Simulation method, device, equipment, medium and product for an aeroengine

Abstract

This disclosure relates to the field of artificial intelligence technology, and in particular to a simulation method, apparatus, equipment, medium, and product for aero-engines. The method includes: acquiring the operating parameters and simulation target of the current aero-engine simulation case, and acquiring similar historical operating parameters; constructing a calibration case based on the historical operating parameters; performing rapid simulation on the calibration case, and generating an analysis report based on the simulation results; updating the parameter values ​​of the model parameters based on the analysis report and a reinforcement learning algorithm to obtain an updated calibration case; repeating rapid simulation on the calibration case until an optimal case is obtained; and performing formal simulation based on the optimal case and the simulation target. By constructing a calibration case for rapid trial calculations and executing a cyclical process of "rapid simulation - generating analysis report - parameter update," the optimal case is identified. The calibration stage incorporates a reinforcement learning algorithm, enabling adaptive iteration of model parameters without manual parameter tuning, thus improving work efficiency.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a simulation method, apparatus, equipment, medium, and product for aero-engines. Background Technology

[0002] During the design process of aero engines, simulation is required to predict their performance, identify problems, and optimize the design. This helps designers understand the product performance of aero engines and provides a basis and support for optimization and modification.

[0003] In the calibration phase of aero-engine simulation, a large number of model parameters, such as spray, turbulence, and combustion, need to be set in advance. The setting of these parameters is highly dependent on the human experience of experts, and the process is cumbersome, time-consuming, and labor-intensive. It is also prone to human error, resulting in inconsistent simulation quality. Therefore, there is an urgent need for a more intelligent and automated aero-engine simulation method. Summary of the Invention

[0004] In view of the above problems, this disclosure is made to provide a simulation method, apparatus, equipment, medium and product for aero-engines.

[0005] According to one aspect of this disclosure, a simulation method for an aero-engine is provided, comprising: Obtain the operating parameters and simulation objectives of the current simulation example of the aero-engine, and retrieve historical operating parameters that match the operating parameters from the historical simulation database; Based on historical operating parameters, a calibration case corresponding to the current case is constructed; the calibration case includes the simulation model required for the current case, as well as the model parameters of the simulation model; The calibration examples are subjected to rapid simulation. Based on the results of the rapid simulation, an analysis report is generated. Based on the analysis report and the preset reinforcement learning algorithm, the parameter values ​​of the model parameters are updated, and the updated calibration examples are obtained. The rapid simulation is repeated on the updated calibration examples until the iteration stopping condition is met, and the optimal example is obtained. The rapid simulation is used to quickly evaluate the effect of the current parameters of the model. Based on the optimal calculation example and simulation objectives, a formal simulation of the aero-engine is carried out.

[0006] The beneficial effects of this scheme are as follows: By retrieving similar historical operating conditions from the historical simulation database, a calibration case can be constructed for rapid trial calculations. Through a cyclical optimization process of "rapid simulation - generating analysis report - parameter update" on the calibration case, the optimal case can be identified, including all operating condition parameters and model parameters, for use in formal simulations. The optimization process in the calibration phase incorporates reinforcement learning algorithms, specifically the Soft Actor-Critic (SAC) algorithm. This reinforcement learning algorithm enables adaptive iterative optimization of model parameters, which can be completed automatically in a short time without repeated manual parameter tuning, significantly improving the efficiency of pre-simulation preparation and ensuring the quality of simulation results.

[0007] Furthermore, according to one aspect of the simulation method of this disclosure, a calibration case corresponding to the current case is constructed based on historical operating parameters, including: Based on the operating parameters and simulation objectives, confirm the simulation model required for the current case, as well as the model parameters of the simulation model; Based on historical operating condition parameters, the initial parameter values ​​of the model parameters are determined and used as the parameter values. Based on the operating parameters, simulation model, and parameter values, construct the calibration case corresponding to the current case.

[0008] In one or more embodiments, the beneficial effects of this solution are as follows: the simulation model includes, for example, a spray model, a turbulence model, a combustion model, and a flame surface treatment method; the model parameters include, for example, turbulent viscosity correlation constants, injection cone angle, particle number and size distribution, fragmentation / merging model parameters, and reaction rate correlation coefficients; the selected model parameters vary depending on the combination of simulation models required for the current example. Determining the initial parameter values ​​of the model through historical operating condition parameters, and using these initial parameter values ​​as the starting point of the calibration cycle, can improve the numerical accuracy of the starting point and ensure optimization quality.

