Prediction of Gas-Liquid Pressurization Performances of Multistage Multiphase Pumps Based on Similarity Laws and Neural Networks北大核心CSCD

It is very important to accurately predict the gas-liquid pressurization performance of multiphase pumps for the economy and safety of oil-gas production. Existent prediction models and methods are limited by narrow parameter ranges and low pump stages. A gas-liquid experimental platform at the industrial level was built, and the gas-liquid pressurization performances of a 25-stage centrifugal multiphase pump were obtained. A prediction method for gas-liquid pressurization performances was proposed for multiphase pumps with high stages at variable rotational speeds. Firstly, the artificial neural network of gas-liquid boosting pressure in the pump with low stages at a constant rotational speed, was constructed. Then, the boosting pressures at variable rotational speeds were converted to the designed condition by the 2-phase similarity law. Finally, based on the isothermal compression hypothesis, the inter-stage flow parameters were updated and the boosting pressures in pumps with high stages were acquired. The relative errors of prediction results of gas-liquid pressurization were less than 15% in pumps with different stage numbers (3~25 stages) and rotational speeds (2 500~3 500 r·min-1). The proposed method can be applied to other types of multiphase pumps, to determine the stage numbers of multiphase pumps and make production evaluation in oil-gas industry. © 2023 Editorial Office of Applied Mathematics and Mechanics. All rights reserved.     

Size-Dependent Effects of Micro-Nano Mindlin Plates Based on the Couple Stress Theory北大核心CSCD

基于偶应力理论,建立了适用于微纳米结构的Mindlin板理论。考虑横向剪切变形和材料的尺度效应并引入长度尺寸参数,推导了各向同性微纳米Mindlin板的本构方程。根据板的平衡条件,进一步推导出用位移函数和转角函数表示的板的屈曲和振动控制方程。通过对位移和转角变量进行空间和时间域上的分离,得出了四边简支（SSSS）和对边简支、对边固支（SCSC）两种边界情况下微纳米板的屈曲和振动问题的解析解。然后利用MATLAB软件进行算例分析,获得了不同尺寸参数、长宽比、厚长比等情况下板的临界屈曲荷载和固有频率。研究结果与已有文献中的结果以及ABAQUS有限元仿真解进行对比,结果表明,不同参数下的三种方法得到的结果均十分接近。算例分析发现,尺度效应对屈曲载荷和固有频率都有显著影响。     

Intelligent Crack Recognition Based on XFEM and GA-BP Neural Networks北大核心CSCD

基于扩展有限元法(XFEM)和经遗传算法(GA)优化的误差反向传播多层前馈(BP)神经网络(GA-BP)算法,建立了识别结构中裂纹的反演分析模型。模型通过XFEM正向分析获得的测点位移数据训练GA-BP神经网络,并在此基础上利用该网络进行裂纹反向识别。通过两个典型算例对模型的可行性和精度进行了验证,并探讨了网格密度、测点布置、输入数据噪声等对网络识别精度的影响。结果表明,该文的方法可反演线弹性断裂力学重点关注的直线裂纹的几何信息且具有较好的容噪性能,此外,GA-BP神经网络的预测精度较传统BP神经网络普遍更高。     

RBF Neural Network Based Prediction on Blade Surface Pressure Fields in Compressors北大核心CSCD

The airflow characteristics of the internal flow path of an aero-engine compressor are complex, and the vortex flow field around the blade is characterized by high pressure, high speed, rotation, and unsteadiness. Therefore, there is an urgent need to calculate and predict the aerodynamic characteristics of the complex flow field around the compressor blade efficiently and accurately. The computational fluid dynamics (CFD) method was used to generate the aerodynamic load distribution on the blade surface under different operating conditions for the study of the complex flow fields around aero-engine blades. The radial based function (RBF) neural network was applied to establish the pressure surface aerodynamic load prediction model, and the neural network modeling method was combined with the flow field calculation. The neural network method can learn and train the CFD-based data set to properly compensate the errors from the CFD, which provides a reference for the effective prediction of the complex flow fields around aero-engine compressor blades. © 2023 Editorial Office of Applied Mathematics and Mechanics. All rights reserved.     

