Fabrication and properties of CNTs/carbon fabric hybrid multiscale composites processed via resin transfer molding technique

Carbon nanotubes (CNTs) are effective fillers/reinforcements regarding improving the properties of polymer. In the present paper, carboxylic acid functionalized CNTs were used to modify epoxy with intent to develop a nanocomposite matrix for hybrid multiscale composites combining benefits of nanoscale reinforcement with well-established fibrous composites. CNTs were dispersed in epoxy by using high energy sonication. At low contents of CNTs, hybrid multiscale composites specimens were manufactured via resin transfer molding (RTM) process. The processibility of CNTs/epoxy systems was explored with respect to their viscosity. The dispersion quality and re-agglomeration behavior of CNTs in epoxy were characterized using optical microscope. A CNTs loading of 0.025 wt% significantly improved the glass transition temperatures (Tg) of the hybrid multiscale composites. Scanning electron microscopy (SEM) was used to examine the fracture surface of the failed specimens. It is demonstrated that the addition of small amount of CNTs (0.025 wt%) to epoxy for the fabrication of multiscale carbon fabric composites via RTM route effectively improves the matrix-dominated properties of polymer based composites. Hybridization efficiency in carbon fiber reinforced composites using CNTs is found to be highly dependent on the changes in the dispersion state of CNTs in epoxy.     

Optimal design for the flexural behaviour of glass and carbon fibre reinforced polymer hybrid composites

A study on the flexural behaviour of hybrid composites reinforced by S-2 glass and T700S carbon fibres in an intra-ply configuration is presented in this paper. The three point bend test in accordance with ASTM D790-07 at various span-to-depth ratios was simulated using finite element analysis (FEA). For the purpose of validation, specimens of selected stacking configurations were manufactured following the hand lay-up process and tested in a three point bend configuration. The validated FEA model was used to study the effects of fibre volume fractions, hybrid ratio and span-to-depth ratio. It is shown that flexural modulus increases when the span-to-depth ratio increases from 16 to 32 but is approximately constant as the span-to-depth ratio further increases. A simple mathematical formula was developed for calculating the flexural modulus of hybrid composites, given the moduli of full carbon and full glass composites, and the hybrid ratio. Flexural strength increases with span-to-depth ratio. Utilisation of hybridisation can improve the flexural strength. A general rule is in order to improve flexural strength, the fibre volume fraction of glass/epoxy plies needs to be higher than that of carbon/epoxy plies. The overall maximum hybrid effect is achieved when the hybrid ratio is 0.125 ([0_G/0_7C]) when both V_fc and V_fg are 50%. The strength increases are 43.46% and 85.57% when compared with those of the full carbon and glass configurations respectively. The optimisation shows that the maximum hybrid effect is 56.1% when V_fc = 47.48% and V_fg = 63.29%.     

A parametric cost model for mineral grinding mills

The adequate cost estimation of mill plants plays a crucial role in the success of feasibility studies of mining projects. Grinding is one of the most important operations in mineral processing plants and assumes a substantial share of the total milling costs. The objective of this work was to develop a set of cost functions for major grinding mill equipment. These cost models were developed using two relatively different techniques: uni-variate regression (UVR) as well as multivariate regression (MVR) based on principal component analysis (PCA). The first is appropriate for the quick estimation of costs in the early stages of project evaluation, while the second method can be helpful in the feasibility study stage. The explanatory variable in UVR was power (P), while in MVR the power and some other variables depending on the type of mill were used. The PCA technique was employed in order to omit the correlation between the independent variables in the multivariate regression. Furthermore, the scale-up factor for all mills has been calculated. The result of the evaluation of the models showed that the mean absolute error rates were less than 9.84% and 11.36% on average for the capital and operating costs of the uni-variate model, and 5.82% and 4.9% for the multivariate model, respectively.     

Numerical investigation of CH4/O2 mixing in Y-shaped mesoscale combustors with/without porous media

A good mixing of reactants is important for non-premixed combustion in miniature combustors. In this paper, mixings of methane and oxygen in Y-shaped mesoscale combustors with and without porous media were compared numerically. The results show that when there is no porous media in the horizontal channel, the mixing becomes worse with the decrease of the included angle between two inlets, or with the increase of inlet velocity. The reason is that for the case without porous media, the dominant mixing mechanism is molecular diffusion under concentration gradients. In contrast, for the case with porous media, due to the mass dispersion effect which becomes more significant with the decrease of channel width, satisfactory mixing can always be attained in the Y-shaped mesoscale combustor. Moreover, fairly good mixing can still be achieved in the horizontal channel of shorter length. All these demonstrate that the porous media greatly promotes the methane/oxygen mixing in the Y-shaped mesoscale combustor, which is beneficial for flame stabilization. Meanwhile, the combustor dimension can be further scaled down because good mixing is possible in the channel with even smaller included angle and shorter length. This is very important for the application of miniature power generation system.     

