Turbine blade cooling structure adjusting and optimizing method and device based on reinforcement learning

Abstract

The invention provides a turbine blade cooling structure optimization method and device based on reinforcement learning, and the method comprises the steps: responding to an input preset design condition, obtaining an initial cooling structure of a turbine blade based on the design condition, responding to the initialization of a preset intelligent agent, and carrying out the optimization of the cooling structure of the turbine blade under a preset reinforcement learning environment framework, the intelligent agent is trained based on a preset reinforcement learning algorithm to obtain a trained intelligent agent, the cooling parameters of the cooling structure are adjusted and optimized based on the trained intelligent agent to obtain an adjusted and optimized cooling parameter, and the adjusted and optimized cooling structure is determined based on the adjusted and optimized cooling parameter. According to the method and the device, a more complex or complete cooling structure can be applied, the cooling parameters of the cooling structure can be automatically adjusted, optimized and determined, and the experience dependence on designers and the time cost in the cooling parameter design process are reduced.

Description

Technical Field

[0001] The present application relates to the field of deep learning technology, and more specifically, to a method and device for tuning a turbine blade cooling structure based on reinforcement learning. Background Art

[0002] In the field of aero engines, turbine blade cooling structures are key structures, and the design of turbine blade cooling structures is very complex and time-consuming, and the design effect is highly dependent on the professional experience of designers. In addition, in order to design a set of parameters that meet the requirements, designers often need to adjust the design parameters multiple times, and then repeat complex steps such as blade modeling, meshing, and simulation calculations, and evaluate whether this set of design parameters is qualified based on the obtained temperature field values. It usually takes designers more than ten days to get a set of parameters that meet the design conditions. However, these parameters affect each other, and in traditional design methods, a high degree of professional experience and knowledge is required to optimize the cooling structure.

[0003] At present, in order to solve the cooling structure optimization problem, the turbine blade cooling structure is often tuned by using heuristic algorithms.

[0004] However, this method is only applicable to simple or local cooling structures, and during the implementation process, designers not only need to iteratively modify the basic structure and cooling structure parameters, but also need to continuously perform simulation calculations to evaluate the cooling effect. This takes a lot of time to search for a better solution and is highly dependent on the designer's experience. Summary of the invention

[0005] In view of this, the purpose of the present application is to provide a method and device for tuning the cooling structure of a turbine blade based on reinforcement learning. By introducing a reinforcement learning algorithm to train an intelligent agent in the design process of the turbine blade cooling structure, and tuning the cooling parameters of the cooling structure according to the trained intelligent agent, a more complex or complete cooling structure can be applied, thereby realizing automatic tuning and determination of the cooling parameters of the cooling structure, and reducing the reliance on the experience of designers and the time cost in the cooling parameter design process.

[0006] In a first aspect, an embodiment of the present application provides a turbine blade cooling structure tuning method based on reinforcement learning, the turbine blade cooling structure tuning method comprising:

[0007] In response to inputting preset design conditions, an initial cooling structure of the turbine blade is obtained based on the design conditions; the design conditions include a blade model configuration file and an operating condition;

[0008] In response to the completion of the preset intelligent agent initialization, the intelligent agent is trained based on a preset reinforcement learning algorithm in a preset reinforcement learning environment framework to obtain a trained intelligent agent; wherein the reinforcement learning algorithm corresponds to a strategy network;

[0009] The cooling parameters of the cooling structure are tuned based on the trained intelligent agent to obtain the tuned cooling parameters, and the tuned cooling structure is determined based on the tuned cooling parameters.

[0010] In a possible implementation manner, in response to inputting a preset design condition, obtaining an initial cooling structure of the turbine blade based on the design condition includes:

[0011] In response to inputting preset design conditions, performing external flow channel simulation calculation based on the design conditions to generate an initial basic structure of the turbine blade;

[0012] A corresponding initial cooling structure is obtained based on the initial basic structure of the turbine blade; wherein the basic structure is used to determine basic parameters of the cooling structure of the turbine blade; and the cooling structure is used to determine cooling parameters of the cooling structure of the turbine blade.

[0013] In a possible implementation, before training the preset intelligent agent based on a preset strategy network in the preset reinforcement learning environment framework, the intelligent agent is initialized by the following steps:

[0014] In a preset reinforcement learning environment framework, the agent resets the environment and reads corresponding hole parameter information and working conditions based on the blade model configuration file, the working conditions and a preset hole parameter file; wherein the hole parameter file includes cooling parameters of the cooling structure;

[0015] The hole parameter information and operating conditions are processed based on a preset full three-dimensional evaluation algorithm to obtain an initial first turbine blade surface temperature after the environment is reset.

[0016] In a possible implementation, the step of training the agent to obtain a trained agent includes:

[0017] The intelligent agent inputs the surface temperature of the first turbine blade and the cooling parameters of the cooling structure read from the hole parameter file into the strategy network to generate a corresponding target action, and hands the target action over to the reinforcement learning environment framework for implementation, so that the reinforcement learning environment framework adjusts the cooling parameters based on the target action;

[0018] In response to the completion of the parameter adjustment of the cooling parameter, the adjusted cooling parameter and the operating condition are calculated based on the full three-dimensional evaluation algorithm to obtain the adjusted surface temperature of the second turbine blade to complete a round of training for the intelligent agent;

[0019] In response to receiving a termination signal generated by a preset reward function being false, a new round of training of the agent is continued based on the surface temperature of the second turbine blade and the adjusted cooling parameters to finally obtain a trained agent.

