Simulation of structural features on mechanochemical synthesis of Al2O3–TiB2 nanocomposite by optimized artificial neural network

In this study, structural features of alumina–titanium diboride nanocomposite (Al₂O₃–TiB₂) were simulated from the mixture of titanium dioxide, boric acid and pure aluminum as raw materials via mechanochemical process using the optimized artificial neural network. The phase transformation and structural evolutions during the mechanochemical process were characterized using X-ray powder diffractometry (XRD). For better understanding the refining crystallite size and amorphization phenomena during the milling, XRD data were modeled and simulated by artificial neural network (ANN). An ANN consisting of three layers of neurons was trained using a back-propagation learning rule. Also, the ANN was optimized by Taguchi method. Additionally, the crystallite size, interplanar distance, amorphization degree and lattice strain were compared for the simulated values and experimental results.     

Statistical modeling of the mechanical alloying process for producing of Al/SiC nanocomposite powders

Mechanical alloying process was modeled by statistical approach for producing of Al/SiC nanocomposite powders. The process variables included two dimensionless variables TV where T and V are milling time and speed, respectively, and P1/P2 where P1 and P2 are balls weight and powders weight, respectively. Responses of the process were crystallite size of the aluminum matrix, lattice strain of the aluminum matrix, and mean particle size of nanocomposite powders. The response variables were obtained by X-ray diffraction patterns (XRD), transmission electron microscopy (TEM), and laser particle size analyzer (LPSA). Two statistical models namely, fixed effects and regression model were developed. Analysis of variance (ANOVA) at 5% levels of significance for fixed effects model and 1% for regression model were performed. Results showed that P1/P2 has a significant effect on the crystallite size, and lattice strain of the aluminum matrix and TV has a significant effect on the crystallite size, and lattice strain of the aluminum matrix as well as mean particle size of nanocomposite powders. ANOVA for regression model showed that the linear effects of TV and P1/P2 variables were significant for crystallite size, lattice strain of the aluminum matrix, and mean particle size of nanocomposite powders. The final regression models were checked and accepted by residual analysis.     

Evaluation of structure and morphology of aluminum powder particles milled at different conditions

Micron sized aluminum powder was milled together with different process control agents (PCAs) including ethanol, graphite and stearic acid for various periods of time. The morphology of powder particles was characterized quantitatively using an image processing program and optical microscopy (OM). Work hardening effect on final particles morphology was evaluated by a number of structural characteristics such as dislocation density and crystallite size calculated by the modified Warren–Averbach method. Normal distribution curves for three morphological parameters of Feret diameter, aspect ratio and roughness, were obtained. The results showed that the type of PCA used during the milling operation was much more effective parameter on morphology, dislocation density and crystallite size of powder particles in comparison with the milling time. The specimens with higher work hardening characteristics showed smaller Feret diameter and aspect ratio. Also, ethanol as a liquid PCA found to be more effective comparing with other solid PCAs.     

基于BP和RBF神经网络对静电纺丝工艺参数的优化研究

针对静电纺丝在制备过程中易受到如聚合物含量、电压、推进速度和接收距离等工艺参数影响的问题,提出一种静电纺丝工艺参数的优化方法,以提升纳米纤维制备效率。以聚乳酸纳米纤维膜为研究对象,采用纤维直径为性能评价指标,设计实验获得训练和测试样本,借助BP(Back Propagation)和RBF(Radial Basis Function)神经网络构建不同工艺参数下的预测模型。结果表明：BP和RBF神经网络模型均能较好的对纤维直径进行预测,但RBF神经网络模型预测精度更高,其平均绝对误差(MAE)为12.125 nm,相对误差不超过7%。RBF神经网络建立的预测模型具有更高的稳定性,模型泛化能力更好,综合预测性能更加优越。所建立的模型可以帮助研究人员制备具有确定纤维直径的静电纺丝纳米纤维膜,实现对工艺参数的优化。     

Microstructural characterization of amorphous and nanocrystalline boron nitride prepared by high-energy ball milling

