为提高降雨预测的准确率,解决现有深度学习降雨预测模型缺乏对多站点气象数据时空关系建模能力的问题,提出基于多任务学习和改进的长短时卷积计算网络降雨预测模型(MTL-LSTC).在长短时记忆网络内部的输入到状态和状态到状态的转换过程加入卷积运算结构,对气象序列数据进行编码,结合多任务学习方法提取出站点间隐藏的交互信息,建立站点间相关性模型,实现基于多站点气象数据的降雨预测.模拟实验结果表明,MTL-LSTC模型预测结果准确率更高且模型更高效,多站点气象数据的利用率也得到较大提升.

To reduce the difficulty in extracting features of an occluded face, a dual-channel Convolutional Neural Network (CNN) model with occlusion perception is proposed.The model is constructed by integrating newly designed occlusiondecision units into VGG16 network, which aims at extractingexpression-related features of the areas that are less occluded.The model employs the transfer learning algorithm to pre-train the parameters of the convolutional layer, which means to alleviate the over-fittingproblem.At the meantime, the expression-related features of the whole facial image are extracted by the modified residual network.Finally, the outputs of theperceptive neural network and residual network arefused in a weighted manner.The experimental results show that the proposed model achieves an accuracy of 97.33% on CK+, 86% on RAF-DB, and 61.06%on SFEW.Compared with traditional OPCNN, ResNet, and VGG16 models, the proposed model exhibits a significant improvement in the accuracy of recognizing the expression of an occluded face.

With the rapid development of deep neural networks and smart mobile devices, the research of lightweight neural network structure has gradually become a hotspot. The essence of lightweight design is to optimize the storage space and improve the running speed without causing any loss to the precision of deep neural networks. Then an introduction to the mainstream methods of lightweight network structure design for deep learning is given, and the innovative features, strengths and weaknesses between the manual design methods, neural network structure search-based design methods and automated model compression-based design methods are compared. The advantages and disadvantages of the high-performance network structures generated by the above methods are also summarized. On this basis, the challenges faced by lightweight network structure design, and its applications and development trends are discussed.

针对浅层神经网络面对温室复杂多变环境因子表征能力低、学习时间长的问题,提出一种基于改进深度信念网络并结合经验模态分解与门控循环单元的温室预测方法。首先,通过经验模态分解将温度环境因子进行信号分解,之后将分解出来的固有模态函数与残差信号进行不同程度的预测;然后,引入神经胶质改进深度信念网络,并将分解信号结合光照和二氧化碳进行多属性的特征提取;最后,将门控循环单元预测的信号分量相加获得最终的预测结果。仿真实验结果表明,与经验模态分解-深度信念网络(EMD-DBN)和深度信念网络-神经胶质链(DBN-g)相比,所提方法的预测误差分别降低了6.25%和5.36%,验证了其在强噪声、强耦合的温室时序环境下预测的有效性和可行性。

高校课堂教学评价是教学质量监控系统的重要组成部分,本文在分析教学质量评价体系的基础上,设计了课堂教学质量评价系统,该系统集成课堂教学视频的直播和点播、同步与异步教学评价以及结果反馈等功能,旨在促进教师的专业发展和课堂教学质量的提升.

降雨是众多气象要素相互作用的结果,气象数据具有类型复杂、不定性、产生速度快、数据量大等特点,增大了降雨预测模型处理气象数据的难度。针对降雨预测方法要求气象要素之间相互独立,但很多气象要素相互关联,降低了预测的准确性和效率的问题,为了提高降雨等级的预测准确性与效率,提出一种基于Map Reduce的改进加权朴素贝叶斯算法(WMNBC)。首先根据气象数据相应条件属性、类属性及其它所有条件属性三者之间的相关程度,提出一种信息密度的概念,并结合对应条件属性取值数目设计一个新的属性赋权方法,然后建立基于MapReduc的WMNBC降雨等级预测模型,最后对实际降雨等级进行仿真测试。仿真结果表明,相比NBC和HWNN算法,WMNBC算法可以更加准确的预测降雨等级并提高预测效率,为降雨等级预测提供了新的途径。