[0020] In a possible implementation, the method further includes:

[0021] In response to receiving a termination signal generated by a preset reward function being false, the agent updates the policy network based on the adjusted cooling parameter and the second turbine blade surface temperature;

[0022] In response to the updated policy network not converging, the training is terminated.

[0023] In a possible implementation, the step of optimizing the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the optimized cooling parameters includes:

[0024] Under the reinforcement learning algorithm, based on the trained agent and the updated policy network, and based on the cooling parameters and basic parameters corresponding to the basic structure, an initial third turbine blade surface temperature is obtained under the reinforcement learning environment framework;

[0025] generating a corresponding second target action based on the third turbine blade surface temperature and the cooling parameter, and delivering the second target action to the reinforcement learning environment framework for implementation, so as to tune the cooling parameter of the cooling structure, and obtain a fourth turbine blade surface temperature and the tuned cooling parameter;

[0026] The change of the surface temperature of the turbine blade is evaluated based on the surface temperature of the fourth turbine blade, and the optimized cooling parameters are determined based on the evaluation result.

[0027] In a possible implementation manner, the evaluating the change of the surface temperature of the turbine blade based on the fourth turbine blade surface temperature, and determining the optimized cooling parameter based on the evaluation result, includes:

[0028] In response to determining that the turbine blade surface temperature rises or the cooling parameter exceeds a preset parameter boundary during the implementation of the second target action based on the fourth turbine blade surface temperature and the initial third turbine blade surface temperature, the agent accumulates the second target action to obtain total modification information for the cooling structure;

[0029] The cooling parameters are modified based on the total modification information to obtain final optimized cooling parameters.

[0030] In a second aspect, an embodiment of the present application further provides a turbine blade cooling structure tuning device based on reinforcement learning, the turbine blade cooling structure tuning device comprising:

[0031] An acquisition module, configured to obtain an initial cooling structure of a turbine blade based on an input preset design condition in response to the input preset design condition; the design condition includes a blade model configuration file and an operating condition;

[0032] A training module, configured to, in response to the completion of the initialization of the preset intelligent agent, train the intelligent agent based on a preset reinforcement learning algorithm in a preset reinforcement learning environment framework to obtain a trained intelligent agent; wherein the reinforcement learning algorithm corresponds to a policy network;

[0033] A tuning module is used to tune the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the tuned cooling parameters, and determine the tuned cooling structure based on the tuned cooling parameters.

[0034] In a possible implementation, the acquisition module is specifically configured to:

[0035] In response to inputting preset design conditions, performing external flow channel simulation calculation based on the design conditions to generate an initial basic structure of the turbine blade;

[0036] A corresponding initial cooling structure is obtained based on the initial basic structure of the turbine blade; wherein the basic structure is used to determine basic parameters of the cooling structure of the turbine blade; and the cooling structure is used to determine cooling parameters of the cooling structure of the turbine blade.

[0037] In a possible implementation, the agent is initialized by the following steps:

[0038] Before the preset agent is trained based on the preset strategy network under the preset reinforcement learning environment framework, the agent resets the environment under the preset reinforcement learning environment framework and reads the corresponding hole parameter information and working condition based on the blade model configuration file, the working condition and the preset hole parameter file; wherein the hole parameter file includes the cooling parameters of the cooling structure;

[0039] The hole parameter information and operating conditions are processed based on a preset full three-dimensional evaluation algorithm to obtain an initial first turbine blade surface temperature after the environment is reset.

[0040] In a possible implementation, the training module is specifically used to:

[0041] The intelligent agent inputs the surface temperature of the first turbine blade and the cooling parameters of the cooling structure read from the hole parameter file into the strategy network to generate a corresponding target action, and hands the target action over to the reinforcement learning environment framework for implementation, so that the reinforcement learning environment framework adjusts the cooling parameters based on the target action;

[0042] In response to the completion of the parameter adjustment of the cooling parameter, the adjusted cooling parameter and the operating condition are calculated based on the full three-dimensional evaluation algorithm to obtain the adjusted surface temperature of the second turbine blade to complete a round of training for the intelligent agent;

[0043] In response to receiving a termination signal generated by a preset reward function being false, a new round of training of the agent is continued based on the surface temperature of the second turbine blade and the adjusted cooling parameters to ultimately obtain a trained agent.

[0044] In a possible implementation manner, the device further includes:

[0045] An updating module, configured to update the strategy network based on the adjusted cooling parameter and the second turbine blade surface temperature in response to receiving a termination signal generated by a preset reward function being false;

[0046] The judgment module is used to terminate the training in response to the updated policy network not converging.

[0047] In a possible implementation manner, the tuning module is specifically used to:

[0048] Under the reinforcement learning algorithm, based on the trained agent and the updated policy network, and based on the cooling parameters and basic parameters corresponding to the basic structure, an initial third turbine blade surface temperature is obtained under the reinforcement learning environment framework;

[0049] generating a corresponding second target action based on the third turbine blade surface temperature and the cooling parameter, and delivering the second target action to the reinforcement learning environment framework for implementation, so as to tune the cooling parameter of the cooling structure, and obtain a fourth turbine blade surface temperature and the tuned cooling parameter;

[0050] The change of the surface temperature of the turbine blade is evaluated based on the surface temperature of the fourth turbine blade, and the optimized cooling parameters are determined based on the evaluation result.

