With the purpose of improving the insufficient ant colony optimization algorithm, including easiness to sink into local optimization and low convergence speed, an improved algorithm based on dichotomy and controlling the amount of pheromone was proposed. The dichotomy was introduced into the iterative process, and the ant pheromone on the inferior path in the last half of each iteration was discarded to accelerate the convergence speed of the algorithm. The 3-opt local optimization was used to improve the accuracy of the solution. The path selection probability of ants was dynamically adjusted by controlling pheromone quantity to avoid premature algorithm. The improved method was applied to the traveling salesman problem. Experimental results show that the proposed method has significant superiorities in optimization ability, reliability, convergence speed and stability.

针对传统模糊C均值聚类算法(FCM)过度依赖初始中心且易陷入局部最优等问题,提出一种基于上下文人工蜂群的模糊C均值聚类算法(CABCFCM).首先,引入人工蜂群算法,用来确定FCM算法的初始聚类中心;其次,采用邻域半径和高斯扰动提升人工蜂群算法的局部搜索能力,并引入上下文多臂赌博机提升算法开发能力;最后,将CABCFCM应用到广告分发业务的推荐模型上.仿真实验结果表明,其准确率明显更高、聚类效果更佳.

食品安全事关人民的身体健康,如何建立可靠的食品流通体系,实现从生产源头到销售终端全程追溯,保障食品质量和安全,具有十分重要的意义.该文利用条形码、QR二维码和RFID等多种物联网技术,对食品生产、加工、贮存、物流运输和销售等环节进行信息化处理,同时结合区块链技术实现了一个去中心化、防篡改的食品溯源系统.该系统包含Web后台管理系统和前端微信小程序,较好地满足了人们对食品安全可溯性的需求.

针对传统K-means算法初始化对参数k与初始中心选取敏感问题,提出一种融合Pearson相似度与最小生成树的K-means算法.首先,采用Hermite插值法对数据集进行均差计算,以此确定参数k;其次,计算数据对象间的Pearson相似度并将其作为无向图边的权值,由最小生成树进一步得到k个初始中心;最后,将本文算法同传统K-means和MST-K算法比较.仿真实验结果表明,本文算法在k与初始中心选取的准确性、聚类效果稳定性方面均表现出明显优势.

如何提高学生的学习自主性是翻转课堂教学过程中需要解决的关键问题。本文以《数据库原理》课程为例，分别从课前准备、课中教学和课后推送三个教学环节，介绍了如何通过案例教学、分组教学、讨论式教学、启发式教学和项目驱动式教学多种教学方法，并结合其它辅助教学手段以提高学生的学习自主性。教学实践表明，这种多教学方法融合的方式能够激发学生的学习兴趣，提高了学习自主性，有效提升了翻转课堂的教学效果。

针对粒子群优化算法容易陷入局部最优、收敛精度不高以及收敛速度较慢的问题,本文提出一种混合分层自主学习量子粒子群优化算法HHQPSO.首先,根据粒子适应度值和迭代次数将种群动态划分为三个阶层:上、下两层粒子分布较少,分别采用局部学习模型和全局学习模型,以增强粒子多样性;中层粒子分布较多,采用混合自适应量子学习模型.其次,在混合量子模型中提出改进差分策略以更新粒子的随机位置,并引入Levy飞行策略以提高算法的收敛精度和收敛速度.最后,分别在9个典型测试函数上对6种改进粒子群算法进行仿真对比实验.实验结果表明,HHQPSO算法在收敛精度、速度和稳定性上均有着较为明显的优势,特别适用于多峰函数寻优.

针对粒子群算法易于陷入早熟、收敛速度慢及收敛精度低的问题,提出了加权变异的WVPSO(WeightedVariation Particle Swarm Optimization)粒子群算法.根据自适应惯性权重和自适应学习因子,平衡了全局搜索和局部搜索能力;基于算术交叉的变异和自然选择机制的替换策略,增加了粒子的多样性,提高了算法的收敛精度;最后加入高斯扰动,使粒子产生震荡,更容易跳出局部最优.仿真实验表明,相比多个具有代表性的群智能进化算法,WVPSO算法在求解精度和收敛速度上效果更佳,并且在高维函数优化问题上具有更好的精度和稳定性.

