Batch process modelling with mixtures of Gaussian processes

In recent years Gaussian processes have attracted a significant amount of interest with the particular focus being that of process modelling. This has primarily been a consequence of their good predictive performance and inherent analytical properties. Gaussian processes are a member of the family of non-parametric Bayesian regression models and can be derived from the perspective of neural networks. Their behaviour is controlled through the structure of the covariance function. However, when applied to batch processes, whose data exhibits different variance structures throughout the duration of the batch, a single Gaussian process may not be appropriate for the accurate modelling of its behaviour. Furthermore there are issues with respect to the computational costs of Gaussian processes. The implementation of a Gaussian process model requires the repeated computation of a matrix inverse whose order is the cubic of the number of training data points. This renders the algorithm impractical when dealing with large data sets. To address these two issues, a mixture model of Gaussian processes is proposed. The resulting prediction is attained as a weighted sum of the outputs from each Gaussian process component, with the weights determined by a Gaussian kernel gating network. The model is implemented through a Bayesian approach utilising Markov chain Monte Carlo algorithms. The proposed methodology is applied to data from a bench-mark batch simulation polymerization process, methyl methacrylate (MMA), and the results are compared with those from a single Gaussian process to illustrate the advantages of the proposed mixture model approach.     

Vibration analysis of cable-driven parallel manipulators

A cable-driven parallel manipulator is a manipulator whose end-effector is driven by a number of parallel cables instead of rigid links. Since cables always have more flexibility than rigid links, a cable manipulator bears a concern of possible vibration. Thus, investigation of vibration of cable manipulators caused by cable flexibility is important for applications requiring high system stiffness or bandwidth. This paper provides a vibration analysis of general 6-DOF cable-driven parallel manipulators. Based on the analysis of the natural frequencies of the multibody system, the study demonstrates that a cable manipulator can be designed stiff enough for special applications like the cable-manipulator based hardware-in-the-loop simulation of contact dynamics. Moreover, under an excitation, a cable may vibrate not only in its axial direction, but also in its transversal direction. The paper also analyzes the vibration of cable manipulators caused by cable flexibilities in both axial and transversal directions. It is shown that the vibration of a cable manipulator due to the transversal vibration of cables can be ignored comparing to that due to the axial flexibility of cables.     

基于多信息融合技术的智能汽车防盗报警系统研究与设计

为完善汽车的防盗系统,提出研制的智能汽车防盗报警器是在ARM嵌入式系统的基础上,利用Dempster--Sharer证据理论融合密码识别、加速度传感、GPS全球定位信息,判断车辆的安全状况,再将车况信息通过GPRS无线通信网络发送到车主手机。实验室测试表明,系统实现了监控基本功能并满足实时性,增强了汽车安全性和智能报警性。     

GPU加速的圆弧轨迹Katsevich锥束CT重建算法

通过改进GPU的渲染到纹理策略加速锥束反投影计算,提高了圆弧轨迹Katsevich锥束CT重建算法的执行效率.在CPU中完成投影数据的滤波处理,并将其重排为四通道浮点纹理载人GPU内存;绑定2个浮点输出纹理到帧缓冲对象的不同绑定点,用于将渲染结果输出到纹理,并同时完成反投影数据的累加计算;将矩形网格编码为四叉树结构用于加速物体切片的绘制.实验结果表明,文中算法在较大锥角时仍能重建出高质量图像,并获得接近于实时的重建速度.     

嵌入式系统的安全技术研究

随着嵌入式系统的网络化发展,嵌入式系统的安全将成为嵌入式系统必须考虑的问题。嵌入式系统的资源有限性使它的安全问题比一般的桌面系统更加复杂。因此,如何构建一个安全的嵌入式系统是个很值得研究的问题。同时,嵌入式系统的安全不是安全嵌入式系统的附加功能,安全嵌入式系统要在整个设计过程中考虑安全因素,从而决定采用哪种安全技术。从安全的整体性出发,研究了嵌入式系统的各种安全技术,总结归纳出几种安全增强方式,并指出其相应的应用特点,为如何建立嵌入式系统的安全提供依据。     

Face anti-spoofing based on multi-modal and multi-scale features fusion

Face anti-spoofing is used to assist face recognition system to judge whether the detected face is real face or fake face. In the traditional face anti-spoofing methods, features extracted by hand are used to describe the difference between living face and fraudulent face. But these handmade features do not apply to different variations in an unconstrained environment. The convolutional neural network (CNN) for face deceptions achieves considerable results. However, most existing neural network-based methods simply use neural networks to extract single-scale features from single-modal data, while ignoring multi-scale and multi-modal information. To address this problem, a novel face anti-spoofing method based on multi-modal and multi-scale features fusion ( MMFF) is proposed. Specifically, first residual network ( Resnet )-34 is adopted to extract features of different scales from each modality, then these features of different scales are fused by feature pyramid network (FPN), finally squeeze-and-excitation fusion ( SEF) module and self-attention network ( SAN) are combined to fuse features from different modalities for classification. Experiments on the CASIA-SURF dataset show that the new method based on MMFF achieves better performance compared with most existing methods.     