[0051] In a possible implementation manner, the tuning module is specifically used to:

[0052] In response to determining that the turbine blade surface temperature rises or the cooling parameter exceeds a preset parameter boundary during the implementation of the second target action based on the fourth turbine blade surface temperature and the initial third turbine blade surface temperature, the agent accumulates the second target action to obtain total modification information for the cooling structure;

[0053] The cooling parameters are modified based on the total modification information to obtain final optimized cooling parameters.

[0054] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to perform the steps of the turbine blade cooling structure tuning method based on reinforcement learning as described in any one of the first aspects.

[0055] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the turbine blade cooling structure tuning method based on reinforcement learning as described in any one of the first aspects are executed.

[0056] The embodiment of the present application provides a method and device for tuning the cooling structure of a turbine blade based on reinforcement learning. In response to inputting preset design conditions, the initial cooling structure of the turbine blade is obtained based on the design conditions. In response to the completion of the preset intelligent agent initialization, the intelligent agent is trained based on a preset reinforcement learning algorithm under a preset reinforcement learning environment framework to obtain a trained intelligent agent. The cooling parameters of the cooling structure are tuned based on the trained intelligent agent to obtain the tuned cooling parameters, and the tuned cooling structure is determined based on the tuned cooling parameters. In the present application, by introducing a reinforcement learning algorithm to train the intelligent agent in the process of designing the cooling structure of the turbine blade, and tuning the cooling parameters of the cooling structure according to the trained intelligent agent, a more complex or complete cooling structure can be applied, and the automatic tuning and determination of the cooling parameters of the cooling structure are realized, which reduces the reliance on the experience and time cost of the designer in the cooling parameter design process.

[0057] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 is a flow chart of a turbine blade cooling structure tuning method based on reinforcement learning according to an embodiment of the present application;

[0060] Figure 2 It is a schematic diagram of the cooling structure tuning process;

[0061] Figure 3 It is a schematic diagram of the iterative interaction process between the agent and the environment;

[0062] Figure 4 is a flow chart of a turbine blade cooling structure tuning method based on reinforcement learning according to another embodiment of the present application;

[0063] Figure 5 A flowchart of a turbine blade cooling structure tuning method based on reinforcement learning according to another embodiment of the present application;

[0064] Figure 6 It is a schematic diagram of the agent training process;

[0065] Figure 7 is a flow chart of a turbine blade cooling structure tuning method based on reinforcement learning according to another embodiment of the present application;

[0066] Figure 8 is a structural schematic diagram of a turbine blade cooling structure tuning device based on reinforcement learning according to an embodiment of the present application;

[0067] Fig. 9 It is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0068] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0069] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0070] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0071] Considering that in the field of aero-engines, the cooling structure of turbine blades is a key structure, and the design of the cooling structure of turbine blades is very complex and time-consuming, the design effect is very dependent on the professional experience of the designer. In addition, in order to design a set of parameters that meet the requirements, designers often need to adjust the design parameters many times, and then repeat the complex steps of blade modeling, meshing, simulation calculation, and evaluate whether this set of design parameters is qualified based on the obtained temperature field values. It usually takes designers more than ten days to get a set of parameters that meet the design conditions. However, these parameters will affect each other, and in traditional design methods, a high degree of professional experience and knowledge is required to optimize the cooling structure.

[0072] At present, in order to solve the cooling structure optimization problem, the turbine blade cooling structure is often tuned by using heuristic algorithms.

[0073] However, this method is only applicable to simple or local cooling structures, and during the implementation process, designers not only need to iteratively modify the basic structure and cooling structure parameters, but also need to continuously perform simulation calculations to evaluate the cooling effect. This takes a lot of time to search for a better solution and is highly dependent on the designer's experience.

[0074] To address this problem, the present application provides a method and device for tuning the cooling structure of a turbine blade based on reinforcement learning. By introducing a reinforcement learning algorithm to train an intelligent agent during the design process of the turbine blade cooling structure, and tuning the cooling parameters of the cooling structure according to the trained intelligent agent, more complex or complete cooling structures can be applied, thereby achieving automatic tuning and determination of the cooling parameters of the cooling structure, and reducing the reliance on the designer's experience and time cost during the cooling parameter design process.

[0075] Figure 1 is a flow chart of a method for optimizing a turbine blade cooling structure based on reinforcement learning according to an embodiment of the present application, such as Figure 1 As shown, the turbine blade cooling structure tuning method based on reinforcement learning in the embodiment of the present application may specifically include:

[0076] S101, in response to inputting a preset design condition, obtaining an initial cooling structure of a turbine blade based on the design condition.

[0077] S102, in response to the completion of the preset intelligent agent initialization, the intelligent agent is trained based on a preset reinforcement learning algorithm in a preset reinforcement learning environment framework to obtain a trained intelligent agent.

[0078] S103, tuning the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the tuned cooling parameters, and determining the tuned cooling structure based on the tuned cooling parameters.

[0079] In the above-mentioned turbine blade cooling structure tuning method based on reinforcement learning, by introducing a reinforcement learning algorithm to train the intelligent agent in the turbine blade cooling structure design process, and tuning the cooling parameters of the cooling structure according to the trained intelligent agent, a more complex or complete cooling structure can be applied, thereby realizing automatic tuning and determination of the cooling parameters of the cooling structure, and reducing the reliance on the designer's experience and time cost in the cooling parameter design process.