In t for equipment. Graphic Abstract

粒子群算法的全局勘探能力和局部开发能力的不协调往往会导致算法收敛不精、陷入早熟.针对上述问题,提出了基于动态种群的双重学习粒子群优化算法(DP-DLPSO).首先将粒子种群划分为勘探子种群和开发子种群,让两个子种群以不同的学习机制引导粒子运动,粒子处于寻优期间时,子群体间无信息交流,如果开发种群陷入局部最优也不会影响到勘探种群的寻优能力;其次,采用动态种群思想,勘探种群将会把好的粒子输送到开发种群中去,增加开发种群中有效解的几率,提高粒子的收敛精度;最后,对开发种群采用高斯扰动策略以提高粒子跳出局部最优的能力.将DP-DLPSO与5个改进粒子群算法进行比较,实验结果表明DP-DLPSO在收敛精度和收敛速度上具有更大的优势.

翻转课堂作为一种先进的教学理念,备受教育界人士的广泛关注.本文立足应用型本科院校的教学现状,首先阐述了翻转课堂教学的必要性,其次分析了翻转课堂教学在实施过程中容易出现的几点误区,最后指出了提升翻转课堂教学质量的关键,并探讨了几点思路.

High‐mobility, high‐stability InGaZnO (IGZO) thin‐film transistors (TFTs) were fabricated with the aid of phenyltriethoxysilane (PTES) self‐assembled monolayers (SAMs) instead of traditional passivation layer. The effect of PTES on the performance IGZO‐TFTs was investigated, systematically. Compared to the IGZO‐TFTs without PTES modification, PTES‐treated IGZO‐TFTs exhibited higher mobility and smaller hysteresis of transfer curves, owing to less adsorption/desorption effect on the IGZO surface and mild self‐assembly process. Meanwhile, IGZO‐TFTs modified with PTES SAMs exhibited a more excellent electrical stability with a threshold voltage shift (ΔVth) of only 0.17 V than the unmodified one with a ΔVth of 5.38 V, which was attributed to the formation of hydrophobic PTES SAMs on the IGZO surface and the well‐ordered interface between PTES and back channel surface. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

DBSCAN聚类算法因其能自动识别不同数量的簇而被广泛使用.然而,由于算法中的两个参数设定需要人工进行干预,因此在很大程度上容易导致聚类效果出现偏差.针对上述问题,本文首先提出了一种基于数据集统计特征的数据矩阵;在此基础上,利用埃尔米特插值法对数据集进行差商分析与计算,使得参数自动确定;最后进行了仿真实验,结果表明本文提出的HI-DBSCAN聚类算法较原始算法在准确性和自适应性方面更佳.

随着云服务的迅猛发展,传统集中式注册的云发现技术容易出现单点失效和性能瓶颈问题.因此,提出了一种基于语义的分布式云服务发现方法.该方法采用多层云本体的方式表示云服务,把服务描述映射到云本体结构中,服务描述中的对等点根据它们的语义相似度组成集群.同时,在具有高相似度的节点之间建立语义边,从而使查询只需在相关的节点中进行,以提高发现效率.实验结果表明:本文提出的方法能够提高云服务发现效率和降低查询时间.

The Graham algorithm is used to establish a convex polygon model for the environment boundary and obstacles to simplify the environment model and improve the efficiency of the algorithm. According to the MAKLINK graph,the model of marine environ-ment is established. Then the initial path is plotted on the established environment model by using the Dijkstra algorithm. Secondly,af-ter completing path planning by use of linear decreasing inertia weight Particle Swarm Optimization,the route is smoothed by deleting redundant points to make it more practical. Aimed at the problem of linear decreasing inertia weight Particle Swarm Optimization,an improved method is proposed. By introducing chaos theory,the population is initialized by chaotic sequence to ensure the quality of the initial particles and the uniformity of distribution. The multi-population mechanism is established to balance the ability of global search and local search,which speeds up the convergence of the algorithm and improves its accuracy. The results of experiments show that the improved algorithm has better performance,especially in the complex environment,which can complete the global path planning quick-ly and get better result.

