CN-122021428-A - Carrier-based aircraft dynamic boundary coupling simulation method and system based on neural network model

CN122021428ACN 122021428 ACN122021428 ACN 122021428ACN-122021428-A

Abstract

The invention relates to a neural network model-based carrier-based aircraft dynamic boundary coupling simulation method and system, and belongs to the technical field of computational fluid dynamics simulation. The method comprises the steps of training an engine neural network proxy model based on test or simulation data, initializing a CFD model and obtaining initial flow information, performing CFD transient calculation and extracting parameters of an outlet flow field of an air inlet channel, inputting the parameters of the outlet flow field of the air inlet channel into the proxy model for time, predicting an updated value of a boundary condition of the engine, giving the predicted value to the corresponding boundary of the CFD, reversely calculating and updating the geometrical profile of a spray pipe according to a preset rule or parameters, and performing loop iteration until the end. The invention realizes the real-time dynamic closed-loop updating of boundary conditions and the geometrical profile of the spray pipe, effectively solves the problems of boundary rigidness, neglecting dynamic feedback, efficiency and accuracy contradiction in the traditional simulation, remarkably improves the authenticity and efficiency of the simulation of the air inlet channel-engine coupling system, and is suitable for temperature distortion prediction and safety evaluation.

Inventors

QIAO WENYOU
DENG YONGHAO
CHE JIEXIAN
CHEN KUIYU
Zeng Daoming
Han Longxin
ZHOU CHUANJIANG
Jiang Zhouyu
YE WEI

Assignees

西南科技大学

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (6)

1. The carrier-based aircraft dynamic boundary coupling simulation method based on the neural network model is characterized by comprising the following steps of: the preparation stage comprises training a neural network agent model based on real operation test data or simulation data of a carrier aircraft engine; Initializing a CFD simulation model comprising a carrier-based aircraft, a deck and a bias flow plate, setting initial boundary conditions and initial nozzle geometric profiles, and performing steady calculation based on the slow engine state to obtain an initial flow field of unsteady simulation; S1, a CFD solver performs calculation of a time step delta t based on an initial boundary condition and an initial spray pipe geometric profile under an initial flow field of unsteady simulation, and extracts total temperature and total pressure of an outlet section of an air inlet channel from a result of the whole calculation domain after the time step delta t is calculated; S2, inputting the total temperature, the total pressure and the current simulation time of the outlet section of the air inlet channel extracted in the step S1 into the neural network proxy model; s3, the neural network agent model predicts and outputs the outlet flow of the air inlet channel and the static pressure, the total temperature and the flow of the inlet of the spray pipe corresponding to the current simulation time according to the total temperature, the total pressure and the current simulation time of the outlet section of the air inlet channel; S4, inputting the inlet channel outlet flow and the jet pipe inlet static pressure, the total temperature and the flow which correspond to the current simulation time obtained in the step S3 into the corresponding inlet channel outlet and jet pipe inlet in the CFD simulation model to update corresponding boundary conditions; S5, returning to the step S1, and performing loop iteration until the simulation is finished.
2. The method according to claim 1, wherein the time-dependent change of the predetermined nozzle throat and outlet area in step S4 is derived from test data by extracting a time-dependent curve of the nozzle throat and outlet cross-sectional area during the test.
3. The method according to claim 1, wherein the nozzle throat and outlet area change law with time calculated by the parameter in step S4 is achieved by: The logic of obtaining the throat area A 8 and the outlet area A 9 of the spray pipe according to the real-time inverse calculation of the working state of the engine in accordance with the pneumatic function relation is that firstly, the throat area A 8 is directly solved according to the critical choking condition by utilizing the total pressure Pt 7 , the total temperature Tt 7 , the mass flow W 7 , the specific heat ratio gamma and the gas constant R of the inlet of the spray pipe, then the Mach number Ma 9 of the outlet of the spray pipe is calculated with the complete expansion as a target based on the determined A 8 、Pt 7 and the environmental backpressure P b , and finally the required outlet area A 9 is calculated through an isentropic area ratio formula.
4. The method of claim 1, wherein the inlet channel outlet flow and the jet inlet static pressure, the total temperature and the flow corresponding to the current simulation time obtained in the step S4 are input to the corresponding inlet channel outlet and jet inlet in the CFD simulation model by calling macro operation or script provided by the CFD solver, and the step S4 is also implemented by calling the movable grid function of the CFD solver by updating the throat and outlet area of the jet pipe in the CFD simulation model.
5. The method of claim 1, wherein the loop iteration process of steps S1 to S5 is performed by external flow integration and automated platform scheduling.
6. A carrier-borne dynamic boundary coupling simulation system based on a neural network proxy model, which adopts the simulation method as set forth in any one of claims 1 to 5, and is characterized by comprising: the data acquisition module is used for extracting the total temperature and the total pressure of the outlet section of the air inlet channel from the CFD solving result; The agent model prediction module is used for receiving the total temperature, the total pressure and the current simulation time of the outlet section of the air inlet channel, calling the neural network agent model to predict and output the outlet flow of the air inlet channel, the static pressure, the total temperature and the flow of the inlet of the spray pipe corresponding to the current simulation time; The boundary condition updating and spray pipe geometric profile updating module is used for inputting the output inlet channel outlet flow corresponding to the current simulation time and the spray pipe inlet static pressure, total temperature and flow to the inlet channel outlet section and the spray pipe inlet section of the CFD model, updating the corresponding boundary conditions, and updating the throat and outlet area of the spray pipe in the CFD simulation model according to the change rule of the throat and outlet area of the spray pipe with time, which is calculated reversely according to the preset or parameter, namely updating the geometric profile of the spray pipe; And the circulation control module is used for controlling the data acquisition module, the agent model prediction module, the boundary condition updating and the spray pipe geometric profile updating module to perform circulation iteration.

