STUDY ON PROPERTY PREDICTION FOR SEALING ALLOYS

This paper describes a model of property prediction for alloys using the mapping function and self-learning ability of artificial neural network. By learning from experimental data, the neural network induces the relationship between composition, processing and properties of alloys, and predicts the properties with given composition and processing parameters of new alloys.The verification of sealing alloys demonstrates that the artificial neural network is an effective method for materials design and properties prediction.

Prediction of mechanical property of E4303 electrode using artificial neural network

Based on the method of artificial neural network, a new approach has been devised to predict the mechanical property of E4303 electrode. The outlined predication model for determining the mechanical property of electrode was built upon the production data. The research leverages a back propagation algorithm as the neural network’s learning rule. The result indicates that there are positive correlations between the predicted results and the practical production data. Hence, using the neural network, predication of electrode property can be realized. For the first time, this research provides a more scientific method for designing electrode.

Prediction of 2A70 aluminum alloy flow stress based on BP artificial neural network

The hot deformation behavior of 2A70 aluminum alloy was investigated by means of isothermal compression tests performed on a Gleeble-1500 thermal simulator over 360～480 ℃ with strain rates in the range of 0.01～1 s-1 and the largest deformation up to 60%. On the basis of experiments, a BP artificial neural network （ANN） model was constructed to predict 2A70 aluminum alloy flow stress. True strain, strain rates and temperatures were input to the network, and flow stress was the only output. The comparison between predicted values and experimental data showed that the relative error for the trained model was less than ±3% for the sampled data while it was less than ±6% for the non-sampled data. Furthermore, the neural network model gives better results than nonlinear regression method. It is evident that the model constructed by BP ANN can be used to accurately predict the 2A70 alloy flow stress.

Evolutionary artificial neural network approach for predicting properties of Cu-15Ni-8Sn-0.4Si alloy

A novel data mining approach,based on artificial neural network(ANN) using differential evolution(DE) training algorithm,was proposed to model the non-linear relationship between parameters of aging processes and mechanical and electrical properties of Cu-15Ni-8Sn-0.4Si alloy.In order to improve predictive accuracy of ANN model,the leave-one-out-cross-validation (LOOCV) technique was adopted to automatically determine the optimal number of neurons of the hidden layer.The forecasting performance of the proposed global optimization algorithm was compared with that of local optimization algorithm.The present calculated results are consistent with the experimental values,which suggests that the proposed evolutionary artificial neural network algorithm is feasible and efficient.Moreover,the experimental results illustrate that the DE training algorithm combined with gradient-based training algorithm achieves better convergence performance and the lowest forecasting errors and is therefore considered to be a promising alternative method to forecast the hardness and electrical conductivity of Cu-15Ni-8Sn-0.4Si alloy.

Artificial Neural Network and Fuzzy Logic Based Techniques for Numerical Modeling and Prediction of Aluminum-5%Magnesium Alloy Doped with REM Neodymium

In this study, the mechanical properties of aluminum-5%magnesium doped with rare earth metal neodymium were evaluated. Fuzzy logic (FL) and artificial neural network (ANN) were used to model the mechanical properties of aluminum-5%magnesium (0-0.9 wt%) neodymium. The single input (SI) to the fuzzy logic and artificial neural network models was the percentage weight of neodymium, while the multiple outputs (MO) were average grain size, ultimate tensile strength, yield strength elongation and hardness. The fuzzy logic-based model showed more accurate prediction than the artificial neutral network-based model in terms of the correlation coefficient values (R).

Neural Network Modeling and Prediction of Surface Roughness in Machining Aluminum Alloys

Artificial neural network is a powerful technique of computational intelligence and has been applied in a variety of fields such as engineering and computer science. This paper deals with the neural network modeling and prediction of surface roughness in machining aluminum alloys using data collected from both force and vibration sensors. Two neural network models, including a Multi-Layer Perceptron (MLP) model and a Radial Basis Function (RBF) model, were developed in the present study. Each model includes eight inputs and five outputs. The eight inputs include the cutting speed, the ratio of the feed rate to the tool-edge radius, cutting forces in three directions, and cutting vibrations in three directions. The five outputs are five surface roughness parameters. Described in detail is how training and test data were generated from real-world machining experiments that covered a wide range of cutting conditions. The results show that the MLP model provides significantly higher accuracy of prediction for surface roughness than does the RBF model.

