CN-121997460-A - Method for optimizing gradient acceleration of wall thickness of water-in shock-resistant aluminum alloy honeycomb of aircraft

CN121997460ACN 121997460 ACN121997460 ACN 121997460ACN-121997460-A

Abstract

The invention discloses a wall thickness gradient acceleration optimization method of a water-in shock-resistant aluminum alloy honeycomb of an aircraft, which comprises the steps of selecting an optimal shape from aluminum alloy honeycombs of quadrilateral, hexagonal and circular three honeycomb units, axially layering, generating a plurality of groups of honeycomb unit wall thickness gradient samples, constructing an optimal shape aluminum alloy honeycomb according to each group of honeycomb unit wall thickness gradient samples, constructing an aircraft fluid-solid coupling simulation model, carrying out impact response simulation at preset water inlet speed to obtain a real sample data set, expanding the real sample data set, training and verifying a GAN-ANN proxy model, embedding the GAN-ANN proxy model into a multi-objective optimization model, minimizing peak acceleration and the quality of the aluminum alloy honeycomb as targets, meeting peak stress constraint, and carrying out iterative solution to obtain the optimal wall thickness of each layer of honeycomb unit. The optimized aluminum alloy honeycomb wall can not only effectively slow down the impact load of water and reduce the stress borne by an aircraft main body, but also realize the light weight of the structure.

Inventors

JIN HAO
SHENG YUCHAO
SHAO CHUN

Assignees

杭州电子科技大学

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (9)

