Combination and application of machine learning and computational mechanics

Li, Xiang; Yan, Zeng; Liu, Zhanli

doi:10.1360/n972019-00005

Cited by 14 publications

(8 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, reference 14 used real material experiment data directly instead of the material constitutive model. 15 A data-driven model of dynamics of rail vehicle was formulated to replace the original mechanical element by machine learning. This method assumed that a set of sample data corresponds to a set of observed outputs, thus predicting the underlying relationship between the inputs and the outputs.…”

Section: Introductionmentioning

confidence: 99%

A numerical method combining finite element and neural network model to study dynamic system prediction of dry friction

Lyu

Zhang

Lyu³

et al. 2022

Proceedings of the Institution of Mechanical Engineers, Part J:

View full text Add to dashboard Cite

Friction wedge is an important damping component of three-piece freight bogies, and the better damping performance is beneficial to improving the stability of the vehicle operation. This paper introduces an effective method for numerical simulation of dry friction system and the related experiment was conducted to verify the correctness of the method. On the basis of conducting the experimental of dry friction model to test the lateral and tangential forces of the dry friction, the dynamic friction coefficient under different speeds and pressures was calculated. The most suitable dry friction model was obtained by comparing the fitting accuracy of different models. The fitting accuracy of the neural network model is above 0.9, which is much higher than other models. Pressure is an important parameter of the friction coefficient and should be taken into account in the model. The dynamic implicit procedure was adopted in the simulation process with Abaqus/Standard solver, the user-subroutine FRIC integrated in the commercial package ABAQUS was coded to study the rate and pressure dependent dynamic friction during the movement of dry friction system. The calculation result is basically consistent with the experiment when the neural network model is combined with the user-subroutine FRIC. The feasibility of the co-simulation analysis method is verified. The neural network model is more accurate and convenient to establish the dynamic friction model, avoiding the difficulty of choosing the the dry friction model. It is verified that the neural network model can be used in finite element analysis, which provides a new idea for the combination of neural network and traditional calculation methods.

show abstract

Section: Introductionmentioning

confidence: 99%

A numerical method combining finite element and neural network model to study dynamic system prediction of dry friction

Lyu

Zhang

Lyu³

et al. 2022

Proceedings of the Institution of Mechanical Engineers, Part J:

View full text Add to dashboard Cite

show abstract

“…Many scholars apply machine learning methods to their research fields, such as signal processing, biomedicine, complex dynamic systems (Brunton et al, 2016), multiphysical phenomena (Rudy et al, 2017), etc. Some commonly used machine learning methods include Perceptron (Rosenblatt, 1958), Genetic Programming (Goldberg, 1989;Banzhaf et al, 1998), Kernel and Nearest-Neighbor Nonparametric Regression (Dudani, 1976;Altman, 1992), Linear Statistical Models (Neter et al, 1996), Adaptive Boosting Algorithm (Freund and Schapire, 1997;Hastie et al, 2009), Support-Vector Machines (Cortes and Vapnik, 1995;Tefas et al, 2002;Veropoulos et al, 2016) and Artificial Neural Network (Ivakhnenko, 1971;Rumelhart et al, 1986;Widrow, 1987;Ge et al, 2004;Hinton and Salakhutdinov, 2006;Li et al, 2019). Rosenblatt (Rosenblatt, 1958) built the perceptron model and described the process of learning behavior in detail which is considered as the precursor to modern artificial network models (Cortes and Vapnik, 1995).…”

Section: Introductionmentioning

confidence: 99%

“…This method is very advantageous while dealing with nonseparable training data (Cortes and Vapnik, 1995) and has been successfully applied to a number of applications, ranging from face recognition (Tefas et al, 2002) to biological data processing for medical diagnosis (Veropoulos et al, 2016). In addition to the various methods mentioned above, Artificial Neural Networks have gradually developed into an important branch in the field of machine learning (Li et al, 2019). From "Perceptron" (Rosenblatt, 1958) to "MADLINE" (Widrow, 1987) to "restricted Boltzmann machine" (Hinton and Salakhutdinov, 2006), with the use of nonlinear functions (Ivakhnenko, 1971) and back-propagation algorithms (Rumelhart et al, 1986), ANN could train a multilayer neural network with a central layer to convert high-dimensional data to low-dimensional codes by extensive tunable parameters through multi-layer network structure which will obtain the nonlinear relationship contained in a large amount of training data and generally receive good classification and regression accuracy.…”

Section: Introductionmentioning

confidence: 99%

Machine learning for rock mechanics problems; an insight

Taleghani

Balushi

et al. 2022

Front. Mech. Eng.

View full text Add to dashboard Cite

Due to inherent heterogeneity of geomaterials, rock mechanics involved with extensive lab experiments and empirical correlations that often lack enough accuracy needed for many engineering problems. Machine learning has several characters that makes it an attractive choice to reduce number of required experiments or develop more effective correlations. The timeliness of this effort is supported by several recent technological advances. Machine learning, data analytics, and data management have expanded rapidly in many commercial sectors, providing an array of resources that can be leveraged for subsurface applications. In the last 15 years, deep learning in the form of deep neural networks, has been used very effectively in diverse applications, such as computer vision, seismic inversion, and natural language processing. Despite the remarkable success in these and related areas, deep learning has not yet been widely used in the field of scientific computing specially when it comes to subsurface applications due to the lack of large amount of data to train algorithms. In this paper, we review such efforts and try to envision future game-changing advances that may impact this field.

show abstract

“…Over the last century, there has been steady advancement in the power of computers; simultaneously, data science and artificial intelligence have undergone rapid developments [1,2]. The effective combination of machine learning and artificial neural network (ANN) has been the focus of ship motion prediction research.…”

Section: Introductionmentioning

confidence: 99%

Real-Time Prediction of Large-Scale Ship Model Vertical Acceleration Based on Recurrent Neural Network

Lin

Zhao

et al. 2020

JMSE

View full text Add to dashboard Cite

In marine environments, ships are bound to be disturbed by several external factors, which can cause stochastic fluctuations and strong nonlinearity in the ship motion. Predicting ship motion is pivotal to ensuring ship safety and providing early warning of risks. This report proposes a real-time ship vertical acceleration prediction algorithm based on the long short-term memory (LSTM) and gated recurrent units (GRU) models of a recurrent neural network. The vertical acceleration time history data at the bow, middle, and stern of a large-scale ship model were obtained by performing a self-propulsion test at sea, and the original data were pre-processed by resampling and normalisation via Python. The prediction results revealed that the proposed algorithm could accurately predict the acceleration time history data of the large-scale ship model, and the root mean square error between the predicted and real values was no greater than 0.1. The optimised multivariate time series prediction program could reduce the calculation time by approximately 55% compared to that of a univariate time series prediction program, and the run time of the GRU model was better than that of the LSTM model.

show abstract

Combination and application of machine learning and computational mechanics

Cited by 14 publications

References 36 publications

A numerical method combining finite element and neural network model to study dynamic system prediction of dry friction

A numerical method combining finite element and neural network model to study dynamic system prediction of dry friction

Machine learning for rock mechanics problems; an insight

Real-Time Prediction of Large-Scale Ship Model Vertical Acceleration Based on Recurrent Neural Network

Contact Info

Product

Resources

About