寻找地震相关地区的时间序列相似性匹配算法

把时间序列相似性匹配的基本概念和方法引入到地震预报的应用中.在分析现阶段时间序列研究成果的基础上,结合大量地震历史源数据和领域专家经验知识,提出了有关地震地区相关性的地震相似度定义和地震序列相似性匹配模型,并通过大量实验模拟对该模型进行了反复验证,实现了基于地震相似度的时间序列相似性匹配算法.同时,通过分析我国地震活动频繁区域近20年来的地震历史数据,应用地震区域序列相似性匹配算法进行了固定时间差的粗粒度和细粒度纵向序列相似性实验分析,取得了可信度较高的实验结果,为地震学预测的应用研究提供了较好的技术支持.     

Kinetic gas molecule optimization for nonconvex economic dispatch problem

This paper presents kinetic gas molecule optimization (KGMO) algorithm to solve economic dispatch problems with non-smooth/non-convex cost functions. KGMO is based on kinetic energy and the natural motion of gas molecules. The effectiveness of the proposed method has been verified on four different non-convex economic dispatch problems with valve-point effects, prohibited operating zones with transmission losses, multiple fuels with valve point effects and the large-scale Korean power system with valve-point effects and prohibited operating zones. The results of the proposed approach are compared with those obtained by other evolutionary methods. It is found that the proposed KGMO based approach is able to provide better solution.     

保税港区服务流程的性能评价

针对保税港区业务背景及服务流程的特点,构建了保税港区服务流程性能管理框架.在此基础上,提出服务流程性能评价模型以及相应的性能评价指标体系.针对保税港区服务流程时效性的特点,提出了相应的实时灰色预测方法,并通过一个应用示例验证了方法的有效性.服务流程性能指标体系综合了保税港区系统中\     

基于阴影集数据选择的可拓神经网络性能改进

为了改进可拓神经网络的性能,提出一种基于阴影集的数据选择方法.通过该方法获取用于训练可拓神经网络的训练样本,进而改进可拓神经网络的性能.针对可拓神经网络的特点,选择核数据和边界数据作为可拓神经网络的训练样本;利用基于阴影集的数据选择方法,可以自动获取核数据和边界数据.实验结果表明,与传统可拓神经网络相比,改进的可拓神经网络不仅节约了训练时间,而且网络的泛化能力和分类识别准确度得到了有效提高.     

高速公路施工区上游过渡区的车辆汇入模型

为确定高速公路施工区上游过渡区长度对高速公路主车道交通流的影响,首先应用实测数据分析施工区上游过渡区主车道的车头时距分布特性,结果表明车头时距分布符合爱尔朗分布;其次在考虑车辆汇入过程中临界间隙变化情况的基础上,利用间隙接受理论和微分法建立了车辆汇入模型,用以计算高速公路上游过渡区的长度;最后通过算例揭示了上游过渡区长度、主车道交通量和汇入速度对车辆汇入概率的影响规律.研究结果表明:车辆汇入模型能较好地反映不同公路几何要素和交通条件情况下行驶车辆的汇入行为,为施工区设计提供一种新方法.     

基于ROI-KNN卷积神经网络的面部表情识别

深度神经网络已经被证明在图像、语音、文本领域具有挖掘数据深层潜在的分布式表达特征的能力. 通过在多个面部情感数据集上训练深度卷积神经网络和深度稀疏校正神经网络两种深度学习模型, 对深度神经网络在面部情感分类领域的应用作了对比评估. 进而, 引入了面部结构先验知识, 结合感兴趣区域(Region of interest, ROI)和K最近邻算法(K-nearest neighbors, KNN), 提出一种快速、简易的针对面部表情分类的深度学习训练改进方案——ROI-KNN, 该训练方案降低了由于面部表情训练数据过少而导致深度神经网络模型泛化能力不佳的问题, 提高了深度学习在面部表情分类中的鲁棒性, 同时, 显著地降低了测试错误率.     

污染源概化对一维模型纳污能力计算的影响分析

分析了污染源概化方式对水体纳污能力计算的影响。结果表明,污染源的概化方式对河段纳污能力计算值有明显影响,且此种影响随来水保证率和综合衰减系数的增加而增大。     

On contact zone models for an electrically impermeable interface crack in a piezoelectric bimaterial

An electrically impermeable interface crack between two semi-infinite piezoelectric planes under remote mechanical tension-shear and electrical loading is studied. Assuming the stresses, strains and displacements are independent on the coordinate x ₂ the expressions for the elastic displacement and potential jumps as well as for the stresses and electrical displacement along the interface via a sectionally holomorphic vector function are found. Introducing an artificial contact zone at the right crack tip and assuming the materials possess the symmetry class 6 mm the problem is reduced for a wide range of bimaterial compounds to a combination of combined Dirichlet–Riemann and Hilbert boundary value problems which are solved analytically. From these solutions clear analytical expressions for characteristic mechanical and electrical parameters are derived. As particular cases of the above mentioned solution the classical (oscillating) and contact zone solutions are obtained. Further, a comparison with an associated solution for an electrically permeable crack has been performed. The fracture mechanical parameters for all models via the remote loads are found analytically and important relationships between these parameters are obtained. Due to these relationships an important algorithm of a numerical method applicable for the investigation of an interface crack in a finite sized piezoelectric bimaterial is suggested.     

Theoretical estimation of loosening pressure on tunnels in soft rocks

This paper presents a theoretical method predicting the development and progress of plastic zones around tunnels in soft rocks and determining the loosening pressure which has to be taken by rock bolts or linings. Influence effects of the mechanical properties of soft rocks on the loosening pressure are examined and discussed. The judgment method used to assess the extent of rock failure is presented according to the ground characteristic curve. The ground characteristic curve shows a lowest point that refers to the minimum support pressure and the controlled tunnel wall strain when the competency factor (i.e. ratio of compressive strength to the initial in-situ pressure) is smaller than a critical value, which can be estimated from the proposed theoretical equations.     

并行学习神经网络集成方法

该文分析了神经网络集成中成员神经网络的泛化误差、成员神经网络之间的差异度对神经网络集成泛化误差的影响,提出了一种并行学习神经网络集成方法;对参与集成的成员神经网络,给出了一种并行训练方法,不仅满足了成员网络本身的精度要求,还满足了它与其余成员网络的差异性要求;另外,给出了一种并行确定集成成员神经网络权重方法．实验结果表明,使用该文的成员神经网络训练方法、成员神经网络集成方法能够构建有效的神经网络集成系统．