[0080] The above exemplary steps of the embodiment of the present application are described below with reference to specific examples:

[0081] S101, in response to inputting a preset design condition, obtaining an initial cooling structure of a turbine blade based on the design condition.

[0082] It should be noted that the present application adopts a reinforcement learning (RL) algorithm to perform turbine blade cooling structure tuning.

[0083] In the embodiment of the present application, the design conditions are the pre-set design conditions of the turbine blade cooling structure. The design conditions include a blade model configuration file and operating conditions. The blade model configuration file includes blade parameters. The blade model configuration file includes at least two-dimensional blade path, blade height, blade maximum and minimum radius, blade circumference, blade radius, blade maximum XY coordinates, blade x point, blade y point, blade midpoint X coordinate, number of blades. The operating conditions include at least gas inlet total pressure, gas outlet static pressure, gas inlet total temperature, cold air inlet total temperature, cold air inlet total pressure, cold air efficiency, and cold air flow upper limit. When the user inputs the preset design conditions, the initial cooling structure of the turbine blade is obtained according to the design conditions for subsequent processing. In other words, if Figure 2 As shown, the design conditions need to be input first, including the input and analysis of the two-dimensional blade shape and the input of the operating conditions, that is, the blade model configuration file and the operating conditions.

[0084] S102, in response to the completion of the preset intelligent agent initialization, the intelligent agent is trained based on a preset reinforcement learning algorithm in a preset reinforcement learning environment framework to obtain a trained intelligent agent.

[0085] In an embodiment of the present application, a reinforcement learning environment framework is a pre-set reinforcement learning environment framework built based on a reinforcement learning algorithm, for example, a gym environment framework, which is described in the present application as an example; a reinforcement learning algorithm corresponds to a policy network, and the present application describes the two reinforcement learning algorithms, PPO and SAC, as examples; a policy network is a pre-set policy network corresponding to a reinforcement learning algorithm for training an intelligent agent. When training an intelligent agent, the intelligent agent needs to be initialized first. After the initialization of the intelligent agent is completed, the intelligent agent is trained based on the preset policy network to obtain a trained intelligent agent for subsequent processing.

[0086] It is understood by those in the field that in reinforcement learning, the agent makes sequential decisions by interacting with the environment. Therefore, the environment that the agent interacts with plays a vital role in the training of reinforcement learning algorithms. In order to design the training interaction environment of reinforcement learning agents, gym, an open source Python environment simulation library, is usually used. gym is an open source Python environment simulation library developed and maintained by the non-profit company OpenAI for developing and comparing the performance of reinforcement learning algorithms. It enables reinforcement learning algorithms and environments to interact and communicate by providing a standard API.

[0087] It should be noted that the initialization of the agent is completed through the following steps: in the preset reinforcement learning environment framework, the agent resets the environment and reads the corresponding hole parameter information and working conditions based on the blade model configuration file, working conditions and preset hole parameter file; based on the preset full three-dimensional evaluation algorithm, the hole parameter information and working conditions are processed to obtain the initial turbine blade surface temperature after the environment is reset. Among them, the hole parameter file includes the cooling parameters of the cooling structure, and the cooling structure parameters include the spoiler column diameter, the air film hole diameter, the air film hole inclination angle, the impact hole diameter, the hole row spacing and the single row hole spacing. The full three-dimensional evaluation algorithm is a pre-set full three-dimensional evaluation algorithm.

[0088] For example, using the prepared gym environment framework, the intelligent agent first resets the environment, that is, copies the necessary files to the training directory, that is, the blade model configuration file, operating conditions and preset hole parameter files, reads these files to obtain the hole parameter information and operating conditions, and processes the hole parameter information and operating conditions based on the full three-dimensional evaluation algorithm to obtain the initial state after environmental processing, that is, the initial surface temperature of the first turbine blade, for subsequent processing.

[0089] It is understood by those in the field that reinforcement learning is a major method of machine learning that aims to learn the best behavior strategy by interacting with the environment. Simply put, the interaction between a machine and the environment means that the machine makes decisions in a state of the environment to generate a set of actions, and acts on the environment, and then the environment changes and feeds back a new state and immediate reward to the machine. In the training of reinforcement learning, the machine will continue to interact with the environment. The ultimate goal is to follow the learned strategy after training, and the expected cumulative reward obtained during the interaction process is the highest. Unlike other machine learning, in reinforcement learning, the machine that makes decisions is called an agent. The agent is different from the model in supervised learning. It learns the behavior strategy and outputs the actions selected under the guidance of the strategy, emphasizing that it can change the environment.

[0090] For example, Figure 3 As shown in Figure 1, the interaction between the agent and the environment is an iterative process. The agent first perceives the state of the current environment and selects the corresponding action based on its own strategy. Once the action is applied to the environment, the environment will feedback a reward signal and cause a state transition. The agent adjusts its strategy based on the state before and after the action, the action applied, and the reward, and selects a new action based on the new state in the next round of interaction. Through this interactive process, the agent gradually learns the optimal decision-making strategy in the current environment.

[0091] S103, tuning the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the tuned cooling parameters, and determining the tuned cooling structure based on the tuned cooling parameters.

[0092] In the embodiment of the present application, the cooling parameters of the cooling structure are tuned based on the trained intelligent agent obtained in step S102 to obtain the tuned cooling parameters, and the tuned cooling structure is determined based on the tuned cooling parameters.

