Surrogate modelling for an aircraft dynamic landing loads simulation using an LSTM AutoEncoder-based dimensionality reduction approach

Lazzara, Michele; Chevalier, Max; Colombo, Michele; García, Jasone Garay; Lapeyre, Corentin; Teste, Olivier

doi:10.1016/j.ast.2022.107629

Cited by 24 publications

(7 citation statements)

References 95 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Schwarzer et al [34] construct a neural network architecture that combines a graph convolutional neural network (GCN) with a recurrent neural network (RNN) to predict fracture propagation in brittle materials. Lazzara et al [35] proposed a dual-phase LSTM Auto-encoder-based surrogate model to predict aircraft dynamic landing response over time. Jahanbakht et al [36] presented an FEA-inspired DNN using an attention transformer to predict the sediment distribution in the wide coral reef.…”

Section: Related Workmentioning

confidence: 99%

Neuro-DynaStress: Predicting Dynamic Stress Distributions in Structural Components

Bolandi¹,

Sreekumar²,

Li³

et al. 2023

Preprint

View full text Add to dashboard Cite

Structural components are typically exposed to dynamic loading, such as earthquakes, wind, and explosions. Structural engineers should be able to conduct real-time analysis in the aftermath or during extreme disaster events requiring immediate corrections to avoid fatal failures. As a result, it is crucial to predict dynamic stress distributions during highly disruptive events in real time. Currently available high-fidelity methods, such as Finite Element Models (FEMs), suffer from their inherent high complexity and are computationally prohibitive. Therefore, to reduce computational cost while preserving accuracy, a deep learning model, Neuro-DynaStress, is proposed to predict the entire sequence of stress distribution based on finite element simulations using a partial differential equation (PDE) solver. The model was designed and trained to use the geometry, boundary conditions and sequence of loads as input and predict the sequences of high-resolution stress contours. The proposed framework's performance is compared to finite element simulations using a PDE solver.

show abstract

Section: Related Workmentioning

confidence: 99%

Neuro-DynaStress: Predicting Dynamic Stress Distributions in Structural Components

Bolandi¹,

Sreekumar²,

Li³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The operator is deterministic, but only allows to consider 2-dimensional paths with simple patterns. Lazzara et al (2022) and Zhang, Hu, and Du (2022) use autoencoders as a non-linear projection operator to extract information from high-dimensional timeseries in order to facilitate their analysis. Jarry, Couellan, and Delahaye (2019) go further into the complexity of the generation method by using Generative Adversarial Networks (GAN), which are capable of reconstructing an aircraft trajectory from a random vector of a smaller dimension.…”

Section: Literature Reviewmentioning

confidence: 99%

“…The vector 𝑦 𝑡 is here the latent representation features that embody the temporal component of the sequence. While the Recurrent Neural Network (RNN) architecture is applied for sequence modelling in most cases (Goodfellow, Bengio, & Courville, 2016;Lazzara et al, 2022), Bai, Kolter, and Koltun (2018) suggest that CNN should also be considered a legitimate option due to many reasons: they are less complicated, less exposed to exploding or vanishing gradients, allow for parallel computation of outputs (unlike RNN), and can achieve cutting-edge performance. The authors exhibit a family of architectures called Temporal Convolutional Networks (TCN) that adapt general CNN for sequence modelling tasks.…”

Section: Temporal Convolutional Networkmentioning

confidence: 99%

Deep Generative Modelling of Aircraft Trajectories in Terminal Maneuvering Areas

et al. 2022

View full text Add to dashboard Cite

“…We can notice the trend to replace classical hydraulic drives with electro-hydrostatic or electro-mechanical drives, aiming the elimination of the centralized hydraulic system. In reference [10], the authors present a study concerning dynamic loads in landing phase. Emergency deploying systems for landing gears are in [11].…”

Section: Introductionmentioning

confidence: 99%

Drive Systems for Landing Gears Numerically Simulated in AMESim Software

Dincă

Corcau

Vaden

2023

Modelling and Simulation in Engineering

View full text Add to dashboard Cite

This work presents two versions of landing gear drives numerically simulated. We follow to obtain the right driving sequence of the legs and doors and the estimation of the landing gear deploy and retract time. Although the studies in this paper do not exhaust all the aspects concerning landing gear hydraulic system functioning, these are useful in the design phase of the aircraft hydraulic systems. Regarding the landing gear hydraulic system, we highlight one version of the drive stall phenomenon. When the hydraulic feeding pressure decreases, it is possible that the forces developed by the hydraulic cylinders are enough to retract completely the legs. The landing gear remains in an intermediate position, and the aircraft cannot accomplish its mission safely. For a second version, logic command functions are synthesized using Karnaugh diagrams, and after that, the numerical simulation is accomplished to validate the solution. We used the AMESim software to perform numerical simulations.

show abstract

Surrogate modelling for an aircraft dynamic landing loads simulation using an LSTM AutoEncoder-based dimensionality reduction approach

Cited by 24 publications

References 95 publications

Neuro-DynaStress: Predicting Dynamic Stress Distributions in Structural Components

Neuro-DynaStress: Predicting Dynamic Stress Distributions in Structural Components

Deep Generative Modelling of Aircraft Trajectories in Terminal Maneuvering Areas

Drive Systems for Landing Gears Numerically Simulated in AMESim Software

Contact Info

Product

Resources

About