[0009] Furthermore, according to one aspect of the simulation method of this disclosure, based on the analysis report and a preset reinforcement learning algorithm, the parameter values ​​of the model parameters are updated, including: The analysis report is quantified into an index vector based on a pre-defined reinforcement learning algorithm; the index vector is used to describe the score corresponding to the parameter value. The indicator vector and the parameter values ​​of the model parameters are input into the reinforcement learning algorithm. The reinforcement learning algorithm constructs a reward function based on the improvement amount and updates the parameter values. The improvement amount is used to represent the difference between the parameter values ​​before and after the model parameter update.

[0010] In one or more embodiments, the beneficial effects of this solution are as follows: By using reinforcement learning algorithms to transform descriptive conclusions in the analysis report into quantifiable index vectors, such as overall flow field scores, hotspot intensity scores, wall collision risk scores, and mixing uniformity scores, complex physical phenomena can be mapped to a finite-dimensional index space, serving as input to the reinforcement learning algorithm. Based on the score feedback, the reinforcement learning algorithm automatically increases or decreases parameter values, automatically completing the cycle of "rapid simulation - generating analysis report - parameter update," achieving adaptive iterative optimization of model parameters without the need for repeated manual parameter tuning.

[0011] Furthermore, according to one aspect of the simulation method of this disclosure, the iteration stopping condition includes: The improvement in the parameter values ​​of the target model is less than a preset threshold, or the improvement in the parameter values ​​of the target model reaches the target range; where the target model parameter is the key parameter in the model parameter.

[0012] In one or more embodiments, the beneficial effect of this solution is that by setting an iteration stopping condition, it is possible to know whether the parameter value has reached the optimal value.

[0013] Furthermore, according to a simulation method of one aspect of this disclosure, an analysis report is generated based on the results of a rapid simulation, including: Based on the results of rapid simulation, flow field cloud maps and monitoring point data are obtained; The operating parameters, simulation targets, flow field cloud maps, and monitoring point data are integrated into multimodal data; Multimodal data is input into a pre-trained multimodal large model, which is then used to identify the flow field structure and generate an analysis report.

[0014] In one or more embodiments, the beneficial effects of this solution are as follows: the multimodal large model is adjusted by the efficient fine-tuning technology (Low-Rank Adaptation, LoRA), and the flow field structure can be identified through the multimodal large model, including hot spot regions, recirculation zones, spray wall impacts, cold flow entrainment, and uneven mixing, and an analysis report for calibration is generated.

[0015] Furthermore, according to one aspect of the simulation method of this disclosure, a rapid simulation of the calibration example is performed, including: The preset number of iteration steps or runtime; wherein the number of iteration steps or runtime is less than the number of steps or runtime required for the actual simulation; By using a flame face substitution model based on a multilayer perceptron, the calibration example is solved according to the number of iteration steps or runtime to predict the physical parameters required for the flame face in real time.

[0016] In one or more embodiments, the beneficial effects of this solution are: a more relaxed convergence criterion and shorter iteration steps or runtime are adopted in the rapid simulation phase to quickly complete a simulation evaluation. The flame surface substitution model based on a multilayer perceptron can replace the traditional flame surface database lookup method, making the acquisition of physical property parameters more lightweight and efficient. The multilayer perceptron is used to fit the combustion flame surface data of aero-engines, which can reduce the memory usage and I / O overhead during combustion model calculations.

[0017] According to another aspect of this disclosure, a simulation apparatus for an aircraft engine is provided, comprising: The acquisition module is used to acquire the operating parameters and simulation targets of the current simulation example of the aero-engine, and to acquire historical operating parameters that match the operating parameters from the historical simulation database. The calibration case construction module is used to construct the calibration case corresponding to the current case based on historical operating condition parameters. The calibration case includes the simulation model required for the current case and the model parameters of the simulation model. The simulation optimization module is used to perform rapid simulations on the calibration simulations. Based on the results of the rapid simulations, an analysis report is generated. Based on the analysis report and the preset reinforcement learning algorithm, the parameter values ​​of the model parameters are updated, and the updated calibration simulations are obtained. The rapid simulations on the updated calibration simulations are repeated until the iteration stopping condition is met, and the optimal simulation is obtained. Among them, the rapid simulation is used to quickly evaluate the effect of the current parameters of the model. The simulation module is used to perform formal simulations of aero-engines based on optimal calculation examples and simulation objectives.