Deformation Behavior Modeling of SMAs Under Cyclic Loading Based on Rational Interpolation北大核心CSCD

提出了一个有限弹塑性模型,用来模拟形状记忆合金(shape memory alloys,SMAs)在循环荷载下的变形行为.首先,通过分析上下屈服阶段形函数的特点,利用有理插值方法给出循环荷载下的应力-应变形函数显式表达,可以精确匹配任意形状的实验数据;其次,基于对数客观率,构建了有限弹塑性J2流模型,耦合了屈服中心的移动和屈服面的增大;再次,从单轴情况出发,推导得到了单个循环下的三个硬化函数显式表达,再引入局部因子和多轴扩展不变量,构造了光滑统一且多轴有效的硬化函数;最后,将模型得到的结果与经典实验结果比较,证明了新方法的有效性.该文创新点如下:第一,通过改进传统的背应力演化方程,使得新模型产生强烈的Bauschinger效应,从而使新方法具备模拟SMAs特殊变形行为的能力;第二,新的光滑统一硬化函数在单个循环下会自动退化,得到精确符合实验数据的结果;第三,利用本构方程推导得到有效塑性功演化规律,而有理插值得到的形函数中包含依赖有效塑性功的参数,给出这些参数方程以后使得模型具备了预测变形的能力.     

Numerical Simulation of Transient Non-Isothermal Viscoelastic Couette Flows Based on the SPH Method北大核心CSCD

Based on the smoothed particle hydrodynamics (SPH) method, the transient non-isothermal viscoelastic flows were numerically simulated. First, the viscoelastic Couette flow based on the Oldroyd-B model under isothermal condition was simulated. Then, the simulation was extended to the non-isothermal case, in which the Reynolds exponential model was adopted to evaluate the dependence of the viscosity and the relaxation time on the temperature. The accuracy and effectiveness of the SPH method for simulating transient non-isothermal viscoelastic flows were verified through comparison with the finite volume method and evaluation of numerical convergence. The different flow characteristics of the non-isothermal flow compared with those of the isothermal flow were discussed. The effects of the temperature dependence coefficient and the Péclet number on the flow physics were analyzed. The numerical results demonstrate that, the SPH method can accurately and effectively simulate transient non-isothermal viscoelastic flow problems. © 2023 Editorial Office of Applied Mathematics and Mechanics. All rights reserved.     

Switching Control of Nonlinear Systems Based on the Quasi-ARX Model and the SVR Algorithm北大核心CSCD

该文基于改进的含有外部输入项的准线性自回归(准ARX)径向基函数(RBF)网络模型和支持向量回归(SVR)算法,提出了一种非线性切换控制方法.改进的准ARX模型非线性部分采用RBF网络.控制系统设计过程分为三个部分:首先,利用聚类方法确定模型的非线性参数;然后,采用线性SVR算法来解决控制系统的鲁棒性问题;接下来,基于控制误差给出切换判定函数,确定切换律给出控制序列.最后通过数值仿真验证了该方法的有效性.     

Identification of Pipeline Inner Wall Geometry Based on the POD⁃RBF Method北大核心CSCD

Based on the proper orthogonal decomposition⁃radial basis function (POD⁃RBF), a geometric identification method for pipeline inner wall was proposed to solve the internal corrosion detection problem of natural gas and oil pipelines. In view of the static magnetic field, the simplified finite element model for the pipelines was established, and the variable⁃geometry sample library was constructed, to realize the response prediction of arbitrary geometry by the POD⁃RBF. The proposed method achieves reduced⁃order analysis and avoids repeated solution of the stiffness matrix due to the geometrical change during the identification process. Hence, it can significantly improve the computation efficiency. Finally, the grey wolf optimization (GWO) algorithm was used to optimize the objective function and avoid the calculation of the sensitivity in the process of geometry change. The numerical examples show that, the proposed method has high efficiency and accuracy in the geometric identification of the pipeline inner wall, with good stability even under introduced noises. © 2023 Editorial Office of Applied Mathematics and Mechanics. All rights reserved.     