Classification and Prediction on Rural Property Mortgage Data with Three Data Mining Methods

The Farmers Property Mortgage Policy is a strategic financial policy in western China, a relatively underdeveloped region. Many contradictions and conflicts exist in the process between the strong demand for the loans by farmers and the strict risk control by the financial institutions. The rural finance corporations should use scientific analysis and investigation of the potential households for overall evaluation of the customers. These include historical credit rating, present family situation, and other related information. Three different data mining methods were applied in this paper to the specifically-collected household data. The objective was to study which factor could be the most important in determining loan demand for households, and in the meanwhile, to classify and predict the possibility of loan demand for the potential customers. The results obtained from the three methods indicated the similar outputs, income level, land area, the way of loan, and the understanding of policy were four main factors which decided the probability of one specific farmer applying for a credit loan. The results also embodied the difference within the three methods for classifying and predicting the loan anticipation for the testing households. The artificial neural network model had the highest accuracy of 91.4 which is better than the other two methods.

人工神经元网络模型预测3D打印部件力学性能的研究

熔融沉积成型(FDM)是一种高效的增材制造技术。将响应面模型与人工神经元网络(ANN)模型相结合,研究了FDM工艺的喷嘴温度、层高和层积角度对尼龙12(PA12)丝材制造部件力学性能的影响。当喷嘴温度、层高和层积角度分别在220~260℃、0.2~0.4 mm、0°~90°之间变化时,部件拉伸强度和缺口冲击强度分别在35.69~60.89 MPa和5.48~19.83 kJ/m^(2)之间。喷嘴温度、层高、层积角度以及层积角度的二阶效应是影响部件拉伸强度的显著因素;喷嘴温度、层积角度以及层积角度的二阶效应是影响缺口冲击强度的显著因素。ANN模型预测拉伸强度和缺口冲击强度的最优结构分别是3-10-5-1和3-25-24-1,预测的拉伸强度和缺口冲击强度均方误差函数(MSE)最低分别为2.54×10^(-4)和2.07×10^(-4),回归系数均在0.97以上。与响应面的二次回归模型相比,ANN模型预测的拉伸强度和缺口冲击强度与实验值的标准偏差分别为0.46和0.32,远低于二次回归模型的2.43和1.58,更适合于优化非线性的FDM工艺。

Prediction of tensile strength of friction stir welded 6061 Al plates

The present paper investigates the prediction of tensile strength after friction stir welding(FSW)using artificial neural network(ANN)in the MATLAB program.The experimental results are used to develop the mathematical model.The combined influence of welding speed,rotation speed,and axial force on the tensile strength of 6061 Al plates is simulated.Results of the tensile test are used to train and test the ANN model.A multi-layer solution is developed using the ANN model to predict tensile strength.Back propagation(BP)method is initially trained using 80%of the experimental data,then,testing is performed with the rest of the data.Results indicate that predicted values are close to the corresponding measured values.

Modeling of mechanical properties of as-cast Mg-Li-Al alloys based on PSO-BP algorithm

Artificial neural networks have been widely used to predict the mechanical properties of alloys in material research.This study aims to investigate the implicit relationship between the compositions and mechanical properties of as-cast Mg-Li-Al alloys.Based on the experimental collection of the tensile strength and the elongation of representative Mg-Li-Al alloys,a momentum back-propagation(BP)neural network with a single hidden layer was established.Particle swarm optimization(PSO)was applied to optimize the BP model.In the neural network,the input variables were the contents of Mg,Li and Al,and the output variables were the tensile strength and the elongation. The results show that the proposed PSO-BP model can describe the quantitative relationship between the Mg-Li-Al alloy's composition and its mechanical properties.It is possible that the mechanical properties to be predicted without experiment by inputting the alloy composition into the trained network model.The prediction of the influence of Al addition on the mechanical properties of as-cast Mg-Li-Al alloys is consistent with the related research results.

基于Attention-ResNet-LSTM混合神经网络的盾构掘进速度预测新方法

针对传统方法存在的盾构性能精准预测阻碍盾构快速掘进技术发展的难题,提出一种基于Attention-ResNet-LSTM混合神经网络的盾构掘进速度预测方法。相比传统的LSTM、GRU等网络预测模型,Attention-ResNet-LSTM模型引入了Attention机制。长距离盾构掘进过程中,针对地层条件存在很大的变异性情况,该模型可自适应更新权重矩阵,让模型面对不同的任务时具有一定的自调节能力,可有效提升预测精度。依托中俄东线天然气管道工程对盾构掘进速度进行了实时预测和验证,且结果表明该方法可分析盾构掘进过程中输入、输出参数之间的相关性,且具有较好的适应性。

基于BP神经网络的TC4钛合金超塑性变形后组织及性能预测研究

分别用VisualFortran语言和MATLAB软件建立了TC4钛合金超塑性变形时变形参数与其力学性能和晶粒尺寸之间的BP神经网络模型,通过用较少的力学性能和晶粒尺寸的试验数据进行训练,进而对其性能进行预测。结果表明,BP神经网络用于材料超塑性变形后的力学性能及晶粒尺寸预测是可行的,其预测误差小于7%。