Description

Carrier-based aircraft dynamic boundary coupling simulation method and system based on neural network model Technical Field The invention relates to the technical field of intersection of aviation engineering and high-performance numerical simulation, in particular to the field of ship-based aircraft surface environment simulation and engine air intake and exhaust system safety analysis. Background In the current carrier-based aircraft take-off process, the high Wen Wei jet flow of the engine impacts the flow deflector to form a complex reflection flow field, and the flow field is sucked by the air inlet channel to cause serious distortion of the outlet section of the air inlet channel, so that the stability and the flight safety of the engine are directly endangered. The existing numerical simulation method has a remarkable bottleneck in the face of simulation of the jet-bias flow plate-air inlet channel-engine strong coupling system. The traditional method has low calculation efficiency, and the transient simulation of the whole system is too long to meet the iteration requirement of engineering design. More importantly, the existing method lacks the dynamic coupling capability of the system, usually adopts preset static boundary conditions, and cannot reflect real-time interaction between an air inlet system and an air outlet system, and cannot process the influence caused by the state change of an engine and the dynamic change of the geometrical profile of a spray pipe, so that the simulation confidence is limited. Aiming at the problems, an innovative simulation method and system are provided. The method and the system realize the unification of high efficiency and high fidelity, can accurately capture the process of the strong coupling system of jet flow, bias flow plate, air inlet channel and engine, update boundary conditions and the geometrical profile of the jet pipe in real time, and remarkably improve the accuracy and efficiency of the safety assessment of the air inlet and outlet system of the carrier-based aircraft. Disclosure of Invention The existing numerical simulation method has obvious defects in the aspects of calculation efficiency, system integration level and dynamic feedback mechanism when processing a complex system of strong coupling, unsteady and dynamic spray pipe geometric profile change of a 'jet flow-bias flow plate-air inlet channel-engine'. Therefore, development of an innovative simulation method and system which have high computing efficiency and high fidelity and can truly reflect the dynamic interaction of the system and the deformation of the geometrical profile of the spray pipe is needed to support the efficient design and safety evaluation of the air intake and exhaust system of the carrier aircraft. In order to achieve the above purpose, the technical scheme of the invention is as follows: The first object of the invention is to provide a carrier aircraft dynamic boundary coupling simulation method based on a neural network model, which comprises the following steps: the preparation stage comprises training a neural network agent model based on real operation test data or simulation data of a carrier aircraft engine; initializing CFD simulation models comprising a carrier-based aircraft, a deck and a bias flow plate, setting initial boundary conditions and an initial geometrical profile of a spray pipe, performing steady calculation based on the slow-running state of the engine, and obtaining an initial flow field of unsteady simulation; S1, a CFD solver performs calculation of a time step delta t based on an initial boundary condition and an initial spray pipe geometric profile under an initial flow field of unsteady simulation, and extracts total temperature and total pressure of an outlet section of an air inlet channel from a result of the whole calculation domain after the time step delta t is calculated; S2, inputting the total temperature, the total pressure and the current simulation time of the outlet section of the air inlet channel extracted in the step S1 into the neural network proxy model; s3, the neural network agent model predicts and outputs the outlet flow of the air inlet channel and the static pressure, the total temperature and the flow of the inlet of the spray pipe corresponding to the current simulation time according to the total temperature, the total pressure and the current simulation time of the outlet section of the air inlet channel; S4, inputting the inlet channel outlet flow and the jet pipe inlet static pressure, the total temperature and the flow which correspond to the current simulation time obtained in the step S3 into the corresponding inlet channel outlet and jet pipe inlet in the CFD simulation model to update corresponding boundary conditions; S5, returning to the step S1, and performing loop iteration until the simulation is finished. Preferably, the time-dependent change rule of the areas of the nozzle throat and the outl