Scale effects on strength of geomaterials, case study: Coal

Scale effects on the strength of coal are studied using a discrete element model. The key point of the model is its capability to discriminate between the “strictly sample size” effect and the “Discrete Fracture Network (DFN) density” effect on the mechanical response. Simulations of true triaxial compression tests are carried out to identify their respective roles. The possible bias due to the discretization size distribution of the discrete element model is investigated in detail by considering low-resolution configurations. The model is shown to be capable of quantitatively reproducing the dependency of the maximum strength on the size of the sample. This relationship mainly relies on the DFN density. For all given sizes, as long as the DFN density remains constant with a uniform distribution or if discontinuities are absent in the considered medium, the maximum strength of the material remains constant.     

Artificial neural network prediction to the hot compressive deformation behavior of Al–Cu–Mg–Ag heat-resistant aluminum alloy

The behavior of the flow stress of Al-Cu-Mg-Ag heat-resistant aluminum alloys during hot compression deformation was studied by thermal simulation test. The temperature and the strain rate during hot compression were 340-500 °C, 0.001 s⁻¹ to 10 s⁻¹, respectively. Constitutive equations and an artificial neural network (ANN) model were developed for the analysis and simulation of the flow behavior of the Al-Cu-Mg-Ag alloys. The inputs of the model are temperature, strain rate and strain. The output of the model is the flow stress. Comparison between constitutive equations and ANN results shows that ANN model has a better prediction power than the constitutive equations.     

Robust airfoil optimization based on improved particle swarm optimization method

A robust airfoil optimization platform is constructed based on the modified particle swarm optimization method (i.e., the second-order oscillating particle swarm method), which consists of an efficient optimization algorithm, a precise aerodynamic analysis program, a high accuracy surrogate model, and a classical airfoil parametric method. There are two improvements for the modified particle swarm method compared with the standard particle swarm method. First, the particle velocity is represented by the combination of the particle position and the variation of position, which makes the particle swarm algorithm a second-order precision method with respect to the particle position. Second, for the sake of adding diversity to the swarm and enlarging the parameter searching domain to improve the global convergence performance of the algorithm, an oscillating term is introduced to the update formula of the particle velocity. At last, taking two airfoils as examples, the aerodynamic shapes are optimized on this optimization platform. It is shown from the optimization results that the aerodynamic characteristic of the airfoils is greatly improved in a broad design range.     

一种应用于CMOS图像传感器的电容阵列PGA研究

研究了一种应用于1 024×1 024CMOS图像传感器中的温度计码电容阵列型可编程增益放大器（PGA）。该PGA采用开关电容阵列结构,通过改变输入与反馈电容值,实现了0～18dB的增益;同时,该PGA具有结构简单、电容阵列面积小、带宽大和反馈系数大等特点,其增益带宽50MHz、总谐波失真（THD）小于50dB。该PGA的信号处理能力完全满足12MHz速率、69dB动态范围CMOS图像传感器内部集成ADC对模拟输入信号的要求。     

流体与结构相互作用问题的FE-SPH耦合模拟

应用有限元(FE)-光滑粒子流体动力学(SPH)耦合法模拟了具有自由表面的不可压流体与结构的相互作用问题.流体和结构分别采用SPH法和有限元法同时求解,两者在交界面处的相互作用通过接触算法进行处理.为了避免隐式计算压力,通过引入人工压缩率,将不可压流体近似为人工可压缩流体.采用FE-SPH耦合法对弹性板在随时间变化的水压作用下的变形以及倒塌水柱冲击弹性结构两个问题进行了模拟.模拟结果与实验结果以及其他已有数值结果符合良好,说明FE-SPH耦合法用于流体与结构相互作用问题的模拟是可行和有效的.     

人造柱状冰特性的研究

由于地形的变化,冰特性的研究对研究冰川、河、湖、海冰的动力过程以及破冰均有重要意义.本文对人造柱状冰的特性进行研究和分析,给出了人造柱状冰的制备过程和利用压力试验机对人造冰特性进行的测定,主要包括冰冻模具的制作、冰试样的制备和加工以及对于纯水冰柱和含沙冰柱的特性研究.在冰柱的冰冻温度为-10℃、-20℃、-30℃和-4...     

PLS分析与RBF神经网络耦合环境模型

鉴于城市大气环境质量受到诸多复杂因素影响,且各因素间存在多重相关性,本文将偏最小二乘(PLS)分析与人工神经网络径向基网络(RBF)耦合,建立偏最小二乘径向基神经网络模型(PLSRBF),应用于贵阳大气环境质量的检验和预测。实例表明:PLSRBF模型可对原多自变量模型进行降维简化,并可有效提取解释变量信息,防止信息丢失,且具有较强的拟合能力。     

Co-authorship model with link strength

We generalize a network formation model for co-authorship introducing the possibility of the connections having different link strengths. Different link strengths represent the fact that authors may put different efforts into different collaborations. To evaluate the model, we consider the notions of efficiency and pairwise stability, which are based on a utility function that measures the benefits for an author to belonging to a certain network structure. We divide the analysis in two cases, considering that link strengths are unbounded or bounded. In the first case, we show that if there are more than two authors in the network, then there is no pairwise stable network. In the second case, we show that the pairwise stable networks consist of completely connected disjoint components where essentially all link strengths are maximal. Regarding efficiency, in both cases, if the number of authors is even, then the unique efficient network structure consists of pairs of connected authors.     

Integrating link-based discrete credit charging scheme into discrete network design problem

In this paper, we present an optimization model for integrating link-based discrete credit charging scheme into the discrete network design problem, to improve the transport performance from the perspectives of both transport network planning and travel demand management. The proposed model is a mixed-integer nonlinear bilevel programming problem, which includes an upper level problem for the transport authority and a lower level problem for the network users. The lower level sub-model is the traffic network user equilibrium (UE) formulation for a given network design strategy determined by the upper level problem. The network user at the lower level tries to minimize his/her own generalized travel cost (including both the travel time and the value of the credit charged for using the link) by choosing his/her route. While the transport authority at the upper level tries to find the optimal number of lanes and credit charging level with their locations to minimize the total system travel time (or maximize the transportation system performance). A genetic algorithm is used to solve the proposed mixed-integer nonlinear bilevel programming problem. Numerical experiments show the efficiency of the proposed model for traffic congestion mitigation, reveal that interaction effects across the tradable credit scheme and the discrete network design problem which amplify their individual effects. Moreover, the integrated model can achieve better performance than the sequential decision problems.     

A class of mixture of experts models for general insurance: Theoretical developments

In the Property and Casualty (P&C) ratemaking process, it is critical to understand the effect of policyholders’ risk profile to the number and amount of claims, the dependence among various business lines and the claim distributions. To include all the above features, it is essential to develop a regression model which is flexible and theoretically justified. Motivated by the issues above, we propose a class of logit-weighted reduced mixture of experts (LRMoE) models for multivariate claim frequencies or severities distributions. LRMoE is interpretable, as it has two components: Gating functions, which classify policyholders into various latent sub-classes; and Expert functions, which govern the distributional properties of the claims. Also, upon the development of denseness theory in regression setting, we can heuristically interpret the LRMoE as a “fully flexible” model to capture any distributional, dependence and regression structures subject to a denseness condition. Further, the mathematical tractability of the LRMoE is guaranteed since it satisfies various marginalization and moment properties. Finally, we discuss some special choices of expert functions that make the corresponding LRMoE “fully flexible”. In the subsequent paper (Fung et al., 2019b), we will focus on the estimation and application aspects of the LRMoE.