[0093] The turbine blade cooling structure tuning method based on reinforcement learning provided in the embodiment of the present application, in response to inputting preset design conditions, obtains the initial cooling structure of the turbine blade based on the design conditions, and in response to the completion of the preset intelligent agent initialization, trains the intelligent agent based on the preset reinforcement learning algorithm under the preset reinforcement learning environment framework to obtain the trained intelligent agent, tunes the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the tuned cooling parameters, and determines the tuned cooling structure based on the tuned cooling parameters. The turbine blade cooling structure tuning method based on reinforcement learning of the present application, by introducing a reinforcement learning algorithm to train the intelligent agent in the turbine blade cooling structure design process, and tuning the cooling parameters of the cooling structure according to the trained intelligent agent, can apply more complex or complete cooling structures, realizes the automatic tuning and determination of the cooling parameters of the cooling structure, and reduces the reliance on the experience and time cost of the designer in the cooling parameter design process.

[0094] Further, such as Figure 4 As shown, in step S101 in the above embodiment, “in response to inputting a preset design condition, obtaining an initial cooling structure of the turbine blade based on the design condition” may specifically include the following steps:

[0095] S401, in response to inputting preset design conditions, performing external flow channel simulation calculation based on the design conditions to generate an initial basic structure of the turbine blade.

[0096] In the embodiment of the present application, when the preset design conditions are input, the outer flow channel simulation calculation is performed based on the design conditions to generate the initial basic structure of the turbine blade for subsequent processing, for example, Figure 2 shown.

[0097] It should be noted that the basic structure is obtained by one of the following steps: selecting the basic structure based on a preset solution database; recommending a set of blade cooling structures as the basic structure based on a preset intelligent recommendation algorithm; and determining the basic structure in response to the input structural parameters of the turbine blade. That is, the determination of the basic structure or the design of the basic structure includes three ways. The first is to select a set of existing design schemes in the solution database (i.e., a database storing basic structure schemes); the second is to use the intelligent recommendation algorithm to recommend a set of blade cooling structure designs. Before running the intelligent recommendation algorithm, a process detection is required to ensure that the input of the algorithm is complete; the third is to manually input the structural parameters of the turbine blade through the user interface to determine.

[0098] S402, obtaining a corresponding initial cooling structure based on the initial basic structure of the turbine blade.

[0099] In the embodiment of the present application, a corresponding initial cooling structure is obtained based on the initial basic structure of the turbine blade obtained in step S401.

[0100] It should be noted that the basic structure is used to determine the basic parameters of the cooling structure of the turbine blade, and the basic parameters include at least the blade wall structure, the inner and outer wall spacing and wall thickness, the number of internal partitions, the position of the internal partitions and the thickness; the cooling structure is used to determine the cooling parameters of the cooling structure of the turbine blade, and the cooling parameters include at least the spoiler column diameter, the impact aperture, the air film aperture, the air film hole angle, the hole row spacing, and the single row hole spacing.

[0101] Further, such as Figure 5 As shown, the step S102 in the above embodiment of “training the agent to obtain the trained agent” may specifically include the following steps:

[0102] S501, the intelligent agent inputs the surface temperature of the first turbine blade and the cooling parameters of the cooling structure read from the hole parameter file into the strategy network to generate a corresponding target action, and hands the target action over to the reinforcement learning environment framework for implementation, so that the reinforcement learning environment framework adjusts the cooling parameters based on the target action.

[0103] In the embodiment of the present application, the reinforcement learning environment framework includes an action space, a state space, and a reward function; the target action is a target action in the set of all actions included in the action space, and these actions characterize the range and possibility of the behavior that the agent can take. When training the agent, the agent inputs the surface temperature of the first turbine blade and the cooling parameters of the cooling structure read from the hole parameter file into the policy network to generate the corresponding target action, and hands the target action to the reinforcement learning environment framework for implementation, so that the reinforcement learning environment framework adjusts the cooling parameters based on the target action to obtain the adjusted cooling parameters. For example, after the agent is initialized, the agent will process the surface temperature of the turbine blade and the cooling parameters of the cooling structure read from the hole file into the state required by the policy network, and input the policy network to generate the corresponding target action. After the agent decides on the action, it will hand over the target action to the gym, and the gym will parse the action and operate the corresponding hole parameter file according to the action of the corresponding dimension to simulate the process of the designer adjusting the parameters to achieve the adjustment of the cooling parameters.

[0104] S502, in response to the completion of the cooling parameter adjustment, the adjusted cooling parameters and operating conditions are calculated based on the full three-dimensional evaluation algorithm to obtain the surface temperature of the second turbine blade after the adjustment, so as to complete a round of training for the intelligent agent.

[0105] In an embodiment of the present application, after the cooling parameters are adjusted in a round of training, the adjusted cooling parameters and operating conditions are calculated based on the full three-dimensional evaluation algorithm to obtain the surface temperature of the second turbine blade after adjustment to complete a round of training for the intelligent agent, which means that a round of training is completed.

[0106] S503, in response to receiving a termination signal generated by a preset reward function being false, a new round of training of the intelligent agent is continued based on the surface temperature of the second turbine blade and the adjusted cooling parameters to finally obtain a trained intelligent agent.

[0107] In an embodiment of the present application, the reward function is a pre-packaged reward function. During a round of training, the intelligent agent generates a corresponding termination signal according to the packaged reward function and judges the termination signal. If the termination signal is false, a new round of training is continued for the intelligent agent based on the surface temperature of the second turbine blade and the adjusted cooling parameters until a trained intelligent agent is finally obtained.