Experimental Study on the Flexural Ductility of BFRP Bar Concrete Beams With Bamboo Fiber and Steel Wire Mesh北大核心CSCD

To study the effects of bamboo fiber and steel wire mesh on the flexural ductility of basalt fiber reinforced polymer（BFRP）bar concrete beams, 7 BFRP bar concrete beams with bamboo fiber and steel wire mesh were tested with different bamboo fiber lengths (0 mm, 30 mm and 45 mm) and different steel wire mesh layout ranges (0, 1/2 maximum bending moment point layout and full beam length layout). The flexural failure tests of the 7 beams were carried out, and the initial crack loads, the crack developments, the ultimate loads and the deformations were detected. The effects of the fiber length and the wire mesh layout range on the crack resistance and the deformation resistance of the specimens were analyzed based on the test data. With the function model, the equivalent yield points of the 7 test beams were obtained, and their ductility coefficients were calculated. The results show that, the addition of bamboo fiber and steel wire mesh increases the cracking loads of BFRP bar concrete beams by 12%~68%, decreases the crack spacings and the crack length development speed, reduces the test beam deformation under the same load, and increases the ductility coefficient by 1.58%~31.75%. © 2023 Editorial Office of Applied Mathematics and Mechanics. All rights reserved.     

Nonlinear Frequency Analysis of FGM Pipes Based on the Homotopy Method北大核心CSCD

Based on the homotopy analysis method, the nonlinear vibration of porous functionally graded material (FGM) conveying pipes under generalized boundary conditions was studied. Based on the power-law distribution of the FGM and the Voigt model, the physical properties of the porous pipe material were described. Under the Euler-Bernoulli beam theory and the von Kármán nonlinear theory, and by means of Hamilton’s variational principle, the dynamic control equations and generalized boundary conditions for porous FGM conveying pipes were established. The homotopy analysis method was used to solve the nonlinear vibration characteristics of the porous FGM conveying pipe under generalized boundary conditions. The numerical results show that, the translation spring has little effect on the critical velocity of instability, while the rotation spring increases the critical velocity of instability, making the system more stable; in the nonlinear system, the viscoelastic coefficient does not change the critical velocity; the pipe length, the power-law exponent and the porosity all influence the nonlinear free vibration of the porous FGM conveying pipe. © 2023 Editorial Office of Applied Mathematics and Mechanics. All rights reserved.     

基于BP神经网络的道路交通事故预测

道路交通事故预测是交通研究的一个重要课题.以我国交通安全状况为研究对象,依据我国道路交通事故的特点,利用神经网络具有自学习、自组织、自适应能力特性,运用神经网络的方法及我国多个年度道路交通事故统计数据,建立了道路交通事故神经网络宏观预测模型,预测精度符合道路交通事故预测的要求.     

基于RBM的神经网络时间序列预测

为了克服神经网络依赖初始化结果,泛化能力不强的缺点,提出一种基于受限玻尔兹曼机(RBM)的神经网络模型.利用无监督学习方法优化神经网络的初始权值和阈值,将RBM与神经网络融合起来,模型与时间序列神经网络做实验对比,结果表明,基于受限的玻尔兹曼机的神经网络模型优于神经网络预测模型,模型可以提高预测的精准度,具有一定的应用意义.     

基于神经网络的股票预测模型

针对股票市场的特征提取困难、预测精度较低等问题,本文基于深度学习算法,构建了一系列用于股票市场预测的神经网络模型,包括基于多层感知机(MLP)、卷积神经网络(CNN)、递归神经网络(RNN)、长短期记忆网络(LSTM)和门控神经单元(GRU)的模型。 针对RNN、LSTM和GRU无法充分利用所参考的时间维度的信息,引入注意力机制(Attention Mechanism) 给各时间维度的信息赋予不同权重,区分不同信息对预测的重要程度,从而提升递归网络模型的性能。上述模型均基于股票数据进行了优化,基于上证指数对各类模型进行了充分的对比实验,探索了模型中重要变量对性能的影响,旨在为基于神经网络的股票预测模型给出具体的优化方向。     

基于CMAC神经网络的舰船运动预报仿真研究

舰船运动是引发多种海面事故的主要原因,为了提高舰载机着舰安全系数、提高舰船战斗力、减少海难等事故的发生,舰船运动极短期预报是至关重要的.首次将CMAC神经网络应用于舰船纵摇运动姿态的极短期预报中,建立了CMAC神经网络舰船运动预报模型.同时对实际的舰船纵摇运动数据进行了数值仿真,证实了CMAC神经网络模型在舰船运动预报中的应用是合理且可行的.     

