data set generating module to obtain a large number of data sets about decision experience of the double-wall cooling structure, wherein the data sets are in the form of current state, action to be executed, the latter state reached by the action to be executed and instant rewards obtained by the action to be executed. The geometric parameters, working condition parameters and the change range of the optimization target description parameters, which are specified by a designer, are input, and the output is a data set in the form of a current state, an executed action, a later state reached by the executed action and an instant rewards obtained by the executed action.
(IV) building and training an intelligent model for optimizing decision-making, namely, step S4;
And on the basis of generating the data set in the step S3, calling an optimization decision intelligent model building and training module to obtain a training convergence action decision network for predicting the optimal action. The input is a data set in the form of a current state, an executed action, a later state reached by the executed action, and instant rewards obtained by the executed action, and the output is an action decision network which is used for predicting the optimal action and is converged by training.
(5) The action decision network assists the optimization of the double-wall cooling structure, namely, step S4;
And (3) placing the action decision network obtained in the step (S4) in a double-wall cooling structure optimization module. And when the initial geometric parameters, the working condition parameters, the temperature and the cooling gas consumption evaluation parameters and the optimization target description parameters of the double-wall cooling structure are known, calling a double-wall cooling structure optimization module to obtain optimized geometric parameter values of the double-wall cooling structure. The input is a trained action decision network, the initial geometric parameters of the double-wall cooling structure, the working condition parameters, the temperature and cooling gas usage evaluation parameters and the optimization target description parameters are known, and the output is the geometric parameter value of the double-wall cooling structure after optimization.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.
Claims (4)
1. The intelligent optimization method of the double-wall cooling structure based on reinforcement learning is characterized by comprising the following steps of:
step S1, parameterizing a double-wall cooling structure;
Parameterizing a double-wall cooling structure to be optimized to obtain geometric parameters, working condition parameters, temperature and cooling gas usage evaluation parameters and optimization target description parameters describing the double-wall cooling structure so as to obtain the parameterized double-wall cooling structure;
step S2, defining reinforcement learning elements of the parameterized double-wall cooling structure under an optimized background;
The definition of the reinforcement learning element includes: status, actions, smart models, environments, rewards, and policies;
the state is defined as a geometric parameter describing a double-wall cooling structure, a working condition parameter, a temperature and cooling gas usage evaluation parameter and an optimization target description parameter;
the motion is the geometric parameter change amount of the double-wall cooling structure;
The intelligent model is an established action decision network for predicting the optimal action by the state and an action scoring network for predicting the maximum possible accumulated rewards by the state and the action;
The environment is an offline numerical simulation or experiment environment;
The rewards are gains of cooling effects before and after executing actions;
the strategy is to select the preferable action corresponding to the highest accumulated rewards;
step S3, generating a plurality of data sets about decision experience of the double-wall cooling structure;
calling a double-wall cooling structure data set generation module on the basis of parameterization of the double-wall cooling structure in the step S1 and clear definition of reinforcement learning state, action, intelligent model, environment, rewards and strategy in the step S2;
the double-wall cooling structure data set generation module is used for obtaining a large number of data sets about decision experience of the double-wall cooling structure;
The data set is in the form of a current state, an executed action, a later state reached by the executed action and an instant reward obtained by the executed action;
s4, building and training an intelligent model for optimizing decision-making;
in step S3, on the basis of the data set generation, an optimization decision intelligent model building and training module is called to obtain a training convergence action decision network for predicting the optimal action;
s5, using the action decision network to assist in optimizing the double-wall cooling structure;
Placing the action decision network obtained in the step S4 in a double-layer wall cooling structure optimization module;
And when the initial geometric parameters, the working condition parameters, the temperature and the cooling gas consumption evaluation parameters and the optimization target description parameters of the double-wall cooling structure are known, invoking the double-wall cooling structure optimization module to obtain the optimized geometric parameter values of the double-wall cooling structure.
2. The reinforcement learning-based intelligent optimization method for a double-wall cooling structure according to claim 1, wherein in step S3, there is a sampling process, and the sampling process specifically refers to: changing the geometric parameters of the initial structure, the working condition parameters, the executed actions, optimizing the target description parameters, and calculating even rewards according to the temperature before and after the execution of the actions and the evaluation parameter values of the cooling gas usage;
And repeating the sampling process, and processing the result into a data set in the form of the current state, the executed action, the last state reached by the executed action and the instant rewards obtained by the executed action.
