• Transactions of Materials Processing
• /
• v.21 no.8
• /
• pp.479-484
• /
• 2012

The objective of this study is to compensate the elastic recovery in the flexible forming process using the predictive models. The target shape was limited to two-dimensional shape having only one curvature radius in the longitudinal-direction. In order to predict the shape error the regression and neural network models were established based on the finite element (FE) simulations. A series of simulations were conducted considering input variables such as the elastic pad thickness, the thickness of plate, and the objective curvature radius. Then, at sampling points in the longitudinal-direction, the shape errors between formed and objective shapes could be calculated from the FE simulations as an output variable. These shape errors were expressed to a representative error value by the root mean square error (RMSE). To obtain the correct objective shape the die shape was adjusted by the closed-loop using the neural network model since the neural network model shows a higher capability of estimating the shape error than the regression model. Finally the experimental result shows that the formed shape almost agreed with the objective shape.

• Structural Engineering and Mechanics
• /
• v.39 no.2
• /
• pp.267-285
• /
• 2011

There can be a significant amount of moment redistribution in composite frames consisting of steel columns and composite beams, due to cracking, creep and shrinkage of concrete. Considerable amount of computational effort is required for taking into account these effects for large composite frames. A methodology has been presented in this paper for taking into account these effects. In the methodology that has been demonstrated for moderately high frames, neural network models are developed for rapid prediction of the inelastic moments (typically for 20 years, considering instantaneous cracking, and time effects, i.e., creep and shrinkage, in concrete) at a joint in a frame from the elastic moments (neglecting instantaneous cracking and time effects). The proposed models predict the inelastic moment ratios (ratio of elastic moment to inelastic moment) using eleven input parameters for interior joints and seven input parameters for exterior joints. The training and testing data sets are generated using a hybrid procedure developed by the authors. The neural network models have been validated for frames of different number of spans and storeys. The models drastically reduce the computational effort and predict the inelastic moments, with reasonable accuracy for practical purposes, from the elastic moments, that can be obtained from any of the readily available software.

• International Journal of Control, Automation, and Systems
• /
• v.4 no.3
• /
• pp.382-393
• /
• 2006

Mechanical system dynamics plays an important role in the area of computational structural biology. Elastic network models (ENMs) for macromolecules (e.g., polymers, proteins, and nucleic acids such as DNA and RNA) have been developed to understand the relationship between their structure and biological function. For example. a protein, which is basically a folded polypeptide chain, can be simply modeled as a mass-spring system from the mechanical viewpoint. Since the conformational flexibility of a protein is dominantly subject to its chemical bond interactions (e.g., covalent bonds, salt bridges, and hydrogen bonds), these constraints can be modeled as linear spring connections between spatially proximal representatives in a variety of coarse-grained ENMs. Coarse-graining approaches enable one to simulate harmonic and anharmonic motions of large macromolecules in a PC, while all-atom based molecular dynamics (MD) simulation has been conventionally performed with an aid of supercomputer. A harmonic analysis of a macroscopic mechanical system, called normal mode analysis, has been adopted to analyze thermal fluctuations of a microscopic biological system around its equilibrium state. Furthermore, a structure-based system optimization, called elastic network interpolation, has been developed to predict nonlinear transition (or folding) pathways between two different functional states of a same macromolecule. The good agreement of simulation and experiment allows the employment of coarse-grained ENMs as a versatile tool for the study of macromolecular dynamics.

• Computers and Concrete
• /
• v.24 no.6
• /
• pp.541-553
• /
• 2019

Alkali-silica reaction (ASR) in concrete can induce degradation in its mechanical properties, leading to compromised serviceability and even loss in load capacity of concrete structures. Compared to other properties, ASR often affects the modulus of elasticity more significantly. Several empirical models have thus been established to estimate elastic modulus reduction based on the ASR expansion only for condition assessment and capacity evaluation of the distressed structures. However, it has been observed from experimental studies in the literature that for any given level of ASR expansion, there are significant variations on the measured modulus of elasticity. In fact, many other factors, such as cement content, reactive aggregate type, exposure condition, additional alkali and concrete strength, have been commonly known in contribution to changes of concrete elastic modulus due to ASR. In this study, an artificial intelligent model using artificial neural network (ANN) is proposed for the first time to provide an innovative approach for evaluation of the elastic modulus of ASR-affected concrete, which is able to take into account contribution of several influence factors. By intelligently fusing multiple information, the proposed ANN model can provide an accurate estimation of the modulus of elasticity, which shows a significant improvement from empirical based models used in current practice. The results also indicate that expansion due to ASR is not the only factor contributing to the stiffness change, and various factors have to be included during the evaluation.

• Structural Engineering and Mechanics
• /
• v.89 no.3
• /
• pp.283-300
• /
• 2024

Machine learning (ML) models based on artificial neural network (ANN) and decision tree (DT) were developed for estimation of axial capacity of concrete columns reinforced with fiber reinforced polymer (FRP) bars. Between the design codes, the Canadian code provides better formulation compared to the Australian or American code. For empirical models based on elastic modulus of FRP, Hadhood et al. (2017) model performed best. Whereas for empirical models based on tensile strength of FRP, as well as all empirical models, Raza et al. (2021) was adjudged superior. However, compared to the empirical models, all ML models exhibited superior performance according to all five performance metrics considered. The performance of ANN and DT models were comparable in general. Under the present setup, inclusion of the transverse reinforcement information did not improve the accuracy of estimation with either ANN or DT. With selective use of inputs, and a much simpler ANN architecture (4-3-1) compared to that reported in literature (Raza et al. 2020: 6-11-11-1), marginal improvement in correlation could be achieved. The metrics for the best model from the study was a correlation of 0.94, absolute errors between 420 kN to 530 kN, and the range being 0.39 to 0.51 for relative errors. Though much superior performance could be obtained using ANN/DT models over empirical models, further work towards improving accuracy of the estimation is indicated before design of FRP reinforced concrete columns using ML may be considered for design codes.