[0018] According to another aspect of this disclosure, a computer device is provided, including a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to implement the method of one aspect above.

[0019] According to another aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the method of one aspect above.

[0020] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method of the above-described aspect.

[0021] It should be understood that both the foregoing general description and the following detailed description are exemplary and intended to provide further illustration of the claimed technology. Attached Figure Description

[0022] The above and other objects, features, and advantages of this disclosure will become more apparent from the more detailed description of the embodiments thereof in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the disclosure and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0023] Figure 1 This is a system architecture diagram illustrating a simulation method for an aero-engine according to an embodiment of the present disclosure.

[0024] Figure 2 This is a flowchart illustrating the workflow of the application configuration module according to an embodiment of this disclosure.

[0025] Figure 3 This is a flowchart illustrating a simulation method for an aero-engine according to an embodiment of the present disclosure.

[0026] Figure 4 This is a schematic diagram of the structure of a simulation device for an aero-engine according to an embodiment of the present disclosure.

[0027] Figure 5 This is a schematic diagram illustrating the structure of a computer device according to an embodiment of the present disclosure.

[0028] Figure 6 This is a schematic diagram illustrating a computer program product according to an embodiment of the present disclosure. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this disclosure more apparent, exemplary embodiments according to this disclosure will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this disclosure, and not all embodiments of this disclosure. It should be understood that this disclosure is not limited to the exemplary embodiments described herein.

[0030] In the design process of aero-engines, simulation is necessary to predict performance, identify problems, and optimize the design. This helps designers understand the product performance of aero-engines and provides a basis and support for optimization and modification. Currently, three-dimensional numerical simulation has become an indispensable and important part of aero-engine development. Its high-confidence predictive capabilities can significantly reduce the cost of trial and error in testing. The simulation displays a wealth of information about the physical processes, helping designers understand product performance from a mechanistic perspective and providing a basis and support for scheme optimization. After introducing simulation calculations into many aspects of aero-engine design, the entire product development cycle has been shortened from more than ten years to a few years, significantly improving product development efficiency.

[0031] However, the following problems still exist in the existing aero-engine simulation process: (1) During the calibration stage of the simulation, a large number of model parameters need to be set in advance, such as spray parameters, turbulence constants, boundary conditions, etc. The specific values ​​of these model parameters are highly dependent on the human experience of experts, which is prone to human error. A large number of repeated calculations are needed to obtain stable and reliable parameter values. (2) The process of constructing simulation examples is complicated and prone to inconsistencies, which can easily lead to inconsistent simulation quality. (3) The lack of intelligent configuration and auxiliary decision-making capabilities during the simulation process results in a long overall simulation cycle and high resource consumption; (4) The analysis of simulation results (such as three-dimensional flow field cloud map, multiphase distribution field, temperature field, etc.) heavily relies on the experience and judgment of senior experts.

[0032] The above description, with reference to the accompanying drawings, illustrates a simulation method, apparatus, device, medium, and product for aero-engines according to embodiments of the present disclosure. By utilizing reinforcement learning algorithms (SAC) to automatically learn parameter optimization strategies before simulation, it reduces repeated trial calculations during the simulation preparation phase and automatically generates optimal simulation cases. By using deep neural networks to perform deep learning on the FGM flame surface database, it significantly reduces the memory usage of model calculations and improves computational efficiency. By integrating large-scale simulation data and textual knowledge, it performs domain fine-tuning of multimodal models, enabling automated analysis and engineering semantic interpretation of simulation results. Through simulation case configuration modules, simulation execution modules, and result analysis modules, it forms an intelligent processing system for aero-engine simulation operations. This system significantly reduces efficiency losses due to reliance on human feedback, reduces manual intervention in repetitive tasks, improves the automation level and consistency of simulation preparation, simulation execution, and result analysis, enhances the predictive reliability and engineering efficiency of aero-engine simulation activities, and promotes the intelligent and standardized development process of aero-engines.