[0108] Furthermore, in response to receiving a termination signal generated by a preset reward function being false, the agent updates the policy network based on the adjusted cooling parameters and the second turbine blade surface temperature; in response to the updated policy network not converging, the training is terminated. Specifically, when the termination signal generated by the preset reward function is false, the policy network is updated based on the adjusted cooling parameters and the second turbine blade surface temperature to obtain an updated policy network. When the update of the entire stage is completed, the agent checks whether the policy network of this stage has converged, and if not, the training is terminated.

[0109] It should be noted that, for example, Figure 6 As shown in the figure, after the initialization, the agent will use the policy network to generate actions and hand them over to the gym environment framework to implement the actions. According to the encapsulated reward function, the state, action, reward, and termination signal tuples are generated and fed back to the agent. If the termination signal is false, the agent uses the new state, that is, the surface temperature of the second turbine blade, to generate a new target action and repeat the above process. If the termination signal is true, the agent will end the current round of trajectory collection and update the policy network according to the obtained data (including the surface temperature of the second turbine blade and the cooling parameters after parameter adjustment). After the update, the environment will be reset to start a new round of interaction. When the update of the entire stage is completed, the agent will check whether the policy network of this stage has converged. If it has converged, the target temperature of the environment will be modified, the latest cooling parameters will be saved, and the next stage of training will continue. If it has not converged, the training will end.

[0110] Thus, the training process of the intelligent agent is completed.

[0111] Further, such as Figure 7As shown, in step S103 in the above embodiment, “optimizing the cooling parameters of the cooling structure based on the trained intelligent agent to obtain optimized cooling parameters” may specifically include the following steps:

[0112] S701, under the reinforcement learning algorithm, based on the trained intelligent agent and the updated policy network, and based on the basic parameters corresponding to the cooling parameters and the basic structure, obtain the initial surface temperature of the third turbine blade in the reinforcement learning environment framework.

[0113] In an embodiment of the present application, under a selected reinforcement learning algorithm, based on the trained intelligent agent trained in the above embodiment, the updated policy network, and the basic parameters corresponding to the cooling parameters and the basic structure, the initial surface temperature of the third turbine blade is obtained in the reinforcement learning environment framework for subsequent processing.

[0114] S702, generating a corresponding second target action based on the surface temperature and cooling parameters of the third turbine blade, and handing the second target action over to the reinforcement learning environment framework for implementation, so as to tune the cooling parameters of the cooling structure and obtain the surface temperature of the fourth turbine blade and the tuned cooling parameters.

[0115] In the embodiment of the present application, the corresponding second target action is generated according to the third turbine blade surface temperature and cooling parameters and is implemented by the reinforcement learning environment framework to tune the cooling parameters of the cooling structure to obtain the fourth turbine blade surface temperature and the tuned cooling parameters. It should be noted that the tuning process is similar to the training process. The intelligent experience will load the corresponding parameters according to the selected reinforcement learning algorithm, use the gym environment framework to obtain the initial state, i.e., the turbine blade surface temperature, make decisions, and deliver the environment for implementation. For details, please refer to the training process in the above embodiment.

[0116] S703, evaluating the change of the surface temperature of the turbine blade based on the surface temperature of the fourth turbine blade, and determining the optimized cooling parameters based on the evaluation result.

[0117] In the embodiment of the present application, the change in the surface temperature of the turbine blade, that is, the increase or decrease in the surface temperature of the turbine blade, is evaluated according to the fourth turbine blade surface temperature, and the optimized cooling parameters are determined based on the evaluation results.

[0118] As a possible implementation, in response to determining that the turbine blade surface temperature rises or the cooling parameter exceeds the preset parameter boundary during the implementation of the second target action based on the fourth turbine blade surface temperature and the initial third turbine blade surface temperature, the intelligent agent accumulates the second target action to obtain the total modification information for the cooling structure; based on the total modification information, the cooling parameter is modified to obtain the final optimized cooling parameter. The parameter boundary is the upper and lower boundaries of the preset cooling parameter, and the total modification information is the total modification information or total change information between the initial cooling parameter of the cooling structure and the final cooling parameter.

[0119] Specifically, when repeating the tuning process, when the intelligent agent detects that the surface temperature of the turbine blade does not decrease but increases after a certain second target action, that is, the surface temperature of the fourth turbine blade is larger than the surface temperature of the third turbine blade, or the cooling parameter at this time is out of bounds, the intelligent agent will accumulate the second target action of each previous round to obtain the total modification information of the cooling structure, and modify the cooling parameters according to the total modification information to obtain the final tuned cooling parameters.

[0120] Thus, the tuning process of cooling parameters is completed.

[0121] It should be noted that this application is described using PPO and SAC reinforcement learning algorithms as examples, but it can also be implemented using other reinforcement learning algorithms, and no excessive restrictions are made on this.

[0122] Figure 8 is a flow chart of a turbine blade cooling structure tuning device based on reinforcement learning according to an embodiment of the present application, such as Figure 8 As shown, the turbine blade cooling structure tuning device 800 based on reinforcement learning in the embodiment of the present application may specifically include:

[0123] The acquisition module 801 is used to obtain the initial cooling structure of the turbine blade based on the input preset design conditions in response to the design conditions; the design conditions include the blade model configuration file and the operating conditions.

