Mishra Bharat Bhushan, Kumar Ajay, Zaburko Jacek, Sadowska-Buraczewska Barbara, Barnat-Hunek Danuta
Department of Civil Engineering, National Institute of Technology Patna, Patna 800005, India.
Faculty of Environmental Engineering, Lublin University of Technology, Nadbystrzycka St. 40B, 20-618 Lublin, Poland.
Materials (Basel). 2021 Jan 14;14(2):395. doi: 10.3390/ma14020395.
In the present work, for the first time, free vibration response of angle ply laminates with uncertainties is attempted using Multivariate Adaptive Regression Spline (MARS), Artificial Neural Network-Particle Swarm Optimization (ANN-PSO), Gaussian Process Regression (GPR), and Adaptive Network Fuzzy Inference System (ANFIS). The present approach employed 2D stochastic finite element (FE) model based on the Third Order Shear Deformation Theory (TSDT) in conjunction with MARS, ANN-PSO, GPR, and ANFIS. The TSDT model used eliminates the requirement of shear correction factor owing to the consideration of the actual parabolic distribution of transverse shear stress. Zero transverse shear stress at the top and bottom of the plate is enforced to compute higher-order unknowns. FE model makes it commercially viable. Stochastic FE analysis done with Monte Carlo Simulation (MCS) FORTRAN inhouse code, selection of design points using a random variable framework, and soft computing with MARS, ANN-PSO, GPR, and ANFIS is implemented using MATLAB in-house code. Following the random variable frame, design points were selected from the input data generated through Monte Carlo Simulation. A total of four-mode shapes are analyzed in the present study. The comparison study was done to compare present work with results in the literature and they were found in good agreement. The stochastic parameters are Young's elastic modulus, shear modulus, and the Poisson ratio. Lognormal distribution of properties is assumed in the present work. The current soft computation models shrink the number of trials and were found computationally efficient as the MCS-based FE modelling. The paper presents a comparison of MARS, ANN-PSO, GPR, and ANFIS algorithm performance with the stochastic FE model based on TSDT.
在本研究中，首次尝试使用多元自适应回归样条法（MARS）、人工神经网络 - 粒子群优化算法（ANN - PSO）、高斯过程回归法（GPR）和自适应网络模糊推理系统（ANFIS）来研究具有不确定性的角铺设层合板的自由振动响应。本方法采用基于三阶剪切变形理论（TSDT）的二维随机有限元（FE）模型，并结合MARS、ANN - PSO、GPR和ANFIS。所使用的TSDT模型由于考虑了横向剪应力的实际抛物线分布，消除了剪切修正因子的需求。通过强制板顶部和底部的横向剪应力为零来计算高阶未知量。有限元模型使其具有商业可行性。使用蒙特卡罗模拟（MCS）FORTRAN内部代码进行随机有限元分析，使用随机变量框架选择设计点，并使用MATLAB内部代码实现MARS、ANN - PSO、GPR和ANFIS的软计算。按照随机变量框架，从通过蒙特卡罗模拟生成的输入数据中选择设计点。本研究共分析了四种振型。进行了对比研究，将本研究结果与文献中的结果进行比较，发现两者吻合良好。随机参数为杨氏弹性模量、剪切模量和泊松比。本研究假设材料属性服从对数正态分布。当前的软计算模型减少了试验次数，并且与基于MCS的有限元建模相比，计算效率更高。本文给出了基于TSDT的随机有限元模型与MARS、ANN - PSO、GPR和ANFIS算法性能的比较。