[0033] To facilitate understanding of this embodiment, a simulation method for aero-engines disclosed in this disclosure will first be described in detail. The execution subject of the simulation method for aero-engines provided in this disclosure is generally a computer device with certain computing capabilities. In some possible implementations, the simulation method can be implemented by the processor calling computer-readable instructions stored in the memory.

[0034] like Figure 1The diagram shown is a system architecture diagram of the simulation method for aero-engines provided in this embodiment of the present disclosure, including a case configuration module 1, a simulation execution module 2, and a result analysis module 3. This embodiment divides the simulation process of aero-engines into a calibration stage and a formal simulation stage. The calibration stage is used to automatically generate the optimal case, and the formal simulation stage outputs a final analysis report for engineers based on the optimal case. Therefore, this embodiment describes each module for each of the two stages.

[0035] First, during the calibration phase, the functions of each module are described below according to the calibration process: The simulation configuration module 1 receives the simulation objective and determined operating parameters input by the user. Based on the user's requirements, it automatically determines the combination of simulation models to be used in the current simulation and configures a set of initial model parameters for these models. It retrieves the historical operating parameters most similar to the current operating condition from the historical simulation database and constructs a calibration simulation corresponding to the current simulation based on these historical operating parameters.

[0036] The operating parameters include: setting inlet boundary conditions (such as total temperature, total pressure, mass flow rate, and velocity distribution), outlet boundary conditions (such as back pressure and mass flow rate distribution), flow distribution of each branch, rotational speed, fuel flow rate, and basic parameters such as medium properties. The simulation models include spray models, turbulence models, combustion models, and corresponding flame surface treatments. Model parameters include, but are not limited to: turbulent viscosity correlation constants, injection cone angle, particle number and size distribution, breakup / merging model parameters, and reaction rate correlation coefficients. The case configuration module only determines "what flow field to calculate" and "how to define the boundaries," without specifying the exact values ​​of the model parameters. Model parameters can be based on empirical default values ​​or selected from a set of typical model parameters from a historical simulation database according to operating condition characteristics, serving as the starting point for the calibration cycle.

[0037] For retrieving historical operating condition parameters, taking aero-engine combustion / spray examples as an example, the retrieval dimensions include nozzle type, combustion mode, inlet pressure and temperature, swirl intensity, flow distribution, and key performance indicators. Based on the retrieved similar operating condition parameters, the initial parameter values ​​of the model are calibrated. For example, the spray angle, particle number, and turbulence model coefficients are adjusted to values ​​close to those of historical operating condition parameters, thereby improving the starting quality of subsequent parameter optimization.

[0038] Simulation Execution Module 2: The simulation configuration module calls the simulation execution module to perform short-cycle fast simulation on the calibration case, that is, to solve the calibration case.

[0039] The calibration case maintains the same simulation model selection as the case in the formal simulation phase. However, when performing rapid simulation on the calibration case, the simulation execution module adopts a more lenient convergence criterion and a shorter runtime or number of iterations to quickly complete a simulation evaluation, obtain flow field cloud maps and monitoring point data, and finally the case configuration module transmits the results of the rapid simulation to the result analysis module.

[0040] To accelerate the iterative optimization during the calibration phase, the simulation execution module employs a flame surface substitution model based on a multilayer perceptron (MLP) to replace the traditional flame surface database lookup method, making the acquisition of combustion flame surface physical parameters more lightweight and efficient. This MLP model is specifically designed for fitting aero-engine combustion flame surface data, reducing memory usage and I / O overhead during combustion model calculations.

[0041] Results Analysis Module 3: Integrates operating parameters, simulation targets, flow field cloud maps, and monitoring point data into multimodal data. Through a multimodal large model fine-tuned by LoRA, the flow field structure is identified, and an analysis report is generated.

[0042] Specifically, the multimodal model can identify typical physical phenomena related to aero-engine combustion chambers, such as hotspot regions, recirculation zones, spray wall impacts, cold flow entrainment, and uneven mixing, and generate analysis reports for calibration.

[0043] Example Configuration Module 1: This module uses a large language model to parse and quantify the natural language text descriptions in the analysis report, transforming "descriptive conclusions" into "index vectors that can be optimized," such as overall flow field scores, hotspot intensity scores, wall collision risk scores, and mixing uniformity scores. This maps complex physical phenomena to a finite-dimensional index space, serving as the input basis for reinforcement learning algorithms. Specifically, the reinforcement learning algorithm is the Soft Actor-Critic (SAC) algorithm. The index vectors and the current parameter values ​​of each model are used as input to the SAC algorithm. An improvement amount is used to construct a reward function, updating the parameter values ​​of each model in the calibration example, ultimately obtaining the updated calibration example. The improvement amount represents the difference between the parameter values ​​before and after the update. The SAC algorithm is only used for CFD preprocessing and calibration of aero-engines and does not participate in controller parameter tuning or general decision-making tasks.

