Article
Computer Science, Software Engineering
Aaditya Chandrasekhar, Krishnan Suresh
Summary: In this paper, an approximate length scale filter strategy for topology optimization (TO) is proposed by extending a density-based TO formulation using neural networks (TOuNN). The proposed method enhances TOuNN with a Fourier space projection to approximately control the minimum and/or maximum length scales. The method does not involve additional constraints and automates sensitivity computations using the neural net's library.
COMPUTER-AIDED DESIGN
(2022)
Article
Mathematics, Interdisciplinary Applications
Chunpeng Wang, Yanping Lian, Ruxin Gao, Feiyu Xiong, Ming-Jian Li
Summary: This paper proposes an artificial neural network (ANN)-based structural topology optimization method that combines traditional gradient-based methods and population-based approaches to efficiently achieve converged design or design diversity. The ANN is used as the structural descriptor and integrated with the solid isotropic material with penalization (SIMP) framework. The weights and bias associated with the ANN are optimized via back-propagation, and novel loss functions are proposed to encode structural performances and topology awareness.
COMPUTATIONAL MECHANICS
(2023)
Article
Engineering, Multidisciplinary
Zeyu Zhang, Yu Li, Weien Zhou, Xiaoqian Chen, Wen Yao, Yong Zhao
Summary: The paper introduces a method for topology optimization using neural networks, the TONR framework, which allows flexible design variable updates and sensitivity analysis. With this approach, optimized structures for different optimization problems can be obtained stably without the need to construct a dataset beforehand.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2021)
Article
Chemistry, Multidisciplinary
Dejun Jiang, Zhaofeng Ye, Chang-Yu Hsieh, Ziyi Yang, Xujun Zhang, Yu Kang, Hongyan Du, Zhenxing Wu, Jike Wang, Yundian Zeng, Haotian Zhang, Xiaorui Wang, Mingyang Wang, Xiaojun Yao, Shengyu Zhang, Jian Wu, Tingjun Hou
Summary: This study compiled the largest metalloprotein-ligand complex dataset and evaluated the docking powers of three competitive docking tools for metalloproteins. A structure-based deep graph model called MetalProGNet was developed to predict metalloprotein-ligand interactions. MetalProGNet outperformed various baselines in internal and external evaluations. The study also employed an atom-atom interaction masking technique to interpret MetalProGNet.
Article
Computer Science, Interdisciplinary Applications
Gorkem Can Ates, Recep M. Gorguluarslan
Summary: This study proposes a two-stage network model for topology optimization, which effectively reduces structural disconnections and pixel-wise errors, enhancing the predictive performance of DNNs. The optimized framework improves network prediction ability while significantly reducing compliance and volume fraction errors.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2021)
Article
Engineering, Civil
Hyogu Jeong, Jinshuai Bai, C. P. Batuwatta-Gamage, Charith Rathnayaka, Ying Zhou, YuanTong Gu
Summary: This paper proposes a novel topology optimization framework called PINNTO, which uses a energy-based PINN to determine deformation states in structural topology optimization. It trains a supervised neural network to respect governing physical laws defined via partial differential equations. The results show that PINNTO can achieve optimized topologies without labeled data or FEA, and can generate comparable designs to SIMP. The framework demonstrates promising capabilities to solve problems when the usage of FEA is challenging or impossible.
ENGINEERING STRUCTURES
(2023)
Article
Computer Science, Interdisciplinary Applications
Jun Yan, Dongling Geng, Qi Xu, Haijiang Li
Summary: This paper proposes a step-to-step training method to improve the prediction accuracy of a deep learning model for real-time structural topology optimization. By increasing the utilization of optimization history information, the method improves the efficiency of model utilization without increasing the sample set size.