基于ARIMA模型与神经网络模型的股价预测

了解并掌握股价运行的规律是许多投资者和学者所关注的领域,采用了ARIMA模型和BP神经网络对百度、阿里巴巴两支股票的收盘价进行建模与预测,并对比了两模型的预测精度,结果表明两种预测模型都达到比较理想的预测精度和短期预测可行的效果.     

基于RN和GN的两种RBF神经网络的工程造价预测模型

建立了调用NEWRB函数的正规化网络RN和基于K-means聚类的广义网络GN的两种RBF‘神经网络的工程造价预测模型,以55个厦门市工程造价案例进行实证分析.结果表明:当调用NEWRB函数构建RBF模型时,其性能主要取决于分布宽度,而基于K-means聚类的RBF神经网络主要取决于重叠系数和隐含层节点数;基于广义网络GN的RBF神经网络模型的训练效果较差,但学习速度更快、预测精度更高.     

基于卷积神经网络模型数值求解双曲型偏微分方程的研究

人工神经网络近年来得到了快速发展,将此方法应用于数值求解偏微分方程是学者们关注的热点问题.相比于传统方法其具有应用范围广泛(即同一种模型可用于求解多种类型方程)、网格剖分条件要求低等优势,并且能够利用训练好的模型直接计算区域中任意点的数值.该文基于卷积神经网络模型,对传统有限体积法格式中的权重系数进行优化,以得到在粗粒...     

基于深度学习LSTM对交通流状态的预测

当今道路交通状态对城市管理和人们出行愈加重要,影响着人类生活的方方面面.以深圳交通为研究对象,由基础车辆数据和道路坐标构建了路网系统,从车辆速度和密度两个方面导出了交通流状态评价指数TSI.利用深度学习长短期记忆神经网络(LSTM)对车辆速度和密度两个指标进行预测,并通过对比极限学习机(ELM),时间序列(ARMA)和BP神经网络,进行仿真实验,结果表明相对于传统预测模型,所采用的LSTM网络具有更优的预测精确度和对远期预测的稳定性.最后利用预测结果计算出更能直观反映出道路交通拥堵情况的TSI指数,为人们提供了准确的交通状态预测.     

COVID-19 Epidemic Prediction Based on Deep Learning

In this paper, a multi-layer gated recurrent unit neural network (multi-head GRU) model is proposed to predict the confirmed cases of the new crown epidemic (COVID-19). We extract the time series relationship in the data, and the rolling prediction method is adopted to ensure the simple structure of the model and achieve higher precision and interpretability. The prediction results of this model are compared with the LSTM model, the Transformer model and the infectious disease model (SIR). The results show that the proposed model has higher prediction accuracy. The mean absolute error (MAE) of epidemic prediction in most countries (the United States, Brazil, India, the United Kingdom and Russia) is respectively 197.52, 68.02, 200.67, 24.78 and 123.50, which is much smaller than the prediction error of the SIR model, LSTM model and Transformer model. For the spread of the COVID-19 epidemic, traditional infectious disease models and machine learning models cannot achieve more accurate predictions. In this paper, we use a GRU model to predict the real-time spread of COVID-19, which has fewer parameters and reduces the risk of overfitting to train faster. Meanwhile, it can make up for the shortcoming of the transformer model to capture local features.     

数据挖掘在房地产价格预测中的应用研究

通过基于数据挖掘理论的粗糙集和神经网络的研究,用属性约简算法约简并提取了影响房地产价格的主要指标因素,对降维后的数据进行网络学习和训练,最后用训练好的的网络检验测试样本.方法使学习训练的速度和识别率提高了,为房地产价格预测提供了一种更为有效和实用的新途径.