基于参数优化的人工神经网络的AZ31镁合金力学性能预测模型

通过力学性能试验测定了不同退火条件下AZ31镁合金的抗拉强度、屈服强度和延伸率,并利用人工神经网络技术建立了对应力学性能的预测模型,其中对模型的优化采用了一种新方法,即参数全排列组合训练。结果表明,基于全排列训练得到的最优参数建立的网络模型具有优良的性能,比经传统试探法构建的模型具有更高的平均相关系数和更低的平均误差,因此能更准确地预测AZ31镁合金在不同退火条件后的力学性能。

人工神经网络在石油分析中的应用研究(Ⅰ)——BP神经网络预测石油馏分临界性质

基于面向对象程序设计的设计思想，定义３种新的神经元结构体类型变量，研究了４种变形ＢＰ神经网络模型，引入神经元非线性敏感度因子、动量因子和阈值的动态修正，在一定程度上克服了传统ＢＰ网络收敛速度慢、易陷于局部极小的缺陷，将４种模型应用于石油馏分临界性质的预测，取得较一般常规非线性处理方法更高的预测精度。

GH99合金TIG焊接接头拉伸性能的人工神经网络预测

利用Matlab7.0软件建立了用于预测GH99高温合金焊接接头拉伸性能的改进算法的多层BP神经网络.以焊接电流、焊接速度、脉冲频率、重熔次数、板厚、装配间隙、坡口与连接形式作为输入参数,抗拉强度、屈服强度和断后伸长率分别作为输出值.结果表明,改进算法的多层BP神经网络能够很好的预测GH99高温合金TIG焊接接头的拉伸性能,抗拉强度、屈服强度与断后伸长率预报值与试验值的平均相对误差分别为-0.76%,1.71%和2.30%.

应用人工神经网络模型预测Ti-10V-2Fe-3Al合金的力学性能

采用人工神经网络方法建立了Ti-10V-2Fe-3Al合金机械性能预测的神经网络模型。模型的输入参数包括变形温度、变形程度、固溶温度、时效温度等热加工工艺参数和热处理制度。模型的输出为钛合金最重要的5个机械性能指标,即抗拉强度、屈服强度、延伸率、断面收缩率和断裂韧性。与传统回归拟合公式相比,该模型具有容错性好、通用性强等优点。该模型可以预测Ti-10V-2Fe-3Al合金在不同热加工工艺参数和热处理制度下的机械性能,也可以用于优化热加工参数和热处理制度。

热轧带钢力学性能预测模型及其应用

为实现对热轧带钢的屈服强度、抗拉强度、断裂延伸率等力学性能的预测及控制,利用人工神经网络技术,分别建立了根据生产工艺参数预测力学性能的质量模型,以及根据力学性能要求对生产工艺参数进行优化的逆质量控制模型.利用质量预测模型,分析得出屈服强度随卷取温度的上升而下降的变化规律,进而可以对组织性能进行在线调整,实现在线应用.

基于人工神经网络的A357合金力学性能预测(英文)

A357铝合金零件一般都需要经过热处理(T6状态)以获得优异的力学性能。这类零件的性能取决于固溶温度、固溶时间、人工时效温度及人工时效时间。在本研究中,建立了基于反向传播(BP)算法的人工神经网络(ANN)模型,对A357合金的力学性能进行预测,研究了热处理工艺对该合金性能的影响。结果表明,所建立的BP模型能够对A357合金的力学性能进行有效且精度高的预测。良好的神经网络预测能力能够直观地反映A357合金的热处理工艺参数对其力学性能的影响。绘制抗拉强度和伸长率的等值线图形有助于清晰地找到抗拉强度和伸长率之间的关系,可为实际生产中热处理工艺参数的选择提供技术支持。

人工神经网络在玻璃配方设计中的应用研究

应用人工神经网络技术,采用 Neuralworks Predict 软件建立 BP 网络模型,通过对 R_2O-MO-Al_2O_3-SiO_2系统玻璃组成与热膨胀系数关系实验数据的训练,以期能预测该系统指定组成的玻璃的热膨胀系数。研究结果表明,所建立的神经网络模型能较正确地反映玻璃氧化物组成与其热膨胀系数之间的规律性。模型对给定组成玻璃热膨胀系数的预测值与实际测试值的相对误差在6.4%以内,表明由神经网络技术建立的这一模型能正确反映 R_2O-MO-Al_2O_3-SiO_2系统玻璃组成与热膨胀系数间的内在规律性。

人工神经网络在钢铁材料力学性能预测方面的应用

人工神经网络模型特别适用于非线性系统 ,具有较好的学习精度和概括能力 ,已成功应用于钢铁材料力学性能的预测。使用人工神经网络模型 ,通过输入合金元素、组织、生产工艺参数可预测钢铁材料的抗拉强度、延伸率、韧性、疲劳和蠕变性能。概要叙述了人工神经网络在预测板材、球墨铸铁的常温力学性能 ,合金结构钢的淬透性 ,高速钢。