Ultra-dense planar metallic nanowire arrays with extremely large anisotropic optical and magnetic properties

A nanofabrication method for the production of ultra-dense planar metallic nanowire arrays scalable to wafer-size is presented. The method is based on an efficient template deposition process to grow diverse metallic nanowire arrays with extreme regularity in only two steps. First, III–V semiconductor substrates are irradiated by a low-energy ion beam at an elevated temperature, forming a highly ordered nanogroove pattern by a “reverse epitaxy” process due to self-assembly of surface vacancies. Second, diverse metallic nanowire arrays (Au, Fe, Ni, Co, FeAl alloy) are fabricated on these III–V templates by deposition at a glancing incidence angle. This method allows for the fabrication of metallic nanowire arrays with periodicities down to 45 nm scaled up to wafer-size fabrication. As typical noble and magnetic metals, the Au and Fe nanowire arrays produced here exhibited large anisotropic optical and magnetic properties, respectively. The excitation of localized surface plasmon resonances (LSPRs) of the Au nanowire arrays resulted in a high electric field enhancement, which was used to detect phthalocyanine (CoPc) in surface-enhanced Raman scattering (SERS). Furthermore, the Fe nanowire arrays showed a very high in-plane magnetic anisotropy of approximately 412 mT, which may be the largest in-plane magnetic anisotropy field yet reported that is solely induced via shape anisotropy within the plane of a thin film.

Open image in new window

     

'98 Escherichia coli SWISS-2DPAGE database update

The combination of two-dimensional polyacrylamide gel electrophoresis (2-D PAGE), computer image analysis and several protein identification techniques allowed the Escherichia coli SWISS-2DPAGE database to be established. This is part of the ExPASy molecular biology server accessible through the WWW at the URL address http://www.expasy.ch/ch2d/ch2d-top.html . Here we report recent progress in the development of the E. coli SWISS-2DPAGE database. Proteins were separated with immobilized pH gradients in the first dimension and sodium dodecyl sulfate-polyacrylamide gel electrophoresis in the second dimension. To increase the resolution of the separation and thus the number of identified proteins, a variety of wide and narrow range immobilized pH gradients were used in the first dimension. Micropreparative gels were electroblotted onto polyvinylidene difluoride membranes and spots were visualized by amido black staining. Protein identification techniques such as amino acid composition analysis, gel comparison and microsequencing were used, as well as a recently described Edman "sequence tag" approach. Some of the above identification techniques were coupled with database searching tools. Currently 231 polypeptides are identified on the E. coli SWISS-2DPAGE map: 64 have been identified by N-terminal microsequencing, 39 by amino acid composition, and 82 by sequence tag. Of 153 proteins putatively identified by gel comparison, 65 have been confirmed. Many proteins have been identified using more than one technique. Faster progress in the E. coli proteome project will now be possible with advances in biochemical methodology and with the completion of the entire E. coli genome.     

Die shape optimisation for net-shape accuracy in metal forming using direct search and localised response surface methods

In this paper, three direct search algorithms, i.e. a modified simplex, random direction search and enhanced Powell’s methods together with a new localised response surface method are presented and applied to solve die shape optimisation problems for achieving net-shape accuracy in metal forming processes. The main motivation is to develop efficient and easy to implement optimisation algorithms in metal forming simulations which often involve complex tool and workpiece interaction and coupled thermal and mechanical analysis. Three case studies are presented including a simple upsetting, a 2D blade forging and a forward extrusion problem. In all cases, the objective was to achieve net-shape accuracy of the formed parts, one important criterion for precision forming. C+ + programs were developed to implement these algorithms and to automatically integrate optimisation computation and forging simulation. The optimisation results from the three case problems show that direct search based methods especially the modified simplex and the localised response surface methods are computationally efficient and robust for net-shape forging and extrusion optimisation problems. It is also suggested that these methods can be used in more complex forging problems where die shape design and optimisation are essential for achieving net-shape accuracy.