Microstructural parameters like crystallite size, lattice strain, stacking faults and dislocation density were evaluated from the X-ray diffraction data of boron nitride (BN) powder milled in a high-energy vibrational ball mill for different length of time (2-120 h), using different model based approaches like Scherrer analysis, integral breadth method, Williamson-Hall technique and modified Rietveld technique. From diffraction line-broadening analysis of the successive patterns of BN with varying milling time, it was observed that overall line broadening was an operative cause for crystallite size reduction at lower milling time (∼5 h), whereas lattice strains were the prominent cause of line broadening at higher milling times (>19 h). For intermediate milling time (7-19 h), both crystallite size and lattice strain influence the profile broadening although their relative contribution vary with milling time. Microstructural information showed that after long time milling (>19 h) BN becomes mixture of nanocrystalline and amorphous BN. The accumulations of defects cause this crystalline to amorphous transition. It has been found that twin fault (β′) and deformation fault (α) significantly contributed to BN powder as synthesized by a high-energy ball-milling technique. Present study consider only three ball-milled (0, 2 and 3 h) BN powder for faults calculation because fault effected reflections (1 0 1, 1 0 2, 1 0 3) disappear with milling time (>3 h). The morphology and particle size of the BN powders before and after ball milling were also observed in a field emission scanning electron microscope (FESEM).     

RBF神经网络在爆破振动强度预测中的应用

在介绍RBF神经网络基本思想的基础上,建立了爆破振动预测模型,用RBF神经网络方法对质点振幅、主振频率及振动持续时间进行预测。用阳泉煤矿主井爆破开挖工程中所监测到的振动数据对模型进行了训练,并对27组数据进行了预测,实测结果和模型预测结果的对比表明,RBF神经网络预测模型能反映影响因素与特征量之间的非线性关系,适用于爆破振动特征参量预测。     

Effect of ball milling on the physical and mechanical properties of the nanostructured Co–Cr–Mo powders

In the present study, Co-based machining chips (P1) and Co-based atomized alloy (P2) has been processed through planetary ball mill in order to obtain nanostructured materials and also to comprise some their physical and mechanical properties. The processed powders were investigated by X-ray diffraction technique in order to determine several microstructure parameters including phase fractions, the crystallite size and dislocation density. In addition, hardness and morphological changes of the powders were investigated by scanning electron microscopy and microhardness measurements. The results revealed that with increasing milling time, the FCC phase peaks gradually disappeared indicating the FCC to HCP phase transformation. The P1 powder has a lower value of the crystallite size and higher degree of dislocation density and microhardness than that of the P2 powder. The morphological and particle size investigation showed the role of initial HCP phase and chemical composition on the final processed powders. In addition results showed that in the first step of milling the crystallite size for two powders reach to a nanometer size and after 12 h of milling the crystallite size decreases to approximately 27 and 33 nm for P1 and P2 powders, respectively.     

基于RBF神经网络的镍基TiN纳米复合镀层显微硬度预测

针对神经网络在预测复合镀层性能方面的应用情况,以及传统的BP神经网络存在缺陷;通过对RBF神经网络的基本原理和特点的研究,建立了利用RBF神经网络对Ni-TiN纳米复合镀层显微硬度进行预测的模型。通过实验数据验证了所建立的RBF神经网络模型具有很高的精确度,其最小相对误差可达0.62%,而且所建立的预测模型具有优化工艺参数的功能,对复合镀层的其它性能进行预测具有指导意义。     

Single step synthesis of nanocrystalline ZnO via wet-milling

In this work, we describe the effect of milling speed on the formation, crystallite size, and lattice parameter of nanocrystalline ZnO in a single step process that is based on wet-milling of metallic Zn in distilled water. The samples were characterized by XRD, TEM, and FTIR spectra. The analyses reveal that although the 150 rpm milled sample exhibits imprints of Zn (OH)₂, 200, 250, 300, and 350 rpm milled samples possess the standard hexagonal ZnO wurtzite structure. The crystallite size and lattice parameters of the samples were calculated from the XRD patterns by applying the Maud refinement procedure. According to the results, average crystallite size of the ZnO nanocrystals is in the range of 27.3-31.4 nm depending on the milling speed.     

Friction mode and shock mode effect on magnetic properties of mechanically alloyed Fe-based nanocrystalline materials

The so called “vario mill” (P4 Fritsch) planetary ball mill has been used to prepare nanocrystalline Fe-10 wt% Ni and Fe-20 wt% Ni alloys from powder mixtures. For both studied alloys, a disordered body cubic centered (BCC) solid solution has been formed after 36 h of milling. The higher the shock power, the larger the lattice parameter of the investigated systems. It has been found that in friction mode process (FMP), the lower the crystallite size, the lower the lattice strain of the prepared alloy. In shock mode process (SMP), the lower the crystallite size, the higher the lattice strain. FMP has been found to induce a soft magnetic behavior for Fe-10% Ni and F-20% Ni alloys. The highest values of coercivity have been found in materials prepared by SMP.     