1. The method for optimizing the gradient acceleration of the wall thickness of the water-in shock-resistant aluminum alloy honeycomb of the aircraft is characterized by comprising the following specific steps of: step S1, designing aluminum alloy honeycombs of quadrilateral, hexagonal and circular three honeycomb units, respectively constructing an aircraft fluid-solid coupling simulation model, and configuring a fairing, the aluminum alloy honeycombs and a material constitutive model and a failure criterion of an aircraft main body; S2, performing impulse response simulation at preset water inflow speed on each aircraft fluid-solid coupling simulation model processed in the step S1, selecting an optimal aluminum alloy honeycomb, axially layering the optimal aluminum alloy honeycomb, generating a plurality of groups of honeycomb unit wall thickness gradient samples, constructing the optimal aluminum alloy honeycomb according to each group of honeycomb unit wall thickness gradient samples, constructing an aircraft fluid-solid coupling simulation model, and performing impulse response simulation at preset water inflow speed to obtain a real sample data set; And S3, expanding a real sample data set, training and verifying a GAN-ANN proxy model, embedding the model into a multi-objective optimization model, taking the peak acceleration and the quality minimization of the aluminum alloy honeycomb as targets, meeting peak stress constraint, and carrying out iterative solution to obtain the optimal wall thickness of each layer of honeycomb unit.
2. The method for optimizing the gradient acceleration of the wall thickness of the water-in impact-resistant aluminum alloy honeycomb of the aircraft according to claim 1 is characterized in that the step S1 specifically comprises the following steps: S11, designing three aluminum alloy honeycombs with quadrilateral, hexagonal and circular honeycomb units, respectively constructing an aircraft fluid-solid coupling simulation model, wherein the quality consistency of the three aluminum alloy honeycombs is controlled by adjusting the wall thickness of the honeycomb unit of each aluminum alloy honeycomb; the construction process of the fluid-solid coupling simulation model of the aircraft comprises the steps of constructing a solid structural domain of the model of the aircraft, which comprises a fairing, an aluminum alloy honeycomb and an aircraft main body, constructing a fluid domain, which comprises a water area and an air area, setting fluid boundary conditions of the fluid domain, and simulating an infinite flow field environment, thereby obtaining the fluid-solid coupling simulation model of the aircraft; Step S12, selecting an aircraft fluid-solid coupling simulation model constructed according to an aluminum alloy honeycomb with hexagonal honeycomb units, selecting grid unit types of a fluid domain and an aircraft model solid structural domain, respectively performing grid division and grid independence verification, and finding out an optimal solution of the grid unit types of the fluid domain and the aircraft model solid structural domain; s13, adopting a grid unit type optimal solution of a fluid domain and an aircraft model solid structural domain to grid-divide the fluid domain and the aircraft model solid structural domain of each aircraft fluid-solid coupling simulation model; Step S14, configuring a material constitutive model and a failure criterion aiming at each aircraft fluid-solid coupling simulation model, wherein the fairing is configured with an elastoplastic constitutive model by adopting a maximum strain failure criterion, the aircraft main body is configured with a line elastoplastic constitutive model, the failure criterion is whether stress distribution exceeds yield strength, and the failure criterion is a multi-criterion coupling failure model covering ductile fracture, shear fracture and necking instability.
3. The method for optimizing the gradient acceleration of the wall thickness of the water-in impact-resistant aluminum alloy honeycomb of the aircraft according to claim 2 is characterized in that the step S2 specifically comprises the following steps: s21, carrying out impulse response simulation on each aircraft fluid-solid coupling simulation model processed in the step S1 at a preset water inlet speed, obtaining peak acceleration and peak stress corresponding to each aircraft fluid-solid coupling simulation model, and selecting an aluminum alloy honeycomb with the minimum peak acceleration and the peak stress smaller than a threshold value as an optimal aluminum alloy honeycomb; S22, axially layering the optimal aluminum alloy honeycomb, sampling the wall thickness of the honeycomb unit of each layer within a set value range, generating a plurality of groups of wall thickness gradient samples of the honeycomb unit, constructing the optimal aluminum alloy honeycomb according to the wall thickness gradient samples of each group of honeycomb unit, constructing an aircraft fluid-solid coupling simulation model, and performing impact response simulation at a preset water inlet speed to obtain a real sample data set.
4. The method for optimizing the gradient acceleration of the wall thickness of a water-in impact-resistant aluminum alloy honeycomb of an aircraft according to claim 3, wherein the preset water-in speed is 50m/s, 100m/s, 150m/s or 200m/s.
5. A method for optimizing wall thickness gradient of water-in shock-resistant aluminum alloy honeycomb of an aircraft comprises the following steps of dividing an optimal aluminum alloy honeycomb into nine equidistant layers along the axial direction, setting the wall thickness range of each layer to be 0.1 mm-0.8 mm, uniformly generating 1000 groups of different wall thickness gradient samples of the honeycomb units in a wall thickness design space by adopting an optimal Latin hypercube sampling method, endowing the wall thickness of each group of wall thickness gradient samples of the honeycomb units with the wall thickness of the optimal aluminum alloy honeycomb to construct a corresponding aircraft fluid-solid coupling simulation model, executing steps S13 and S14, finally performing impact response simulation under preset water-in speed, extracting corresponding peak acceleration and peak stress after the impact response simulation of the aircraft fluid-solid coupling simulation model corresponding to each group of the wall thickness gradient samples of the honeycomb units is completed, and combining the quality of the optimal aluminum alloy honeycomb corresponding to the wall thickness gradient samples of the honeycomb units and the wall thickness of the honeycomb units to form a real sample data set.
6. The method for optimizing the gradient acceleration of the wall thickness of the water-in impact-resistant aluminum alloy honeycomb of the aircraft according to claim 1, wherein the step S3 specifically comprises the following steps: Step S31, a generating type countermeasure network consisting of a generator G and a discriminator D is constructed, training is carried out, a virtual sample is generated by the trained generator G, and the virtual sample and a real sample are combined into an extended data set; Step S32, dividing the extended data set into a training set, a verification set and a test set, constructing an ANN agent model, inputting 9 nodes of a layer, corresponding to wall thickness parameters of 9 layers of honeycomb units, outputting 2 nodes of a layer, corresponding to peak acceleration and peak stress, adopting a Bayesian regularized back propagation algorithm to optimize the ANN agent model, taking mean square error as a loss function, monitoring error through the verification set after each iteration, stopping training if the error does not drop continuously for multiple times, and obtaining a trained GAN-ANN agent model; S33, constructing a multi-objective optimization model, designing an optimization objective to minimize peak acceleration and aluminum alloy honeycomb quality, wherein the constraint condition is that peak stress is smaller than a preset value, the design variable is 9 layers of honeycomb unit wall thickness parameters, the value range is 0.1 mm-0.8 mm, and a non-dominant ranking genetic algorithm II is selected as an optimization algorithm; In the step S34, 16 initial populations are randomly generated in a designed variable value range in the multi-objective optimization model, a new generation population is generated through selection, intersection and variation operations, a GAN-ANN proxy model is called to predict peak acceleration and peak stress of each population, individuals meeting the peak stress smaller than a preset value in each population are ordered according to a non-dominant ordering genetic algorithm II, then individuals which can be subjected to non-dominant ordering of each population are extracted, no other individuals can be subjected to dominant ordering, the sets of the individuals are pareto fronts, the new population and the historical pareto fronts are combined in an iterative process and are ordered according to the non-dominant ordering genetic algorithm II again to obtain the new pareto fronts, convergence is judged when the rate of the pareto fronts exceeds the volume change rate by more than 1% continuously, and from the final pareto fronts, the individual with the minimum Euclidean distance with ideal points of peak acceleration and aluminum alloy honeycomb quality of 0 is selected as the optimal wall thickness of honeycomb units of each layer.
7. The method for accelerating and optimizing the wall thickness gradient of the water-in impact-resistant aluminum alloy honeycomb of the aircraft is characterized in that a CNN model is adopted by the generator and the discriminator, a standard normal distribution random noise vector Z is taken as an input by the generator, the generator is converted into 1X 3X 128 tensors through a projection layer, then the 1X 11X 1 sample vector is finally output through the up sampling of a plurality of groups of transposed convolution layers, normalization layers and LeakyReLU activation functions, the output characteristic value is constrained to [ -1,1] to match a real sample normalization range, the normalized real sample data set and the 1X 11X 1 data vector output by the generator are taken as the input by the discriminator, the characteristics are extracted through three layers of one-dimensional convolution layers, leakyReLU activation functions and a Dropout layer, and finally the probability that an output sample is a real sample is activated through Sigmoid.
8. The method for accelerating and optimizing the wall thickness gradient of the water-in impact-resistant aluminum alloy honeycomb of the aircraft according to claim 6, wherein the generating type countermeasure network adopts an Adam optimization algorithm, and the generator minimization loss function and the arbiter maximization loss function are as follows: Wherein, the The representation generator minimizes the loss function and, The standard normal distribution is represented by the formula, The samples generated by the generator are represented, The representation arbiter gives the probability that the sample generated by the generator is a true sample, Representation of Is not limited to the desired one; Representing that the arbiter maximizes the loss function, The representation arbiter gives the probability that x is a true sample, The probability distribution of x is represented by, Representation of Is expected to be improved.
9. The method for accelerating and optimizing the wall thickness gradient of the water-in impact-resistant aluminum alloy honeycomb of the aircraft according to claim 6, wherein the expression of the multi-objective optimization model is as follows: in the formula, Represents the axial i-th layer honeycomb unit wall thickness of the aluminum alloy honeycomb, Representing the peak acceleration, m representing the mass of the aluminum alloy honeycomb, Representing peak stress.