[0124] The training module 802 is used to train the intelligent agent based on a preset reinforcement learning algorithm in a preset reinforcement learning environment framework in response to the completion of the preset intelligent agent initialization to obtain a trained intelligent agent; wherein the reinforcement learning algorithm corresponds to a policy network.

[0125] The tuning module 803 is used to tune the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the tuned cooling parameters, and determine the tuned cooling structure based on the tuned cooling parameters.

[0126] In a possible implementation, the acquisition module is specifically configured to:

[0127] In response to inputting preset design conditions, performing external flow channel simulation calculation based on the design conditions to generate an initial basic structure of the turbine blade;

[0128] Based on the initial basic structure of the turbine blade, a corresponding initial cooling structure is obtained; wherein the basic structure is used to determine basic parameters of the cooling structure of the turbine blade; and the cooling structure is used to determine cooling parameters of the cooling structure of the turbine blade.

[0129] In a possible implementation, the agent is initialized by the following steps:

[0130] Before training the preset intelligent agent based on the preset strategy network in the preset reinforcement learning environment framework, the intelligent agent resets the environment and reads the corresponding hole parameter information and working condition based on the blade model configuration file, working condition and the preset hole parameter file in the preset reinforcement learning environment framework; wherein the hole parameter file includes the cooling parameters of the cooling structure;

[0131] The hole parameter information and operating conditions are processed based on a preset full three-dimensional evaluation algorithm to obtain the initial surface temperature of the first turbine blade after the environment is reset.

[0132] In a possible implementation, the training module is specifically used to:

[0133] The intelligent agent inputs the surface temperature of the first turbine blade and the cooling parameters of the cooling structure read from the hole parameter file into the strategy network to generate a corresponding target action, and hands the target action to the reinforcement learning environment framework for implementation, so that the reinforcement learning environment framework adjusts the cooling parameters based on the target action;

[0134] In response to the completion of the cooling parameter adjustment, the adjusted cooling parameters and operating conditions are calculated based on the full three-dimensional evaluation algorithm to obtain the adjusted surface temperature of the second turbine blade to complete a round of training for the intelligent agent;

[0135] In response to receiving a termination signal generated by a preset reward function being false, a new round of training of the intelligent agent is continued based on the surface temperature of the second turbine blade and the adjusted cooling parameters to finally obtain a trained intelligent agent.

[0136] In a possible implementation, the turbine blade cooling structure tuning device based on reinforcement learning further includes:

[0137] An updating module, configured to update the policy network based on the adjusted cooling parameters and the surface temperature of the second turbine blade in response to receiving a termination signal generated by a preset reward function being false;

[0138] The judgment module is used to terminate the training in response to the updated policy network not converging.

[0139] In a possible implementation, the tuning module is specifically used to:

[0140] Under the reinforcement learning algorithm, based on the trained agent and the updated policy network, and based on the basic parameters corresponding to the cooling parameters and the basic structure, the initial surface temperature of the third turbine blade is obtained under the reinforcement learning environment framework;

[0141] generating a corresponding second target action based on the surface temperature and cooling parameters of the third turbine blade, and delivering the second target action to the reinforcement learning environment framework for implementation, so as to tune the cooling parameters of the cooling structure, and obtain the surface temperature and the optimized cooling parameters of the fourth turbine blade;

[0142] The change in the surface temperature of the turbine blade is evaluated based on the surface temperature of the fourth turbine blade, and the optimized cooling parameters are determined based on the evaluation result.

[0143] In a possible implementation, the tuning module is specifically used to:

[0144] In response to determining that the turbine blade surface temperature rises or the cooling parameter exceeds a preset parameter boundary during the implementation of the second target action based on the fourth turbine blade surface temperature and the initial third turbine blade surface temperature, the agent accumulates the second target action to obtain total modification information for the cooling structure;

[0145] The cooling parameters are modified based on the total modification information to obtain the final optimized cooling parameters.

[0146] The turbine blade cooling structure tuning device based on reinforcement learning provided in the embodiment of the present application, in response to inputting preset design conditions, obtains the initial cooling structure of the turbine blade based on the design conditions, and in response to the completion of the preset intelligent agent initialization, trains the intelligent agent based on the preset reinforcement learning algorithm under the preset reinforcement learning environment framework to obtain the trained intelligent agent, tunes the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the tuned cooling parameters, and determines the tuned cooling structure based on the tuned cooling parameters. The turbine blade cooling structure tuning device based on reinforcement learning of the present application, by introducing the reinforcement learning algorithm to train the intelligent agent in the process of turbine blade cooling structure design, and tuning the cooling parameters of the cooling structure according to the trained intelligent agent, can apply more complex or complete cooling structures, realizes the automatic tuning and determination of the cooling parameters of the cooling structure, and reduces the dependence on the experience and time cost of the designer in the cooling parameter design process.

[0147] like Fig. 9As shown, an electronic device 900 provided in an embodiment of the present application includes: a processor 901, a memory 902 and a bus, wherein the memory 902 stores machine-readable instructions executable by the processor 901. When the electronic device is running, the processor 901 communicates with the memory 902 through the bus, and the processor 901 executes the machine-readable instructions to perform the steps of the turbine blade cooling structure tuning method based on reinforcement learning as described above.

[0148] Specifically, the above-mentioned memory 902 and processor 901 can be general-purpose memory and processor, which are not specifically limited here. When the processor 901 runs the computer program stored in the memory 902, it can execute the above-mentioned turbine blade cooling structure tuning method based on reinforcement learning.