[0044] Based on the updated calibration case, the case configuration module calls the simulation execution module again to perform short-cycle rapid simulation on the calibration case. That is, by repeatedly executing the cyclic process of "short-cycle rapid simulation → multimodal large model analysis → large language model quantization → reinforcement learning algorithm parameter update", the case configuration of the calibration case gradually converges towards a direction with better indicators and more reasonable physics.

[0045] When the improvement of several key model parameters falls below a preset threshold or reaches the target range, it indicates that the convergence condition (i.e., the iteration stopping condition) has been met, the loop exits, and the calibration process ends. At this point, the case configuration module outputs the final calibration case configuration (i.e., the optimal case), including boundary conditions, flow distribution, and parameter settings for each simulation model, for use in the subsequent formal simulation phase.

[0046] Then, in the formal simulation phase, the detailed description of each module is as follows: Case configuration module 1: Executes the entire calibration phase described above and finally outputs the optimal case (i.e., the official case). Specifically, it is equivalent to embedding a simulation execution module and a result analysis module within the case configuration module to complete the entire calibration phase process.

[0047] like Figure 2 The diagram shown is a flowchart of the simulation configuration module, including: S201: Parse user input and select simulation model.

[0048] S202: Retrieve similar historical operating conditions.

[0049] S203: Construct a calibration example.

[0050] S204: Call the simulation execution module to perform a short-cycle simulation on the calibration example.

[0051] S205: Call the results analysis module to perform multimodal analysis and generate an analysis report.

[0052] S206: Quantify the analysis report to obtain an indicator vector.

[0053] S207: Input the index vector into the reinforcement learning algorithm for parameter update.

[0054] S208: Determine whether the convergence condition is met. If yes, proceed to S209; otherwise, return to S204.

[0055] S209: Output the optimal example.

[0056] Simulation Execution Module 2: Used to execute the formal simulation work based on the optimal simulation case (formal simulation case) output by the simulation case configuration module. Specifically, it includes the following: Receive the case configuration and complete the solution initialization: Receive the formal case file output by the case configuration module, read the mesh, boundary conditions, flow distribution, physical properties and solution control parameters, complete the consistency verification and solution initialization, and establish monitoring points and output configuration; this module is limited to CFD numerical solutions for aero-engine combustion chambers and related components.

[0057] Enabling combustion-related solving capabilities and loading a flame surface substitution model: When formal calculations involve combustion solving and flame surface property acquisition, a multilayer perceptron-based flame surface substitution model is enabled during the solving process to predict flame surface-related properties, replacing the traditional database lookup and interpolation process, thereby reducing the overhead of flame surface data loading and querying. This substitution model is limited to the scenario of calculating flame surface properties for aero-engine combustion. After obtaining the corresponding data type, this data is stored in the topology information. The data is then filled into one of the generated containers (point information, surface topology information, and volume topology information). At this point, each process has completely analyzed the data in its temporary storage area. Thus, combining the three pieces of information from all processes yields the overall mesh topology information.

[0058] Execute CFD solution and record process data: Perform steady-state or unsteady-state solution according to the formal simulation settings, continuously record residuals, physical quantities at key monitoring points and necessary operation logs until the preset convergence conditions or calculation termination conditions are met.

[0059] Output simulation results and contour maps: Output simulation results such as temperature field, velocity field, pressure field, spray distribution, and necessary cross-sectional / statistical features, and generate multimodal results (including contour maps and corresponding numerical feature sequences) for post-processing. Output and transfer simulation results and metadata (operating conditions, model configuration summary, monitoring information, etc.) to the results analysis module according to a preset data format to support subsequent contour map recognition and analysis report generation based on LoRA-based fine-tuned multimodal large models.