Description

Method for optimizing gradient acceleration of wall thickness of water-in shock-resistant aluminum alloy honeycomb of aircraft Technical Field The invention belongs to the field of water-in anti-impact honeycomb optimization, and particularly relates to an aircraft water-in anti-impact aluminum alloy honeycomb wall thickness gradient acceleration optimization method based on small sample machine learning. Background The main operational environments of cross-domain aircraft bodies include air flight and underwater navigation. The sharp increase in media density during the transition from air to water subjects the structure to transient, high amplitude impact loads. These loads can cause severe deformation, yielding, collapse or breakage of the main structure and irreversible damage to sensitive on-board equipment, compromising safety and function. Thus, reducing impact loads during high speed water entry for cross-domain aircraft is a critical engineering challenge. Disclosure of Invention The invention provides an aircraft water-in shock-resistant aluminum alloy honeycomb wall thickness gradient acceleration optimization method based on small sample machine learning for solving the problems in the prior art. The technical scheme adopted by the invention is as follows: The invention relates to a method for optimizing gradient acceleration of a wall thickness of a water-in shock-resistant aluminum alloy honeycomb of an aircraft, which comprises the following steps: step S1, designing aluminum alloy honeycombs of quadrilateral, hexagonal and circular three honeycomb units, respectively constructing an aircraft fluid-solid coupling simulation model, and configuring a fairing, the aluminum alloy honeycombs and a material constitutive model and a failure criterion of an aircraft main body; S2, performing impulse response simulation at preset water inflow speed on each aircraft fluid-solid coupling simulation model processed in the step S1, selecting an optimal aluminum alloy honeycomb, axially layering the optimal aluminum alloy honeycomb, generating a plurality of groups of honeycomb unit wall thickness gradient samples, constructing the optimal aluminum alloy honeycomb according to each group of honeycomb unit wall thickness gradient samples, constructing an aircraft fluid-solid coupling simulation model, and performing impulse response simulation at preset water inflow speed to obtain a real sample data set; And S3, expanding a real sample data set, training and verifying a GAN-ANN proxy model, embedding the model into a multi-objective optimization model, taking the peak acceleration and the quality minimization of the aluminum alloy honeycomb as targets, meeting peak stress constraint, and carrying out iterative solution to obtain the optimal wall thickness of each layer of honeycomb unit. Preferably, step S1 specifically includes: And S11, designing three aluminum alloy honeycombs with quadrangular, hexagonal and circular honeycomb units, and respectively constructing an aircraft fluid-solid coupling simulation model, wherein the quality consistency of the three aluminum alloy honeycombs is controlled by adjusting the wall thickness of the honeycomb unit of each aluminum alloy honeycomb. The construction process of the fluid-solid coupling simulation model of the aircraft comprises the steps of constructing a solid structural domain of the model of the aircraft comprising a fairing, an aluminum alloy honeycomb and an aircraft body, constructing a fluid domain comprising a water area and an air domain, setting fluid boundary conditions of the fluid domain, and simulating an infinite flow field environment, thereby obtaining the fluid-solid coupling simulation model of the aircraft. And S12, selecting an aircraft fluid-solid coupling simulation model constructed according to an aluminum alloy honeycomb with hexagonal honeycomb units, selecting grid unit types of a fluid domain and an aircraft model solid structural domain, respectively performing grid division and grid independence verification, and finding out an optimal solution of the grid unit types of the fluid domain and the aircraft model solid structural domain. And S13, performing grid division on the fluid domain and the solid structural domain of the aircraft fluid-solid coupling simulation model by adopting the optimal solution of the grid cell type of the fluid domain and the solid structural domain of the aircraft model. Step S14, configuring a material constitutive model and a failure criterion aiming at each aircraft fluid-solid coupling simulation model, wherein the fairing is configured with an elastoplastic constitutive model by adopting a maximum strain failure criterion, the aircraft main body is configured with a line elastoplastic constitutive model, the failure criterion is whether stress distribution exceeds yield strength, and the failure criterion is a multi-criterion coupling failure model covering ductile fracture, shear fracture