X-Ray diffraction study of the microstructure of C60 fullerite powder after grinding in a planetary ball mill

The profiles of diffraction peaks of C60 fullerite were used to follow changes in the microstructure of C60 powder during grinding in a planetary ball mill at a rotation rate of 500 rpm and milling time of 8 h. The results demonstrate that grinding for just 1 h leads to a significant decrease in crystallite size. With increasing milling time, the average crystallite size varies little, whereas the lattice strain increases considerably.     

Powder Metallurgical Fabrication and Characterization of Nanostructured Porous NiTi Shape-Memory Alloy

Production of NiTi alloy from elemental powders was conducted by mechanical alloying (MA) and sintering of the raw materials. Effects of milling time and milling speed (RPM) on crystallite size, lattice strain, and XRD peak intensities were investigated by X-ray analysis of the alloy. Powder compaction and sintering time and temperature effects on pore percentage of the as-mixed and the mechanically alloyed samples were empirically evaluated. The crystallite size of the mechanically alloyed Ni₅₀Ti₅₀ samples decreased with MA duration and with the milling speed. Depending on the crystal structure of the raw materials, the lattice strain increases with the milling duration. Metallographic studies proved the existence of martensitic B19' after sintering of both the as-mixed and the mechanically alloyed samples. Its amount was, however, greater for the former. Sintering lowered the porosity of the samples; no matter what powder (as-mixed or mechanically alloyed) was used. The porosity was greater, however, for the MA powders. This difference seemed to be due to the sharper liquid phase sintering effect of the as-mixed samples.     

Using GA–ANN algorithm to optimize soft magnetic properties of nanocrystalline mechanically alloyed Fe–Si powders

In this investigation a theoretical model based on artificial neural network (ANN) and genetic algorithm (GA) has been developed to optimize the magnetic softness in nanocrystalline Fe–Si powders prepared by mechanical alloying (MA). The ANN model was used to correlate the milling time, chemical composition, milling speed, and ball to powders ratio (BPR) to coercivity and crystallite size of nanocrystalline Fe–Si powders. The GA–ANN combined algorithm was incorporated to find the optimal conditions for achieving the minimum coercivity. By comparing the predicted values with the experimental data it is demonstrated that the combined GA–ANN algorithm is a useful, efficient and strong method to find the optimal milling conditions and chemical composition for producing nanocrystalline Fe–Si powders with minimum coercivity.     

基于RBF神经网络的色空间转换模型

研究了RBF神经网络的结构及算法,应用RBF神经网络建立了打印机的色空间转换模型.根据实验数据,对网络结构进行了优化,通过比较不同参数时网络的性能,确定最优网络参数.最后对所建模型进行了仿真验证,验证结果表明,预测数据与实测数据的色差较小,说明该模型具有实用价值.     

Particle fracture and plastic deformation in vanadium pentoxide powders induced by high energy vibrational ball-mill

An X-ray powder profile analysis in vanadium pentoxide powder milled in a high energy vibrational ball-mill for different lengths of time (0–250 h), is presented. The strain and size induced broadening of the Bragg reflection for two different crystallographic directions ([001] and [100]) was determined by Warren-Averbach analysis using a pattern-decomposition method assuming a Pseudo-Voigt function. The deformation process caused a decrease in the crystallite size and a saturation of crystallite size of ∼ 10 nm was reached after severe milling. The initial stages of milling indicated a propensity of size-broadening due to fracture of the powder particles caused by repeated ball-to-powder impact whereas with increasing milling time microstrain broadening was predominant. WA analysis indicated significant plastic strain along with spatial confinement of the internal strain fields in the crystallite interfaces. Significant strain anisotropy was noticed in the different crystallographic directions. A near-isotropy in the crystallite size value was noticed for materials milled for 200 h and beyond. The column-length distribution function obtained from the size Fourier coefficients progressively narrowed down with the milling time.     