[0060] Results Analysis Module 3: After the formal simulation is completed: The results analysis module re-analyzes the contour maps and numerical data, generating a final analysis report for engineers, including the identification of key physical phenomena, explanations of possible causes, and suggestions for structural or operational condition adjustments. Specifically, it includes: The module receives simulation results and constructs multimodal input samples: It receives simulation result data output from the simulation execution module, including at least CFD contour maps of temperature, velocity, pressure, and spray distribution, along with their corresponding monitoring point data, cross-sectional data, or statistical features. It also receives the operating conditions and configuration metadata for the simulation example (such as configuration descriptions, boundary condition summaries, flow distribution, and spray / combustion model selection information). The module performs unified scale normalization, viewpoint / cross-section normalization, or key region clipping on the contour maps, and organizes the numerical monitoring features and textual operating condition information into multimodal input samples aligned with the contour maps. This ensures that subsequent model inference can interpret the physical phenomena in the contour maps within the same semantic context.

[0061] A LoRA-tuned multimodal large-scale model is used for physical structure identification: A pre-trained multimodal large-scale model is loaded, and neighborhood-tuned weights obtained using LoRA (low-rank adaptation) are used to enable the multimodal large-scale model to identify CFD cloud maps related to the combustion chamber / spray / combustion of aero-engines. The multimodal large-scale model performs inference with a combination of "cloud map + operating condition text + numerical features" as input, and outputs the identification results of key physical structures, including but not limited to configuration, recirculation zone, separation zone, cold flow entrainment, hot spot region, combustion inhomogeneity region, spray wall impact risk region, and insufficient mixing region, and provides corresponding regional location information.

[0062] Generate engineering semantic interpretation and anomaly cause inference: After identifying key structures, combine operating condition text and monitoring point trends to perform engineering semantic interpretation of phenomena, forming an organization of "phenomenon-evidence-possible cause". For example, when the multimodal large model identifies local high temperature concentration and abnormal temperature fluctuation at monitoring points, it gives the explanation of "enhanced local hot spots" and associates it with possible causes such as uneven flow distribution, enhanced recirculation structure, or insufficient mixing; when it identifies that the spray is enriched near the wall and the oil droplet distribution is abnormal, it gives the explanation of "high risk of spray hitting the wall" and associates it with possible factors such as the setting of the spray cone angle, particle size distribution, or local flow deflection.

[0063] A structured final analysis report will be generated, outlining key conclusions: The identification and interpretation results will be organized into a structured final analysis report, which will include at least: a list of key phenomena in this case study, a description of key areas (including location information), main anomaly types and their possible causes, an explanation of influencing factors related to the operating conditions / configuration, and recommendations for engineers (e.g., adjusting spray-related parameter ranges, checking local flow distribution, and monitoring temperature / equivalent ratio changes in specific cross-sectional areas). The final analysis report will be output in both readable text and parsable structured field formats for easy reading by engineers.

[0064] Based on the above embodiments, this embodiment also provides a simulation method for aero-engines, such as... Figure 3 The diagram shows the flowchart of the simulation method, including S301-S304: S301: Obtain the operating parameters and simulation objectives of the current simulation example of the aero-engine, and retrieve historical operating parameters that match the operating parameters from the historical simulation database.

[0065] S302: Based on historical operating parameters, construct the calibration case corresponding to the current case; The calibration case includes the simulation model required for the current case, as well as the model parameters of the simulation model.

[0066] S303: Perform rapid simulation on the calibration example, generate an analysis report based on the results of the rapid simulation, update the parameter values ​​of the model parameters based on the analysis report and the preset reinforcement learning algorithm, and obtain the updated calibration example. Repeat the rapid simulation on the updated calibration example until the iteration stopping condition is met to obtain the optimal example. Among them, rapid simulation is used to quickly evaluate the effect of the current parameters of the model.

[0067] S304: Based on the optimal calculation case and simulation objectives, conduct formal simulation of the aero-engine.

[0068] In one or more embodiments, a calibration case corresponding to the current case is constructed based on historical operating condition parameters, including: Based on the operating parameters and simulation objectives, confirm the simulation model required for the current case, as well as the model parameters of the simulation model; Based on historical operating condition parameters, the initial parameter values ​​of the model parameters are determined and used as the parameter values. Based on the operating parameters, simulation model, and parameter values, construct the calibration case corresponding to the current case.