3. The reinforcement learning-based intelligent optimization method for a double-wall cooling structure according to claim 2, wherein in step S4, an action decision network and an action scoring network are established as intelligent models;
the input of the action decision network is a parameter representing the state, and the output is the optimal action predicted by the network;
the inputs of the action scoring network are states and actions, and the output is a scoring value of the maximum cumulative rewards available for the actions.
4. The reinforcement learning-based double-wall cooling structure intelligent optimization method according to claim 3, wherein in step S4, the intelligent model is trained using the data set in the experience playback buffer based on the offline reinforcement learning basic principle;
after training convergence, the action decision network is saved so as to be utilized in the optimization of the double-wall cooling structure;
the input of step S4 is a data set in the form of a current state, an executed action, a later state reached by the executed action, and an instant prize obtained by the executed action, and the output is an action decision network for predicting an optimal action, which is converged by training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410178806.3A CN117973215A (en) | 2024-02-16 | 2024-02-16 | Intelligent optimization method for double-wall cooling structure based on reinforcement learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410178806.3A CN117973215A (en) | 2024-02-16 | 2024-02-16 | Intelligent optimization method for double-wall cooling structure based on reinforcement learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117973215A true CN117973215A (en) | 2024-05-03 |
Family
ID=90857848
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410178806.3A Pending CN117973215A (en) | 2024-02-16 | 2024-02-16 | Intelligent optimization method for double-wall cooling structure based on reinforcement learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117973215A (en) |
-
2024
- 2024-02-16 CN CN202410178806.3A patent/CN117973215A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108153998B (en) | Full three-dimensional pneumatic optimization design method for centrifugal blower impeller | |
| CN106777482A (en) | A kind of structure Multidisciplinary design optimization method based on mesh parameterization | |
| CN111898212B (en) | Impeller mechanical profile design optimization method based on BezierGAN and Bayesian optimization | |
| CN113361047B (en) | Impeller high-dimensional optimization and model selection method for multilayer sensing structure | |
| CN113158355B (en) | All-condition optimization design method for low-temperature liquid expander | |
| CN113268000B (en) | Soft switching method for multi-model predictive control of aircraft engine | |
| Jasa et al. | Wind turbine blade design with airfoil shape control using invertible neural networks | |
| CN117648763A (en) | CST-EGO multi-parameter optimization design method of synergistic jet wing profile | |
| CN117973215A (en) | Intelligent optimization method for double-wall cooling structure based on reinforcement learning | |
| Li et al. | Optimization design of a compressor cascade airfoil using a Navier-stokes solver and genetic algorithms | |
| CN113294297B (en) | Variable weight adjusting method for wind turbine generator nonlinear model prediction torque control | |
| CN113036769B (en) | Static voltage stability analysis method and system for power system | |
| CN112052621B (en) | Supercritical carbon dioxide impeller machinery supercritical region prediction and control method based on CNN | |
| CN114545928A (en) | Heterogeneous vehicle queue control method based on self-triggering distributed predictive control | |
| Guo et al. | An efficient surrogate-assisted differential evolution algorithm for turbomachinery cascades optimization with more than 100 variables | |
| CN112632728A (en) | Turbine mechanical blade profile design and performance prediction method based on deep learning | |
| Zheng et al. | A Research on Aero-engine Control Based on Deep Q Learning | |
| Song et al. | Knowledge-based aero-thermal multi-disciplinary design optimization of a high temperature blade | |
| WO2022252206A1 (en) | Aero-engine surge active control system based on fuzzy switching of controllers | |
| CN117131636B (en) | Data analysis-based multi-body turbine blade position efficiency optimization method | |
| CN112925201B (en) | Model prediction control-based fuel control method for ship propulsion system | |
| Yang et al. | Deep learning-based total impulse prediction method in ignition process of solid rocket motor | |
| Zhao | Optimization Design of Electrical Axial Fan Based on NSGA-II Algorithm | |
| Goriachkin et al. | Method for Improving Multi-stage Compressor Parameters using the IOSO Optimizer. | |
| CN117874998A (en) | Collaborative design method for gradient distribution and braiding structure of thermophysical parameters of CMC (CMC) material |
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 |