基于RBF神经网络的自动包装机温度控制算法研究

目的针对传统热封工艺中温度调节PID算法参数过度依赖人工经验的缺点,提出一种RBF神经网络与PID算法相结合的具有参数自适应的热封温度控制算法。方法使用控制系统的输出误差作为代价函数,采用最小均方误差(LMS)调整权值与偏置参数,并通过中心自组织算法实现径向基函数中心和中心宽度的动态调节,在Matlab软件中的Simulink子系统中建立仿真模型进行算法验证,并与传统PID控制算法进行比较。结果仿真结果表明,径向基神经网络与传统PID算法的结合使得系统输出响应在动态性能和静态性能方面均优于传统PID,在系统上升时间、调节时间等方面均优于增量式数字PID。结论将RBF神经网络PID算法应用于自动包装机,避免了传统热封工艺中PID控制算法参数不能适应于复杂变换控制环境的问题,神经网络PID算法的自适应性强,实现了热封温度变化下PID参数的自动调整,在一定程度上提升了生产效率和包装设备的智能化水平。     

Structural evolution in nano-crystalline Cu synthesized by high energy ball milling

Nano-crystalline copper with a mean crystallite size of 27 nm was synthesized through solid state reduction of Cu₂O by graphite using high energy planetary ball mill. The structural and morphological changes during mechanical milling were investigated by X-ray diffraction (XRD), scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The mean crystallite size and residual strain were determined by XRD peak broadening using the Williamson–Hall approximation. It was found that the reaction is completed in a manner like a nucleation and growth process. Although the crystallite size and internal strain changes in Cu₂O were regular during mechanical milling, there was an irregularity in both parameters in Cu particles. This irregularity was probably due to the progressive formation of copper during milling.     

Pore size prediction of polyethersulfone ultrafiltration membranes using artificial neural networks

This paper reports the performance of two different artificial neural networks (ANN), Multi Layer Perceptron (MLP) and Radial Basis Function (RBF) compared to conventional software for prediction of the pore size of the asymmetric polyethersulfone (PES) ultrafiltration membranes. ANN has advantages such as incredible approximation, generalization and good learning ability. The MLP are well suited for multiple inputs and multiple outputs while RBF are powerful techniques for interpolation in multidimensional space. Three experimental data sets were used to train the ANN using polyethylene glycol (PEG) of different molecular weights as additives namely as PEG 200, PEG 400 and PEG 600. The values of the pore size can be determined manually from the graph and solve it using mathematical equation. However, the mathematical solution used to determine the pore size and pore size distribution involve complicated equations and tedious. Thus, in this study, MLP and RBF are applied as an alternative method to estimate the pore size of polyethersulfone (PES) ultrafiltration membranes. The raw data needed for the training are solute separation and solute diameter. Values of solute separation were obtained from the ultrafiltration experiments and solute diameters ware calculated using mathematical equation. With the development of this ANN model, the process to estimate membrane pore size could be made easier and faster compared to mathematical solutions.     

Microstructural characterization of Mg-SiC nanocomposite synthesized by high energy ball milling

High-energy ball milling is successfully used to produce magnesium matrix nanocomposites reinforced with SiC nanoparticles. Changes in morphology and microstructural features of the milled powders were characterized in order to highlight advantages of the mechanical milling process and evaluate the role of the SiC nanoparticles. It was observed that with increasing volume fraction of SiC nanoparticles, a finer nanocomposite powder with more uniform particle size distribution is obtained. A homogeneous distribution of SiC nanoparticles, even up to 10% volume fraction, in magnesium matrix after 25?h milling was confirmed by elemental mapping and TEM results. The analysis of the XRD patterns accompanied by dark-field TEM images revealed that magnesium crystallites refine to fine nanocrystalline sizes after the mechanical milling. The results showed that the crystallite size of the magnesium matrix reduced with increasing SiC nanoparticle content in addition to the induced lattice strain.     

基于RBF网络的步行式底盘内力计算优化方法

为了确定步行式底盘局部结构在作业时的最大受力状态,提出了一种基于RBF神经网络的两级优化模型求解方法,第一级优化模型用逐步二次规划法找到局部结构在给定位置参数下的最大受力状态,通过正交试验设计,利用RBF网络构造出局部结构界面最大受力状态与位置参数之间的非线性映射关系;第二级优化模型用GA求解RBF网络的最大值,并通过二分法不断缩小位置参数的搜索空间,提高RBF网络的逼近水平。研究表明,计算结果可为步行式底盘设计提供理论依据,该方法是解决复杂结构系统中非线性、多变量优化问题的有效手段。