[0069] In one or more embodiments, based on the analysis report and a preset reinforcement learning algorithm, the parameter values ​​of the model parameters are updated, including: The analysis report is quantified into an index vector based on a pre-defined reinforcement learning algorithm; the index vector is used to describe the score corresponding to the parameter value. The indicator vector and the parameter values ​​of the model parameters are input into the reinforcement learning algorithm. The reinforcement learning algorithm constructs a reward function based on the improvement amount and updates the parameter values. The improvement amount is used to represent the difference between the parameter values ​​before and after the model parameter update.

[0070] In one or more embodiments, the iteration stopping condition includes: The improvement in the parameter values ​​of the target model is less than a preset threshold, or the improvement in the parameter values ​​of the target model reaches the target range; where the target model parameter is the key parameter in the model parameter.

[0071] In one or more embodiments, an analysis report is generated based on the results of a rapid simulation, including: Based on the results of rapid simulation, flow field cloud maps and monitoring point data are obtained; The operating parameters, simulation targets, flow field cloud maps, and monitoring point data are integrated into multimodal data; Multimodal data is input into a pre-trained multimodal large model, which is then used to identify the flow field structure and generate an analysis report.

[0072] In one or more embodiments, rapid simulation of the calibration example includes: The preset number of iteration steps or runtime; wherein the number of iteration steps or runtime is less than the number of steps or runtime required for the actual simulation; By using a flame face substitution model based on a multilayer perceptron, the calibration example is solved according to the number of iteration steps or runtime to predict the physical parameters required for the flame face in real time.

[0073] According to another aspect of the embodiments of this disclosure, a simulation device for an aircraft engine is provided, such as... Figure 4 As shown, the device includes: The acquisition module 401 is used to acquire the operating parameters and simulation targets of the current simulation example of the aero-engine, and to acquire historical operating parameters that match the operating parameters from the historical simulation database. The calibration case construction module 402 is used to construct the calibration case corresponding to the current case based on historical operating condition parameters; wherein, the calibration case includes the simulation model required for the current case, and the model parameters of the simulation model; The simulation optimization module 403 is used to perform rapid simulation on the calibration simulation, generate an analysis report based on the results of the rapid simulation, update the parameter values ​​of the model parameters based on the analysis report and the preset reinforcement learning algorithm, and obtain the updated calibration simulation. The rapid simulation is repeated on the updated calibration simulation until the iteration stopping condition is met to obtain the optimal simulation. The rapid simulation is used to quickly evaluate the effect of the current parameters of the model. Simulation module 404 is used to perform formal simulations of aero-engines based on optimal calculation examples and simulation objectives.

[0074] The simulation device and simulation method provided in this disclosure are based on the same inventive concept and have the same beneficial effects as the methods they employ, operate, or implement.

[0075] This disclosure also provides a computer device for performing the above-described simulation method. Please refer to... Figure 5 It illustrates a schematic diagram of a computer device provided by some embodiments of this disclosure. For example... Figure 5 As shown, the computer device 5 includes: a processor 500, a memory 501, a bus 502, and a communication interface 503. The processor 500, the communication interface 503, and the memory 501 are connected via the bus 502. The memory 501 stores a computer program that can run on the processor 500. When the processor 500 runs the computer program, it executes the simulation method provided in any of the foregoing embodiments of this disclosure.

[0076] The memory 501 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this device network element and at least one other network element is achieved through at least one communication interface 503 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0077] Bus 502 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. Memory 501 is used to store programs. After receiving an execution instruction, processor 500 executes the program. The simulation method disclosed in any of the foregoing embodiments of this disclosure can be applied to processor 500, or implemented by processor 500.

[0078] The processor 500 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 500 or by instructions in software form. The processor 500 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPTA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 501. The processor 500 reads the information in memory 501 and, in conjunction with its hardware, completes the steps of the above method.

[0079] The computer device and simulation method provided in this disclosure are based on the same inventive concept and have the same beneficial effects as the methods they employ, operate, or implement.

[0080] This disclosure also provides a computer-readable storage medium corresponding to the simulation method provided in the foregoing embodiments. The computer-readable storage medium is an optical disc, on which a computer program (i.e., a computer program product) is stored. When the computer program is run by a processor, it executes the simulation method provided in any of the foregoing embodiments.

[0081] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0082] The computer-readable storage medium provided in the above embodiments of this disclosure and the simulation method provided in the embodiments of this disclosure are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.