ENGINEERING WITH COMPUTERS
(2023)
Article
Computer Science, Interdisciplinary Applications
Van Bang Dinh, Ngoc Le Chau, Nam T. P. Le, Thanh-Phong Dao
Summary: This paper proposes a new multi-phases optimization design method for compliant mechanisms, integrating topology optimization with finite element method, intelligent modeling, and neural network algorithm. The method is used to design a new compliant mechanism and predict its behavior by optimizing parameters. The results of size optimization show that the proposed method has higher accuracy and efficiency compared to other methods.
ENGINEERING WITH COMPUTERS
(2021)
Article
Computer Science, Interdisciplinary Applications
Ren Kai Tan, Chao Qian, Kangjie Li, Dan Xu, Wenjing Ye
Summary: Topology optimization is a systematic approach for obtaining optimal performance in structural design, but it can be computationally expensive and deep learning models lack generalizability. This work proposes an adaptive, scalable deep learning-based model-order-reduction method using MapNet to accelerate large-scale topology optimization. The method allows simulations to be performed at a coarser mesh, reducing computational time, and introduces domain fragmentation to improve the method's transferability and scalability.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2022)
Article
Engineering, Multidisciplinary
Zeyu Zhang, Wen Yao, Yu Li, Weien Zhou, Xiaoqian Chen
Summary: With the rapid development of artificial intelligence (AI) technology, scientific research has entered a new era of AI. The cross development between topology optimization (TO) and AI technology has been receiving continuous attention. This paper introduces the concept of Implicit Neural Representations from AI into the TO field and establishes a novel TO framework called TOINR.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2023)
Article
Agricultural Engineering
Yiping Li, Linda A. Nuamah, Yashuai Pu, Haikuo Zhang, Eyram Norgbey, Amechi S. Nwankwegu, Patrick Banahene, Robert Bofah
Summary: In this study, an RBFNN model was used to optimize the hydraulic loading rate, hydraulic retention time, and mass loading rates for enhanced removal of nitrogen and phosphorus in a treatment wetland system. The model achieved a high accuracy in predicting total nitrogen and total phosphorus removal efficiencies, suggesting the feasibility of using RBFNN modelling for optimizing treatment wetlands.
BIORESOURCE TECHNOLOGY
(2022)
Article
Computer Science, Interdisciplinary Applications
Yi Xing, Liyong Tong
Summary: In this work, a machine learning-assisted structural optimization (MLaSO) scheme is proposed to accelerate the computational speed of structural optimization. A new machine learning model is used to predict the update of the optimization quantity during the optimization process, eliminating the need for finite element analysis and sensitivity analysis. The MLaSO scheme can be easily integrated into different structural optimization methods and does not require additional training datasets.
STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
(2022)
Article
Telecommunications
Guillermo Bernardez, Jose Suarez-Varela, Albert Lopez, Xiang Shi, Shihan Xiao, Xiangle Cheng, Pere Barlet-Ros, Albert Cabellos-Aparicio
Summary: This paper introduces a distributed ML-based framework called MAGNNETO, which leverages Multi-Agent Reinforcement Learning and Graph Neural Networks to solve Traffic Engineering optimization problems in ISP networks. Experimental results show that MAGNNETO achieves comparable performance to existing optimizers with significantly lower execution times, and demonstrates strong generalization capability.
IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING
(2023)
Article
Thermodynamics
Ali Sohani, Siamak Hoseinzadeh, Saman Samiezadeh, Ivan Verhaert
Summary: An enhanced design for a solar still desalination system was employed to develop artificial neural network (ANN) models, with FF and RBF types identified as the best structures for predicting distillate production and water temperature. Error analysis on data not used for ANN model development showed varying errors in different months.
JOURNAL OF THERMAL ANALYSIS AND CALORIMETRY
(2022)
Article
Chemistry, Physical
Praveen S. Vulimiri, Hao Deng, Florian Dugast, Xiaoli Zhang, Albert C. To
Summary: This research introduces a novel topology optimization method that uses neural style transfer to optimize both structural performance and geometric similarity. The user can control the influence of neural style transfer on structural performance, allowing for an ideal compromise in design.