一种无先验地图的移动机器人导航方法

为了提高移动机器人在复杂环境下的无先验地图导航算法模型训练速度及导航成功率,提出一种基于深度确定性策略梯度(deep deterministic policy gradient, DDPG)的移动机器人导航方法。利用2D激光雷达的均匀分布测距信息,降低环境噪声的干扰及高维度环境信息的计算量;采用人工势场法构建移动机器人从初始位置到目标点过程的奖励函数;通过Actor-Critic网络结构提高模型训练的稳定性和泛化能力。实验结果表明,提出的方法具有模型训练速度快、导航成功率高及泛化能力强等优点。     

硝酸盐氮产污规律研究

利用大型可变坡土槽和室内人工降雨,模拟没有任何作物的裸土坡地在不同雨强下非点源硝酸盐氮产污过程.通过多次全过程的降雨量、径流、水质的同步观测,分析污染物浓度随降雨过程的动态变化规律.在物理机理研究的基础上提出降雨径流污染物浓度计算模型.     

一种基于免疫的敏捷供应链形成算法

敏捷供应链形成是敏捷制造的首要问题,主要研究如何在众多候选企业中选择合作伙伴,形成最优供应链。该文通过引入多目标约束条件,优化了问题的数学模型,并设计了一种基于免疫的敏捷供应链形成算法,最后给出一个典型算例,以说明该算法的有效性和实用性。     

Vaults and the major vault protein: Novel roles in signal pathway regulation and immunity

The unique and evolutionary highly conserved major vault protein (MVP) is the main component of ubiquitous, large cellular ribonucleoparticles termed vaults. The 100 kDa MVP represents more than 70% of the vault mass which contains two additional proteins, the vault poly (ADP-ribose) polymerase (vPARP) and the telomerase-associated protein 1 (TEP1), as well as several short untranslated RNAs (vRNA). Vaults are almost ubiquitously expressed and, besides chemotherapy resistance, have been implicated in the regulation of several cellular processes including transport mechanisms, signal transmissions and immune responses. Despite a growing amount of data from diverse species and systems, the definition of precise vault functions is still highly complex and challenging. Here we review the current knowledge on MVP and vaults with focus on regulatory functions in intracellular signal transduction and immune defence. Received 27 June 2008; received after revision 25 July 2008; accepted 30 July 2008     

Antimicrobial skin peptides and proteins

Human skin is permanently exposed to microorganisms, but rarely infected. One reason for this natural resistance might be the existence of a ‘chemical barrier’ consisting in constitutively and inducibly produced antimicrobial peptides and proteins (AMPs). Many of these AMPs can be induced in vitro by proinflammatory cytokines or bacteria. Apart from being expressed in vivo in inflammatory lesions, some AMPs are also focally expressed in skin in the absence of inflammation. This suggests that non-inflammatory stimuli of endogenous and/or exogenous origin can also stimulate AMP synthesis without inflammation. Such mediators might be ideal ‘immune stimulants’ to induce only the innate antimicrobial skin effector molecules without causing inflammation. Received 9 August 2005; received after revision 21 October 2005; accepted 16 November 2005     

Olfactory ensheathing cells are attracted to, and can endocytose, bacteria

Olfactory ensheathing cells (OECs) have been shown previously to express Toll-like receptors and to respond to bacteria by translocating nuclear factor-kappaB from the cytoplasm to the nucleus. In this study, we show that OECs extended significantly more pseudopodia when they were exposed to Escherichia coli than in the absence of bacteria (p=0.019). Co-immunoprecipitation showed that E. coli binding to OECs was mediated by Toll-like receptor 4. Lyso-Tracker, a fluorescent probe that accumulates selectively in lysosomes, and staining for type 1 lysosome-associated membrane proteins demonstrated that endocytosed FITC-conjugated E. coli were translocated to lysosomes. They appeared to be subsequently broken down, as shown by transmission electron microscopy. No obvious adherence to the membrane and less phagocytosis was observed when OECs were incubated with inert fluorescent microspheres. The ability of OECs to endocytose bacteria supports the notion that OECs play an innate immune function by protecting olfactory tissues from bacterial infection.     

基于神经网络的综采工作面技术经济指标预测

应用人工神经元网络系统理论 ,基于神经网络的自学习方法 ,在对西山煤电集团东曲矿综采工作面大量实际资料进行统计分析的基础上 ,就工作面日进度、日产量、回采工效率、坑木消耗、配件消耗等 1 1项综合技术经济指标进行预测 ,预测结果精确度较高与实际相吻合 .为解决煤矿综采工作面计划、生产、管理的预测决策问题提供了一种新的研究方法.     

基于神经网络的Lyapunov指数谱的计算

利用 BP神经网络的非线性函数逼近能力 ,由实验观察数据列计算系统的 Lyapunov指数谱 .实例计算表明 ,此种方法精度较高且计算量较小 ,有重要的实际意义.     

基于Rogowski线圈的新型甲烷传感器抗干扰性研究

在对传统的甲烷传感器进行应用与分析基础上,介绍了一种基于Rogowski线圈新型甲烷传感器。经过理论分析建立了该传感器数学模型,在现场实际应用中针对新型甲烷传感器干扰信号来源,对新型甲烷传感器抗干扰能力进行了理论分析以及实际信号处理设计。     

紫外分光光度法-人工神经网络在4种嘌呤同时测定中的应用

将紫外分光光度法与人工神经网络相结合, 在波长200～300　nm不经分离同时测定腺嘌呤、 黄嘌呤、 次黄嘌呤和鸟嘌呤的混合物. 讨论了pH值对体系的影响, 并对网络参数进行选定. 方法简便、 快速, 相对标准偏差小于3.7%.