Pressure prediction on a variable-speed pump controlled hydraulic system using structured recurrent neural networks

This paper presents a study to predict the pressures in the cylinder chambers of a variable-speed pump controlled hydraulic system using structured recurrent neural network topologies where the rotational speed of the pumps, the position and the average velocity of the hydraulic actuator are used as their inputs. The paper elaborates the properties of such networks in extended time periods through detailed simulation- and experimental studies where black-box modeling approaches generally fail to yield acceptable performance. As alternative estimation techniques, both linear- and extended Kalman filters are considered in this paper. The estimation properties of the devised network models are comparatively evaluated and their potential application areas are discussed in detail.     

Continuous generating grinding method for beveloid gears and analysis of grinding characteristics

Continuous generating grinding has become an important gear processing method owing to its high efficiency and precision. In this study, an adaptive design model is proposed for the continuous generation of beveloid gears in common gear grinding machines. Based on this model, a method for determining the installation position and grinding kinematics is developed alongside an analytical meshing model for grinding contact trace and derivation of key grinding parameters. By combining these aspects, a general mathematical model for the continuous generation of beveloid gears is presented, comprising the entire grinding process from worm wheel dressing to the evaluation of grinding deviation. The effects of the worm and dressing wheel parameters on the grinding deviation were analysed, facilitating the development of an approach to improve the grinding accuracy. The presented procedure represents a novel design method for the continuous generation of beveloid gears in common gear grinding machines, facilitating the appropriate selection of worm and dressing wheel parameters.The full text can be downloaded at https://link.springer.com/article/10.1007/s40436-022-00388-z     

Combining empirical mode decomposition and deep recurrent neural networks for predictive maintenance of lithium-ion battery

Predictive maintenance of lithium-ion batteries has been one of the popular research subjects in recent years. Lithium-ion batteries can be used as the energy supply for industrial equipment, such as automated guided vehicles and battery electric vehicles. Predictive maintenance plays an important role in the application of smart manufacturing. This mechanism can provide different levels of pre-diagnosis for machines or components. Remaining useful life (RUL) prediction is crucial for the implementation of predictive maintenance strategies. RUL refers to the estimated useful life remaining before the machine cannot operate after a certain period of operation. This study develops a hybrid data science model based on empirical mode decomposition (EMD), grey relational analysis (GRA), and deep recurrent neural networks (RNN) for the RUL prediction of lithium-ion batteries. The EMD and GRA methods are first adopted to extract the characteristics of time series data. Then, various deep RNNs, including vanilla RNN, gated recurrent unit, long short-term memory network (LSTM), and bidirectional LSTM, are established to forecast state of health (SOH) and the RUL of lithium-ion batteries. Bayesian optimization is also used to find the best hyperparameters of deep RNNs. Experimental results with the lithium-ion batteries data of NASA Ames Prognostics Data Repository show that the proposed hybrid data science model can accurately predict the SOH and RUL of lithium-ion batteries. The LSTM network has the optimal results. The proposed hybrid data science model with multiple artificial intelligence-based technologies also demonstrates critical digital-technology enablers for digital transformation of smart manufacturing and transportation.     

Force control-based vibration suppression in robotic grinding of large thin-wall shells

Vibration suppression is a major difficulty in the grinding of low-stiffness large thin-wall shells. The paper proposes that effective workpiece vibration control can be performed by a novel force-controlled end-effector integrated into a robotic grinding workcell. First, a dynamics model is built to capture the characteristics and vibration suppression mechanism of force control-based robotic grinding, then a novel force control-based vibration suppression method is designed for grinding large thin-wall shells, and three robotic grinding tests are conducted to validate the effects of the new method and the grinding performance of the force control-based robotic grinding workcell. The results are: 75% reduction in the amplitude of workpiece vibration; effective suppression of non-tool passing frequency; stable grinding of large thin-wall shells remarkably enhancing grinding depth up to 0.3 mm per pass, grinding depth error less than ±0.1 mm, and significant improvement of the workpiece surface quality up to Ra=0.762 μm.     

基于改进U-Net网络的齿轮点蚀测量

针对U-Net存在的小目标分割精度低、计算复杂度高、收敛慢的问题,构建基于空洞卷积和重构采样单元的U-Net网络(DSU-Net).在DSU-Net中,为增大图像特征提取的感受野并融合多尺度信息,设计具有不同膨胀率的空洞卷积层;针对池化过程丢失大量语义信息的缺点,构建将池化与卷积相结合的采样单元,并运用深度可分离卷积进行特征提取,从而增强神经网络的特征提取能力并降低计算成本.两个公开医学图像数据集的实验结果表明,在IoU、Dice Coeff和F1 Score三个评价指标上,DSU-Net较U-Net、ResU-Net、R2U-Net和U-Net++有着更好的分割性能.最后,将DSU-Net应用于齿轮点蚀的视觉测量,结果表明所提出方法能够更加精确地计算出齿轮点蚀面积率,从而解决了齿轮接触疲劳试验中高效准确检测齿轮失效的难题.