焊接过程的RBF神经网络建模

介绍了RBF神经网络的性能和算法结构 ,建立了RBF神经网络在船舶焊接过程中用于焊接变形预测分析的模型 ,并探讨了其应用和发展趋势     

基于学习的证券市场专家预测意见合成研究

针对证券市场的专家预测意见的特点，研究了将学习、证据理论与协同学理论引入证券市场专家预测意见合成的理论与方法．讨论了将专家的历史预测数据用基本可信数表示作为学习样本，对初选专家进行聚类分析选择有协同效应的专家．该方法用神经网络优化基本可信数的修正系数，再用Dempster合成规则进行预测意见的合成．实验结果表明该方法应用于解决实际问题时具有较好的效果．     

广义正定矩阵的分析与应用

分析了广义正定矩阵,导出了矩阵广义正定的若干新判定准则,并以应用实例研究了其在神经网络中的应用.     

光电转换体系太阳能聚集器

In standard P-V systems, the high cost and low quantum efficiency of silicon cells has held back widespread adoption of this technology for decades. To generate electrical power from solar energy in a cost effective way, most methods require the concentration of the sunlight in order to increase the efficiency of energy conversion. Most systems use parabolic mirrors to focus sunlight to either a line （trough systems） or a small volume （dishes）. One of the reasons solar power still remains little utilised, is the high cost of the collection systems. Light focussing devices using reflection usually take the form of a curved rein＇or. In contrast, this paper describes the new design of a device consisting of an array of small angled reflecting m/trot facets located in a planar form. The three dimensional angle of each facet in the array is a function of its position in the array, and is calculated to be such that for a parallel beam striking the array, each facet will reflect the light in such a way as to form a focal point region.     

识别混凝土板内部缺陷的人工神经网络算法

以一维有缺陷混凝土板为研究对象，分别采用Leverberg-Marcluardt和径向基神经网络算法，对缺陷的深度与厚度进行识别，从而实现对混凝土板内部缺陷的三维重构，称为红外CT模拟．两类神经网络算法的识别结果表明：Leverberg-Marcluardt神经网络较径向基神经网络具有更好的收敛精度与计算效率．     

Therapeutic applications of bone marrow-derived stem cells in liver transplantation for end-stage liver diseases

Today, liver transplantation (LT) is the only established treatment for end-stage liver diseases. The de- velopment of LT, including OLT, cadaveric LT, split LT, living donor LT (LDLT), brings hopes to patients with these diseases. However, increasing donor shortage, rejection and life-long immunosuppression with its side effects are the major limitations of this therapy strategy. Bone marrow-derived stem cells (BMDSCs) are capable of differentiating into hepatocyte-like cells and contribute to liver injury repair. The microenvironment of liver injury caused by rejection, ischemia/reperfusion, loss of liver mass, recurrence of HCV and "small-for-size syndrome" after LT can attract a variety of bone marrow-derived stem cell population to the peripheral circulation and then migration to the injury liver to promote the hepatic function restoration. Additionally, BMDSCs can also take part in the functional regeneration of living donor liver after LDLT. This participation in liver regeneration may be associated to the interac- tion between SDF-1and its receptor CXCR4, involving HGF, IL-8, MMP9, and VEGF/VEGFR-2. BMDSC with its bio-characteristics could maintain the allograft tolerance from different angles and in different ways. In conclusion, BMDSCs transplantation, as a new assistant therapeutic method for LT, will ex- pand the space of LT, and provide more survival opportunities for the patients suffering liver diseases in the future.     

螺旋霉素发酵过程动态建模的神经网络方法

对文题所述方法,根据实验数据进行仿真研究,获得了满意的结果。为高度复杂的非线性生化过程的模型化提供了一条新的途径。     

Intelligent Control Mechanism of Part Picking Operations of Automated Warehouse

This paper studies the part picking operations of a ut omated warehouse. It assumed the demand of picking orders of automated warehouse are dynamic generated. Once the picking orders of certain period of time are kn own, it is necessary to decide an efficient order picking sequence and routing t o minimize the total travel distance to complete those orders. Assumed there are n i items to be picked in order O i. Each item in the picking ord er is located in different locations in the warehouse. Since i...     

神经网络在可靠性分析中的应用

应用人工神经网络对舰炮发生故障的发射炮弹数进行了预测。借助前馈神经网络对非线性函数的逼近能力，对发生故障的发射炮弹数进行了单步预测，并与使用灰色数列预测的结果进行了对比，验证了人工神经网络进行可靠性预测的可行性。