分析了目前GIS实验教学存在的弊端,探讨了工程驱动下的GIS实验教学模式设计与实践.根据南昌工程学院的"水利特色",将基础性实验、启发性实验、设计性实验、工程性实验紧密联系,以为我校水文与水资源工程专业《地理信息系统》实验课为例,设计了其实验课.

降雨是引发洪水灾害、山体崩塌、滑坡和泥石流的主要因素,完善水雨情监测系统越来越重要,作为水雨情监测系统中的雨量测量技术也同样受到广泛关注.对水雨情监测系统中典型雨量测量技术的总结和比较,有利于开发者有效地选择测量技术,也有利于研究者对测量技术的改进提高.本文介绍了几种应用广泛的雨量测量技术,并分析了各自的特性,以便于开发者和研究者对雨量测量技术的选择与研究.

在地理信息系统教学的上机实验中,通过分析区域化变量理论,选取塔里木河流域日平均降水的Hurst指数H1与其他属性为实验数据,以ArcGIS 10.2软件平台为基础,利用地统计学内插方法——联合克利金法(CoKriging)对指数H1进行空间插值.结果表明,CoKriging可以精确地展现流域降水变化的长记忆性空间分布规律.

Information-Centric Vehicular Ad Hoc Network (IC-VANET) is a promising network architecture for the future intelligent transport system. Video streaming applications over IC-VANET not only enrich infotainment services, but also provide the drivers and pedestrians real-time visual information to make proper decisions. However, due to the characteristics of wireless link and frequent change of the network topology, the packet loss seriously affects the quality of video streaming applications. In this paper, we propose a REceiver-Driven loss reCOvery Mechanism (REDCOM) to enhance video dissemination over IC-VANET. A Markov chain based estimation model is introduced to capture the real-time network condition. Based on the estimation result, the proposed REDCOM recovers the lost packets by requesting additional forward error correction packets. The REDCOM follows the receiver-driven model of IC-VANET and does not require the infrastructure support to efficiently overcome packet losses. Experimental results demonstrate that the proposed REDCOM improves video quality under various network conditions.

The traditional approach to information sharing focuses on attaching attributes and policies to an object as it is disseminated from producers to consumers. In contrast, group-centric sharing brings subjects and objects together to facilitate sharing in a group. In such contexts, subject and object join a group without constraints and authorisation is just influenced by the temporal ordering of subject and object in the group. We believe that a subject must get some privileges while joining a group and authorisation is also influenced by the trust level of subject and the security level of object. In this paper, a novel model for Group-Centric Secure Information Sharing g-SIS is proposed, which considers both the temporal ordering of group membership and the trust level of subject and security level of object. We also present its temporal logic specification. At last, a usage scenario is given to illustrate the practical application in subscription system.

Cloud computing offers a wide range of luxuries, such as high performance, rapid elasticity, on-demand self-service, and low cost. However, data security continues to be a significant impediment in the promotion and popularization of cloud computing. To address the problem of data leakage caused by unreliable service providers and external cyber attacks, an enhanced erasure code-based security mechanism is proposed and elaborated in terms of four aspects: data encoding, data transmission, data placement, and data reconstruction, which ensure data security throughout the whole traversing into cloud storage. Based on the mechanism, we implement a secure cloud storage system (SCSS). The key design issues, including data division, construction of generator matrix, data encoding, fragment naming, and data decoding, are also described in detail. Finally, we conduct an analysis of data availability and security and performance evaluation. Experimental results and analysis demonstrate that SCSS achieves high availability, strong security, and excellent performance.