基于LSTM的浮选钼精矿品位软测量模型

战略性稀有金属钼矿品位低,组分复杂、嵌布粒度细等特点,其有价金属分离回收难。浮选作为微细粒钼矿分离回收的主要选矿方法之一,其浮选钼精矿品位一直是选厂的关键性产品指标。国内大多数选厂采取轮班制采样,人工化验得到精矿品位结果,但此方式严重滞后于浮选工艺,难以满足对生产过程进行实时监测和操作指导。LSTM是一种特殊的循环神经网络,引入门机制有效的传递或选择性遗忘长时间序列中的信息,解决RNN中的长期依赖、梯度消失和爆炸问题。本文分析整理东坡选厂中各平台源数据,结合选厂浮选工艺及机理,筛选出多个影响浮选钼精矿品位的变量作为模型输入;将输入变量进行异常值判定,缺失值填充和数据降噪等数据预处理,建立高质量浮选钼精矿品位数据库;软测量模型采用PyCharm软件编码,使用BatchNorm批量规范化处理样本数据,加入Dropout正则化防止过拟合,建立基于LSTM的浮选钼精矿品位软测量模型,通过前向传播算法更新神经网络结构参数,并于Linear模型和CNN模型的预测性能指标结果比较。结果表明：基于LSTM的浮选钼精矿品位软测量模型预测准确度高,样本数据误差波动平稳,浮动范围小,模型泛化能力强,模型平均绝对百分比误差MAPE为1.13%,均方根误差RMSE为0.7049%,决定系数R2为0.8763,实现了浮选钼精矿品位的在线预测。     

The effect of MWS polarisation on the morphology and electro-optical behaviour of normal-mode polymer-stabilised cholesteric textures

We investigate the process and the resulting properties of the curing electric-field-induced segregation of reactive monomers (containing polar groups) in a host liquid crystal. The obtained dielectric spectrum of these cholesteric liquid crystals (CLCs) reveals polarisation at low frequencies. This is a strong indication that charge carriers accumulate at the interface between the sample and the electrodes in these compounds and consistent with Maxwell–Wagner–Sillars (MWS) polarisation. A polymer film forms on the substrate with its thickness determined by the MWS polarisation strength. The SEM images show that the polymer network becomes less dense when MWS polarisation increases. The results are a higher operating voltage and faster response time.     

Hydrogels of chemically cross-linked and organ-metallic complexed interpenetrating PEG networks

The aim of the present work was to prepare a well-defined hydrogel of chemically cross-linked and organ-metallic complexed interpenetrating PEG networks. The hydrogel was synthesized via the reaction of copper(I)- catalyzed 1,3-dipolar azide-alkyne cycloaddition(CuA AC) with poly(ethylene glycol)-dopamine(PEG-DA)("Click Chemistry") followed by complexation with Fe~(3+) ions to crosslink the polymeric network. The chemical composition and morphology of the resulting hydrogels were characterized by Fourier transform infrared spectroscopy(FTIR), ~1H-NMR and scanning electron microscopy(SEM). Swelling ratio, mechanical strength, conductivity, and degradation behaviors of the hydrogels were also studied. The effect of the polymer chain length on properties of hydrogels was explored. The compressive strength of hydrogels could reach as high as 13.1 MPa with a conductivity of 2.2 × 10~(-5) S·cm~(-1). The hydrogels also exhibited excellent thermal stability even at a temperature of 300 °C, whereas degradation of the hydrogel after 7 weeks was observed under a physiological condition. In addition, the hydrogel exhibited a good biocompatibility based on its in vivo performance through an in vivo subcutaneous implantation model. No inflammation and no obvious abnormality of the surrounding tissue were observed when the hydrogel was subcutaneously implanted for 2 weeks. This work is a step towards creating a new pathway to synthesize hydrogels of interpenetrating networks which could be of important applications in the future research.     

Simple Chemical Route Synthesis of Fe2O3 Nanoparticles and its Application for Adsorptive Removal of Congo Red from Aqueous Media: Artificial Neural Network Modeling

Nanocrystalline Fe₂O₃ powder was synthesized by a simple chemical route involving FeCl₃ and NaOH. The Fe₂O₃ powder thus prepared was characterized using x-ray diffraction study, scanning electron microscopy, and Fourier transform infrared spectroscopy. The adsorption properties of crystalline Fe₂O₃ powder have been investigated with an aim to explore a possible low cost and efficient way to remove Congo red (CR) from waste water. Fe₂O₃ powder was found as an excellent adsorbent for CR from aqueous medium. Adsorption capacity as much as 203.66 mg g^?1 is reported at room temperature. Effect of different experimental parameters such as reaction pH, initial CR dye concentration, adsorbent dose, and reaction temperature were studied on adsorption capacity of Fe₂O₃ powder and modeled by artificial neural network (ANN). Optimal ANN structure (4–5–1) shows minimum mean squared error (MSE) of 0.00235 and determination coefficient (R²) of 0.991 with Levenberg–Marquardt algorithm. Isotherm analysis of experimental data exhibited better fit to the Langmuir isotherm. The adsorption process was found to follow second-order kinetics as depicted by the analysis of experimental results. Thermodynamic study shows that the adsorption process is endothermic, spontaneous, and thermodynamically favorable in the temperature range of 27°C to 60°C.     

