基于卷积神经网络的高分遥感影像单木树种分类

树种调查一直面临着成本高、效率低、精度不高等问题.利用遥感手段能大大提高树种类型调查的工作效率、节省成本;卷积神经网络(CNN)虽然已经在自然图像分类领域取得了许多突破,但是较少有人将CNN模型用于单木树种分类.基于上述考虑,搭建出CNN模型,并与高分遥感影像相结合,进行单木树种分类.在利用高分影像半自动化构建单木树种...     

Variable length signal detection framework for cognitive radio systems

The cognitive radio (CR) system opportunistically utilizes the frequency bands temporally unoccupied by the primary user. In the CR system, the energy or cyclostationary detector is used to detect the primary user signal. For protecting the primary user tightly, the signal detection time can be very long, which leads to inefficiency in the CR system. Thus, we propose a novel signal detector that greatly reduces the average detection time. The proposed detector periodically decides whether it terminates the detection process or receives more input signal for more information. Therefore, the proposed detector has variable detection time. We will call the proposed detector the variable length signal detector (VLSD). The VLSD is designed by using a partially observable Markov decision process framework for optimal performance. We present the numerical results showing that the VLSD requires much smaller average detection time compared with the traditional fixed length signal detector to achieve a given detection error probability. Copyright © 2012 John Wiley & Sons, Ltd.     

机器学习理论在入侵检测技术中的应用研究

入侵检测系统是主动保障网络信息安全的重要方法。本文针对大规模、高带宽网络环境下,入侵检测技术存在的不足,提出将机器学习理论应用到入侵检测系统中。文章简要介绍几种适合用于入侵检测系统中的机器学习算法,并建立基于机器学习理论的入侵检测系统框架。利用机器学习的算法不仅能检测到一些已知的攻击,还可以通过自我学习检测到未知的攻击。     

煤矿危险源辨识与风险管理信息系统开发与探究

信息处理的滞后和不全面严重威胁着南梁煤矿安全预防工作的顺利进行。通过对危险源进行辨识,通过危险源等级的划分,完成基于风险矩阵的风险评估方法,有效的对危险源的风险有效的进行了管理,为危险源评估提供了有效的评估方式,对危险源进行分级分类,监测预警和控制,建立危险源辨识动态管理系统,在综合分析近年来安全管理信息化系统研究成果的基础上,进行煤矿危险源辨识及风险管理信息系统研究。     

延时控制差异并联算法与自适应加权融合算法

为提高延时控制系统现有决策算法的精度,提出了基于遗传算法的差异并联决策算法,利用遗传算法对各个延时电路的决策增量进行优化求解;为改善系统决策算法的鲁棒性,提出一种自适应加权融合算法,在对实测数据进行分析的基础上确定合理的权值分配策略。利用数值算例对精度、可靠度、鲁棒性和实时性的对比表明:与原有算法相比,所提两种算法具有更优的综合性能,其中自适应加权融合算法性能较为均衡,而差异并联算法具有较强的抑制异常测量值能力,适用于对鲁棒性要求较高的场合。     

一种自适应的无线射频识别系统中的防冲突算法(英文)

A novel anti-collision algorithm in RFID wireless network is proposed.As it is put forward on the basis of collision tree(CT)and improved collision tree(lCT) anti-collision protocols,we call it adaptive collision tree protocol(ACT).The main novelty of this paper is that the AD strategy is introduced and used in ACT to decrease collisions and improve the tag system throughput.AD strategy means that query strings will divide into two or four branches adaptively according to the label quantity.This scheme can decrease both depth of query and collision timeslots,and avoid producing too much idle timeslots at the same time.Both theoretical analysis and simulation results indicate that the novel proposed anticollision protocol ACT outperforms the previous CT and ICT protocols in term of time complexity,system throughput,and communication complexity.     

面源红外干扰弹质心干扰反舰导弹的决策方法

面源红外诱饵已经发展成为对抗红外成像制导导弹的一种有效手段。以面源红外干扰弹质心干扰反舰导弹为例,设计了相关的仿真模型,包括导弹模型、舰艇模型、面源干扰弹模型、干扰弹发射决策模型和发射机动决策对干扰效果影响的评估模型等,并依据质心干扰仿真计算流程,对干扰弹发射和舰艇机动决策方案进行了寻优仿真,提出了实施面源红外干扰时优选的干扰弹发射和舰艇规避机动决策原则与方法。     

基于模糊综合评价法的城市人居环境评价体系研究

为了对城市人居环境质量做出科学、客观的评价,提出一种改进的模糊综合评价模型,将模糊数学理论应用于环境质量的评价中,结合城市人居环境的特点建立多目标指标体系,采用层次分析法与群体决策相结合来确定指标权重,应用贴近度法确定评价等级。并且结合具体的实例对模型进行了应用,获得了较准确、客观的评价结果,为促进城市人居环境质量的改善提供可靠的决策支持。     

A Distributed Multicast Routing Scheme for Multi-Layered Satellite IP Networks

In this paper, a distributed multicast routing scheme is introduced for multi-layered satellite IP networks, which include GEO, MEO, and LEO layers. This scheme aims to minimize the total cost of multicast trees in the satellite network. Multicast trees are constructed and maintained in the dynamic satellite network topology in a distributed manner. Simulation results are provided to evaluate the performance of the new scheme in terms of end-to-end delay and multicast tree cost.     

Hybrid verification integrating HOL theorem proving with MDG model checking

In this paper, we describe a hybrid tool for hardware formal verification that links the HOL (higher-order logic) theorem prover and the MDG (multiway decision graphs) model checker. Our tool supports abstract datatypes and uninterpreted function symbols available in MDG, allowing the verification of high-level specifications. The hybrid tool, HOL-MDG, is based on an embedding in HOL of the grammar of the hardware modeling language, MDG-HDL, as well as an embedding of the first-order temporal logic L_mdg used to express properties for the MDG model checker. Verification with the hybrid tool is faster and more tractable than using either tools separately. We hence obtain the advantages of both verification paradigms.