[0149] Corresponding to the above-mentioned turbine blade cooling structure tuning method based on reinforcement learning, an embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned turbine blade cooling structure tuning method based on reinforcement learning are executed.

[0150] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application. In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0151] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0152] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0153] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the deployment method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0154] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for optimizing turbine blade cooling structure based on reinforcement learning, characterized in that: The turbine blade cooling structure tuning method comprises: In response to inputting preset design conditions, an initial cooling structure of the turbine blade is obtained based on the design conditions; the design conditions include a blade model configuration file and an operating condition; In response to the completion of the preset intelligent agent initialization, the intelligent agent is trained based on a preset reinforcement learning algorithm in a preset reinforcement learning environment framework to obtain a trained intelligent agent; wherein the reinforcement learning algorithm corresponds to a strategy network; The cooling parameters of the cooling structure are tuned based on the trained intelligent agent to obtain the tuned cooling parameters, and the tuned cooling structure is determined based on the tuned cooling parameters.

2. The method according to claim 1, characterized in that The step of obtaining an initial cooling structure of the turbine blade based on the preset design conditions in response to inputting the preset design conditions comprises: In response to inputting preset design conditions, performing external flow channel simulation calculation based on the design conditions to generate an initial basic structure of the turbine blade; A corresponding initial cooling structure is obtained based on the initial basic structure of the turbine blade; wherein the basic structure is used to determine basic parameters of the cooling structure of the turbine blade; and the cooling structure is used to determine cooling parameters of the cooling structure of the turbine blade.

3. The method according to claim 2, characterized in that Before training the preset intelligent agent based on the preset strategy network in the preset reinforcement learning environment framework, the intelligent agent is initialized by the following steps: In a preset reinforcement learning environment framework, the agent resets the environment and reads corresponding hole parameter information and working conditions based on the blade model configuration file, the working conditions and a preset hole parameter file; wherein the hole parameter file includes cooling parameters of the cooling structure; The hole parameter information and operating conditions are processed based on a preset full three-dimensional evaluation algorithm to obtain an initial first turbine blade surface temperature after the environment is reset.

4. The method according to claim 3, characterized in that The step of training the intelligent agent to obtain a trained intelligent agent includes: The intelligent agent inputs the surface temperature of the first turbine blade and the cooling parameters of the cooling structure read from the hole parameter file into the strategy network to generate a corresponding target action, and hands the target action over to the reinforcement learning environment framework for implementation, so that the reinforcement learning environment framework adjusts the cooling parameters based on the target action; In response to the completion of the parameter adjustment of the cooling parameter, the adjusted cooling parameter and the operating condition are calculated based on the full three-dimensional evaluation algorithm to obtain the adjusted surface temperature of the second turbine blade to complete a round of training for the intelligent agent; In response to receiving a termination signal generated by a preset reward function being false, a new round of training of the agent is continued based on the surface temperature of the second turbine blade and the adjusted cooling parameters to ultimately obtain a trained agent.

5. The method according to claim 4, characterized in that The method further comprises: In response to receiving a termination signal generated by a preset reward function being false, the agent updates the policy network based on the adjusted cooling parameter and the second turbine blade surface temperature; In response to the updated policy network not converging, the training is terminated.

6. The method according to claim 5, characterized in that The step of optimizing the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the optimized cooling parameters includes: Under the reinforcement learning algorithm, based on the trained agent and the updated policy network, and based on the cooling parameters and basic parameters corresponding to the basic structure, an initial third turbine blade surface temperature is obtained under the reinforcement learning environment framework; generating a corresponding second target action based on the third turbine blade surface temperature and the cooling parameter, and delivering the second target action to the reinforcement learning environment framework for implementation, so as to tune the cooling parameter of the cooling structure, and obtain a fourth turbine blade surface temperature and the tuned cooling parameter; The change of the surface temperature of the turbine blade is evaluated based on the surface temperature of the fourth turbine blade, and the optimized cooling parameters are determined based on the evaluation result.

7. The method according to claim 6, characterized in that The step of evaluating the change of the surface temperature of the turbine blade based on the fourth turbine blade surface temperature and determining the optimized cooling parameter based on the evaluation result includes: In response to determining that the turbine blade surface temperature rises or the cooling parameter exceeds a preset parameter boundary during the implementation of the second target action based on the fourth turbine blade surface temperature and the initial third turbine blade surface temperature, the agent accumulates the second target action to obtain total modification information for the cooling structure; The cooling parameters are modified based on the total modification information to obtain final optimized cooling parameters.

8. A turbine blade cooling structure tuning device based on reinforcement learning, characterized in that: The device comprises: An acquisition module, configured to obtain an initial cooling structure of a turbine blade based on an input preset design condition in response to the input preset design condition; the design condition includes a blade model configuration file and an operating condition; A training module, configured to, in response to the completion of the initialization of the preset intelligent agent, train the intelligent agent based on a preset reinforcement learning algorithm in a preset reinforcement learning environment framework to obtain a trained intelligent agent; wherein the reinforcement learning algorithm corresponds to a policy network; A tuning module is used to tune the cooling parameters of the cooling structure based on the trained intelligent agent to obtain the tuned cooling parameters, and determine the tuned cooling structure based on the tuned cooling parameters.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the turbine blade cooling structure tuning method based on reinforcement learning are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, executes the steps of the turbine blade cooling structure tuning method based on reinforcement learning as described in any one of claims 1 to 7.

Images