分类问题的在线顺序双并联快速学习机

双并联前馈神经网络模型是单层感知机和单隐层前馈神经网络的混合结构,本文构造了一种双并联快速学习机算法,与其他类似算法比较,提出的算法能利用较少的隐层单元及更少的待定参数,获得近似的学习性能.数值实验表明,对很多实际分类问题,提出的算法具备更佳的泛化能力,因而可以作为快速学习机算法的有益补充.     

Synthesis of Hydrophobic Cationic Chitosan Flocculant and Its Sludge Dewatering Property

The hydrophobic cationic chitosan (HTCC) flocculant was synthesized by reacting chitosan with epoxy propyl trimethyl ammonium chloride (EPTAC) and (2,3-epoxy propyl) dodecyl dimethyl ammonium chloride (EDC), to improve its waste sludge dewaterability. By FTIR and ¹H NMR analysis, EPTAC and EDC were confirmed to be grafted onto the molecular chains of chitosan (CTS) successfully. With the increase of EDC content, G′ of HTCC aqueous solution with 1:1.5:1.5 mole ratio of CTS:EPTAC:EDC consistently showed higher values than G″ over the entire frequency range explored, indicating the formation of a physically cross-linked network. The crystallization ability of HTCC decreased compared with pristine CTS. The long alkyl chains on the EDC molecules enhanced the hydrophobicity of HTCC, and better dewaterability of the waste sludge was achieved through the promoted bridging action of HTCC.     

Ensuring Quality of Experience in the Connected Home: A New Battleground for Quality of Experience Differentiation

ABSTRACT

The connected home is a critical part of the network—but one that has seen little in the way of true investment in recent years. With customer loyalty and churn directly linked to quality of experience, this is a part of the network that no players in the supply chain can afford to ignore any longer. Beyond broadband experience, the addition of smart technology to the connected home also represents one of the biggest revenue growth areas in the consumer telecoms market for some time. From both a churn-reduction and a revenue-growth perspective, investment in the connected home network must accelerate. This article outlines the key trends identified by Ovum in the connected home space for 2016 and beyond.     

Experimental Investigation on Transmission Control Protocol Throughput Behavior in Optical Fiber Access Networks

ABSTRACT

This article describes an experimental investigation on the behavior of transmission control protocol in throughput measurements to be used in the verification of the service-level agreement between the Internet service provider and user in terms of line capacity for ultra-broadband access networks typical of fiber-to-the-x architectures. It is experimentally shown different conditions in high bandwidth-delay product links where the estimation of the line capacity based on a single transmission control protocol session results are unreliable. Simple equations reported in this work, and experimentally verified, point out the conditions in terms of packet loss, time delay, and line capacity, that allow consideration of the reliability of the measurement carried out with a single transmission control protocol session test by adopting a suitable measurement time duration.     

Psychoacoustics of in-car switch buttons: From feelings to engineering parameters

The car interior is becoming quieter and other sounds are now exposed to user perception, such as the sound produced by interface buttons when actuated. So, the functional role of the button sound on interface operation and its aesthetic and emotional role on the user experience are now more important than before. However, little research and design effort has been paid to understand how to design buttons that produce a pleasant sound. Moreover, the button’s sound requirements received by interface manufacturers are ill-defined, insufficient or even inexistent, and consequently their conversion into specifications for manufacturing is problematic and leads to long and costly development processes. The purpose of this paper is to contribute to identify relevant acoustic parameters that explain the users sound preferences. Data on preference subjective judgments were collected and buttons acoustic signals were measured allowing the development of preference models based on partial least squares regression and neural networks methods. The former was successful in selecting the relevant parameters to describe the preference ratings of the buttons sound. The later, dealing with the non-linear nature of acoustic perception, was able to predict preferences based on the relevant parameters.     

Discovery Through Gossip

We study randomized gossip‐based processes in dynamic networks that are motivated by information discovery in large‐scale distributed networks such as peer‐to‐peer and social networks. A well‐studied problem in peer‐to‐peer networks is resource discovery, where the goal for nodes (hosts with IP addresses) is to discover the IP addresses of all other hosts. Also, some of the recent work on self‐stabilization algorithms for P2P/overlay networks proceed via discovery of the complete network. In social networks, nodes (people) discover new nodes through exchanging contacts with their neighbors (friends). In both cases the discovery of new nodes changes the underlying network — new edges are added to the network — and the process continues in the changed network. Rigorously analyzing such dynamic (stochastic) processes in a continuously changing topology remains a challenging problem with obvious applications. This paper studies and analyzes two natural gossip‐based discovery processes. In the push discovery or triangulation process, each node repeatedly chooses two random neighbors and connects them (i.e., “pushes” their mutual information to each other). In the pull discovery process or the two‐hop walk, each node repeatedly requests or “pulls” a random contact from a random neighbor and connects itself to this two‐hop neighbor. Both processes are lightweight in the sense that the amortized work done per node is constant per round, local, and naturally robust due to the inherent randomized nature of gossip. Our main result is an almost‐tight analysis of the time taken for these two randomized processes to converge. We show that in any undirected n‐node graph both processes take rounds to connect every node to all other nodes with high probability, whereas is a lower bound. We also study the two‐hop walk in directed graphs, and show that it takes time with high probability, and that the worst‐case bound is tight for arbitrary directed graphs, whereas Ω(n²) is a lower bound for strongly connected directed graphs. A key technical challenge that we overcome in our work is the analysis of a randomized process that itself results in a constantly changing network leading to complicated dependencies in every round. We discuss implications of our results and their analysis to discovery problems in P2P networks as well as to evolution in social networks. © 2016 Wiley Periodicals, Inc. Random Struct. Alg., 48, 565–587, 2016