• Steel and Composite Structures
• /
• v.42 no.6
• /
• pp.733-745
• /
• 2022

This study proposed a robust artificial intelligence (AI) model based on the social behaviour of the imperialist competitive algorithm (ICA) and artificial neural network (ANN) for modelling the deflection of reinforced concrete beams, abbreviated as ICA-ANN model. Accordingly, the ICA was used to adjust and optimize the parameters of an ANN model (i.e., weights and biases) aiming to improve the accuracy of the ANN model in modelling the deflection reinforced concrete beams. A total of 120 experimental datasets of reinforced concrete beams were employed for this aim. Therein, applied load, tensile reinforcement strength and the reinforcement percentage were used to simulate the deflection of reinforced concrete beams. Besides, five other AI models, such as ANN, SVM (support vector machine), GLMNET (lasso and elastic-net regularized generalized linear models), CART (classification and regression tree) and KNN (k-nearest neighbours), were also used for the comprehensive assessment of the proposed model (i.e., ICA-ANN). The comparison of the derived results with the experimental findings demonstrates that among the developed models the ICA-ANN model is that can approximate the reinforced concrete beams deflection in a more reliable and robust manner.

• Advances in nano research
• /
• v.15 no.2
• /
• pp.165-174
• /
• 2023

The study investigated the effect of geometric structures of nano-patterned surfaces, such as peak sharpness, height, width, aspect ratio, and spacing, on mechano-bactericidal properties. Here, in silico models were developed to explain surface interactions with Escherichia coli. Numerical solutions were performed based on the finite element method and verified by the artificial neural network method. An E. coli cell adhered to the nano surface formed elastic and creep deformation models, and the cells' maximum deformation, maximum stress, and maximum strain were calculated. The results determined that the increase in peak sharpness, aspect ratio, and spacing values increased the maximum deformation, maximum stress, and maximum strain on E. coli cell. In addition, the results showed that FEM and ANN methods were in good agreement with each other. This study proved that the geometrical structures of nano-patterned surfaces have an important role in the mechano-bactericidal effect.

• Computers and Concrete
• /
• v.33 no.6
• /
• pp.739-754
• /
• 2024

This study presents an innovative AI-driven approach to assess the ultimate axial load in Double-Skinned Profiled Steel sheet Composite Walls (DPSCWs). Utilizing a dataset of 80 entries, seven input parameters were employed, and various AI techniques, including Linear Regression, Polynomial Regression, Support Vector Regression, Decision Tree Regression, Decision Tree with AdaBoost Regression, Random Forest Regression, Gradient Boost Regression Tree, Elastic Net Regression, Ridge Regression, and LASSO Regression, were evaluated. Decision Tree Regression and Random Forest Regression emerged as the most accurate models. The top three performing models were integrated into a hybrid approach, excelling in accurately estimating DPSCWs' ultimate axial load. This adaptable hybrid model outperforms traditional methods, reducing errors in complex scenarios. The validated Artificial Neural Network (ANN) model showcases less than 1% error, enhancing reliability. Correlation analysis highlights robust predictions, emphasizing the importance of steel sheet thickness. The study contributes insights for predicting DPSCW strength in civil engineering, suggesting optimization and database expansion. The research advances precise load capacity estimation, empowering engineers to enhance construction safety and explore further machine learning applications in structural engineering.

• Korea-Australia Rheology Journal
• /
• v.14 no.4
• /
• pp.161-174
• /
• 2002

A new technique for numerical calculation of viscoelastic flow based on the combination of Neural Net-works (NN) and Brownian Dynamics simulation or Stochastic Simulation Technique (SST) is presented in this paper. This method uses a "universal approximator" based on neural network methodology in combination with the kinetic theory of polymeric liquid in which the stress is computed from the molecular configuration rather than from closed form constitutive equations. Thus the new method obviates not only the need for a rheological constitutive equation to describe the fluid (as in the original Calculation Of Non-Newtonian Flows: Finite Elements St Stochastic Simulation Techniques (CONNFFESSIT) idea) but also any kind of finite element-type discretisation of the domain and its boundary for numerical solution of the governing PDE's. As an illustration of the method, the time development of the planar Couette flow is studied for two molecular kinetic models with finite extensibility, namely the Finitely Extensible Nonlinear Elastic (FENE) and FENE-Peterlin (FENE-P) models.P) models.

• ETRI Journal
• /
• v.36 no.6
• /
• pp.960-967
• /
• 2014

In this paper, an analytical framework is proposed for the optimization of network performance through joint congestion control, channel allocation, rate allocation, power control, scheduling, and routing with the consideration of fairness in multi-channel wireless multihop networks. More specifically, the framework models the network by a generalized network utility maximization (NUM) problem under an elastic link data rate and power constraints. Using the dual decomposition technique, the NUM problem is decomposed into four subproblems - flow control; next-hop routing; rate allocation and scheduling; power control; and channel allocation - and finally solved by a low-complexity distributed method. Simulation results show that the proposed distributed algorithm significantly improves the network throughput and energy efficiency compared with previous algorithms.