稀土萃取分离过程软测量方法的研究

分析了稀土组分含量在线检测技术进展缓慢的原因,分析了稀土萃取分离过程机制,找出影响稀土组分含量检测的主要因素,确定辅助变量。利用从包钢稀土厂采集的数据通过newrb函数创建径向基网络来实现稀土组分含量的软测量。文中通过流程图详细阐述了建模方法,采用50组样本数据对所建立的模型进行计算机仿真验证,仿真结果证明了模型的可行性。     

Simultaneous Titration of Ternary Mixtures of Pb(II), Cd(II) and Cu(II) with Potentiometric Electronic Tongue Detection

An automatic titration method is reported to resolve ternary mixtures of transition metals (Pb²⁺, Cd²⁺ and Cu²⁺) employing electronic tongue detection and a reduced number of pre‐defined additions of EDTA titrant. Sensors used were PVC membrane selective electrodes with generic response to heavy‐metals, plus an artificial neural network response model. Detection limits obtained were ca. 1 mg L^?1 for the three target ions and reproducibilities 3.0 % for Pb²⁺, 4.1 % for Cd²⁺ and 5.2 % for Cu²⁺. The system was applied to contaminated soil samples and high accuracy was obtained for the determination of Pb²⁺. In the determination Cd²⁺ and Cu²⁺, sample matrix showed a significant effect.     

参数灵敏度分析的神经网络方法及其工程应用

在系统分析中，参数灵敏度分析不仅为判断各系统参数的重要性大小提供了依据，量化的灵敏度指标也是后续参数估计的前提。然而，在多效实际系统中，系统参数与系统状态间的显式函数关系不易得到，导致一阶灵敏度指标无法直接求取。简化的单因素分析方法亦存在模型粗糙、精度不高的缺点。本文研究采用人工神经网络的高精度泛化映射，通过少量样本的训练，建立复杂系统中多个系统参数与系统状态间的近似映射关系，继而推导得到统一的灵敏度计算列式。简单结构的神经网络方法和解析方法的对比计算显示了方法的有效性和可靠性。最后，应用该法对某斜拉桥结构的荷载参数和刚度参数进行了考查，得到一般性结论。     

基于网络技术的综合导航系统研究

通过对局域网的分析,提出了基于1553B数据总线技术综合导航系统设计方案,并对其可行性进行了充分论证,这种方案在很大程度上保证了系统的可靠性和实时性。     

STABILITY ANALYSIS OF HOPFIELD NEURAL NETWORKS WITH TIME DELAY

ＩｎｔｒｏｄｕｃｔｉｏｎＩｎｒｅｃｅｎｔｙｅａｒｓｔｈｅｄｙｎａｍｉｃｂｅｈａｖｉｏｒｏｆｔｈｅｆｏｌｌｏｗｉｎｇＨｏｐｆｉｅｌｄｎｅｕｒａｌｎｅｔｗｏｒｋｓｍｏｄｅｌｗｉｔｈｔｉｍｅｄｅｌａｙｈａｖｅｂｅｅｎｉｎｖｅｓｔｉｇａｔｅｄｔｈｏｒｏｕｇｈｌｙＣｉ ｘｉ(ｔ) =-ｘｉＲｉ + ∑ｎｊ=1Ｔｉｊｆｊ(ｘｊ(ｔ-τｉｊ) ) +Ｉｉ　　 (ｉ=1 ,2 ,… ,ｎ) ,( 1 )ｗｈｅｒｅＲｉ,ＣｉａｎｄＩｉｒｅｐｒｅｓｅｎｔｒｅｓｉｓｔａｎｃｅ ,ｃａｐａｃｉｔａｎｃｅａｎｄｅｌｅｃｔｒｉｃｃｕｒｒｅｎｔ.Ｒｉ,Ｃｉａｒｅｐａ…     

Filtration in a Porous Granular Medium: 2. Application of Bubble Model to 1-D Column Experiments

This paper describes the formulation of a quasi-1-D network model, referred to as the ‘bubble model’, and its application for simulating particle transport and filtration through a granular filter bed. The model comprises a series of homogeneous sites linked through bundles of cylindrical bonds that represent flow pathways through distributions of pores and pore throats. This model incorporates pore scale processes of particle sieving and infiltration are based on numerical simulations described in a companion paper. The modeling of infiltration is further refined based on detailed experimental observations and measurements of the filtration of a dilute suspension of acrylic particles through a column of glass beads reported by Yoon et al. (2005 Water Resour. Res., to appear). Their data distinguish (a) between the collection of particles on grain surfaces and at grain-to-grain contact points, and (b) between particles that are fully entrapped and those that are hindered (temporarily collected) and can later become detached. These effects are represented by two parameters that characterize the probability of attachment and are linked to the surface roughness of the grains; one that describes the minimum particle size that can be fully entrapped, and one that describes the detachment rate. These parameters can be readily calibrated from conventional measurements of effluent concentration and effluent particle size distribution. Detailed comparisons with the data reported by Yoon et al. show that the proposed bubble model is able to achieve reliable predictions of the spatial distribution of particles within the filter bed following phases of particle injection and washing.     

Delay-independent exponential stability of stochastic Cohen–Grossberg neural networks with time-varying delays and reaction–diffusion terms

Different from the approaches used in the earlier papers, in this paper, the Halanay inequality technique, in combination with the Lyapunov method, is exploited to establish a delay-independent sufficient condition for the exponential stability of stochastic Cohen–Grossberg neural networks with time-varying delays and reaction–diffusion terms. Moreover, for the deterministic delayed Cohen–Grossberg neural networks, with or without reaction–diffusion terms, sufficient criteria for their global exponential stability are also obtained. The proposed results improve and extend those in the earlier literature and are easier to verify. An example is also given to illustrate the correctness of our results.     

A new microstructure-based constitutive model for human blood

A new constitutive equation for whole human blood is derived using ideas drawn from temporary polymer network theory to model the aggregation and disaggregation of erythrocytes in normal human blood at different shear rates. Each erythrocyte is represented by a dumbbell. The use of a linear spring law in the dumbbells leads to a multi-mode generalized Maxwell equation for the elastic stress and both the relaxation times and viscosities are functions of a time-dependent structure variable. An approximate constitutive equation is derived by choosing a single mode corresponding to the cell aggregate size where the largest number of cells are to be found. This size is identified in the case of steady flows. The model exhibits shear-thinning, viscoelasticity and thixotropy and these are clearly related to the microstructural properties of the fluid. Agreement with the experimental data of Bureau et al. [M. Bureau, J.C. Healy, D. Bourgoin, M. Joly, Rheological hysteresis of blood at low shear rate, Biorheology 17 (1980) 191–203] in the case of a simple triangular step shear rate flow is convincing.     

Composite and filament models for the mechanical behaviour of elastomeric materials

In this work we present a composite model, which combines the approach of Poisson's function with the filament theory and requires three material parameters. We also suggest the form for a strain-energy function that approximates the constitutive equations of the composite model. Furthermore, a simple asymptotic analysis allows us to reduce the number of material constants to only two, thus, forming a new filament model. The predictive capability of the two models to reproduce the mechanical behaviour of elastomeric materials in deformation experiments is evaluated against the extensive data of Kawabata et al. (Macromolecules 14 (1981) 154). The models give excellent agreement in not only uniaxial and equibiaxial but also non-equibiaxial extension. Although being rather more simplistic in comparison with some successful network models involving non-Gaussian chain statistics, the two models conform much more closely to the classical experimental data of Treloar (Trans. Faraday Soc. 40 (1944) 59).