[0083] This disclosure also provides a computer program product; please refer to [reference needed]. Figure 6 The computer program product 600 carries program code, namely computer program 601. The instructions included in the computer program 601 can be used to execute the steps of the simulation method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0084] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0085] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0086] The block diagrams of devices, apparatuses, devices, and systems disclosed herein are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0087] Additionally, as used herein, the "or" used in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0088] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0089] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0090] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0091] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A simulation method for aero-engines, characterized in that, include: Obtain the operating parameters and simulation target of the current simulation case of the aero-engine, and obtain historical operating parameters that match the operating parameters from the historical simulation database; Based on the historical operating parameters, a calibration case corresponding to the current case is constructed; wherein, the calibration case includes the simulation model required for the current case, and the model parameters of the simulation model; The calibration example is subjected to rapid simulation, and an analysis report is generated based on the results of the rapid simulation. Based on the analysis report and a preset reinforcement learning algorithm, the parameter values ​​of the model parameters are updated, and the updated calibration example is obtained. The rapid simulation is repeated on the updated calibration example until the iteration stopping condition is met, and the optimal example is obtained. The rapid simulation is used to quickly evaluate the effect of the current parameters of the model. Based on the optimal computational example and the simulation objective, a formal simulation of the aero-engine is performed.

2. The simulation method as described in claim 1, characterized in that, Based on the historical operating parameters, a calibration case corresponding to the current case is constructed, including: Based on the operating parameters and the simulation objective, the simulation model required for the current case and the model parameters of the simulation model are determined. Based on the historical operating condition parameters, the initial parameter values ​​of the model parameters are determined, and the initial parameter values ​​are used as the parameter values. Based on the operating parameters, the simulation model, and the parameter values, a calibration case corresponding to the current case is constructed.

3. The simulation method as described in claim 1, characterized in that, Based on the analysis report and the preset reinforcement learning algorithm, update the parameter values ​​of the model parameters, including: The analysis report is quantified into an index vector based on a preset reinforcement learning algorithm; wherein the index vector is used to describe the score corresponding to the parameter value; The indicator vector and the parameter values ​​of the model parameters are input into the reinforcement learning algorithm. The reinforcement learning algorithm constructs a reward function based on the improvement amount and updates the parameter values. The improvement amount represents the difference between the parameter values ​​before and after the model parameter update.

4. The simulation method as described in claim 3, characterized in that, The iteration stopping conditions include: The improvement in the parameter values ​​of the target model is less than the preset threshold, or the improvement in the parameter values ​​of the target model reaches the target range; The target model parameters are the key parameters in the model parameters.

5. The simulation method as described in claim 1, characterized in that, Based on the results of the rapid simulation, an analysis report is generated, including: Based on the results of the rapid simulation, flow field cloud maps and monitoring point data are obtained; The operating parameters, the simulation target, the flow field cloud map, and the monitoring point data are integrated into multimodal data; The multimodal data is input into a pre-trained multimodal large model, which is then used to identify the flow field structure and generate an analysis report.

6. The simulation method as described in claim 1, characterized in that, Performing rapid simulation on the calibration example includes: The preset number of iteration steps or runtime is less than the number of steps or runtime required for the actual simulation. By using a flame surface substitution model based on a multilayer perceptron, the calibration example is solved according to the number of iterations or the runtime, so as to predict the physical properties required for the flame surface in real time.

7. A simulation device for aircraft engines, characterized in that, include: The acquisition module is used to acquire the operating parameters and simulation targets of the current simulation example of the aero-engine, and to acquire historical operating parameters that match the operating parameters from the historical simulation database. The calibration case construction module is used to construct a calibration case corresponding to the current case based on the historical operating condition parameters; wherein, the calibration case includes the simulation model required for the current case, and the model parameters of the simulation model; The simulation optimization module is used to perform rapid simulation on the calibration simulation, generate an analysis report based on the results of the rapid simulation, update the parameter values ​​of the model parameters based on the analysis report and a preset reinforcement learning algorithm, and obtain the updated calibration simulation. The rapid simulation is repeated on the updated calibration simulation until the iteration stopping condition is met to obtain the optimal simulation. The rapid simulation is used to quickly evaluate the effect of the current parameters of the model. The simulation module is used to perform formal simulation of the aero-engine based on the optimal calculation case and the simulation target.

8. A computer embedded device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 6.

Images