For the problems of the data classification with low efficiency caused by the large amount of meteoro-logical data and the highly complex of heterogeneous data,a K-nearest neighbor combined classifier and a distributed parallel processing method were adopted to obtain the agro-meteorological disaster classification model.Firstly,according to the grade index of meteorological disaster,an exponential formula of the agro-meteorological disaster was proposed.And then,a parallel K-nearest neighbor combined classifier was used to complete the statistical classification of agro-meteorological disaster index.Finally,the ranking information of meteorological disaster which has been classified was analyzed to assess the disaster risk of crops.The evaluated meteorological disaster information can guide the agricultural production for farmers reasonably and reduce property loss.The experiment simulation shows that,faced with various and huge agro-meteorological data,the parallel K-nearest neighbor combined classifier is faster and more precise.

In the study of prehistoric settlement sites,the settlement sites of different research areas are different in spatial and temporal scales,so that a lot of experts recognize that the consistency of the grading assessment standards are difficult to obtain.In this paper,K-means clustering method is used to replace the expert scoring process,which is supplemented by image analysis and SPSS data exploration percentile method.We analyze the five environmental driving factors such as topography,soil,slope aspect,slope and elevation of the prehistoric settlement sites in Zhengzhou-Luoyang area,and determine the classification criteria and grade values of each factor.The results show that the five factors form the appropriate five degrees(high,higher,medium,lower and low) based on the number of sites in the four periods and K-means clustering method.The topography and soil in order of preference correlation coefficient of three adjacent periods are gradually increasing,which means that the degree of preference of prehistoric human remains to be solidified.In the Longshan Period,it's significant to study the distribution of settlement sites with modem landform type data,which can be used to provide reference for the location of the new settlement site.Besides,prehistoric human's selection for hilly topography types will continue to decline,while for the selection of loess plateau,cinnamon soil and southeast slope has come increasing.The preference of loess terrace like plain,mixed soil and the gentler slope is still in a high level.The results have a reference value for the study of distribution of prehistoric settlement sites and the excavation of new sites.

The dissemination of internet-based information changed the previous pattern of geography, while micro-blog has formed an important platform for exchanging information. Using the method of social network analysis ( SNA) and GIS visualization, this paper tries to analyze the path of information transmission in micro-blog from the perspective of geography. Eventually the results show as fol-lows.①The spatial structure and the social economy are relatively consistent. The number of in-out degree in network is consistent with the development of the economic level, while situations of dissemination are different in certain central cities.②The space of network information has the geographical features. Those cities as center-nodes dominate the direction of public opinion in the surrounding areas, even nationwide direction.③In network, the dissemination number of public opinion is regional during the same time frame.

Aiming at the problem that the current classification algorithm has low generalization ability and insufficient precision,a combination classification model combining Adaboost algorithm and Back-Propagation (BP) neural network was proposed.Multiple neural network weak classifiers were constructed and weighted,which were linearly combined into a strong classifier.The improved Adaboost algorithm aimed to optimize the normalization factor.The sample weight update strategy was adjusted during the lifting process,to minimize the normalization factor,increasing the number of weak classifiers while reducing the error upper bound estimate was ensured,and the generalization ability and classification accuracy of the final integrated strong classifier was improved.A daily precipitation model of 6 sites in Jiangsu province was selected as the experimental data,and 7 precipitation models were established.Among the many factors influencing the rainfall,12 attributes with large correlation with precipitation were selected as the forecasting factors.The results show that the improved Adaboost-BP combination model has better performance,especially for the site 58259,and the overall classification accuracy is 81％.Among the 7 grades,the prediction accuracy of class-0 rainfall is the best,and the accuracy of other types of rainfall forecast is improved.The theoretical derivation and experimental results show that the improvement can improve the prediction accuracy.

Aiming at the problems that the purpose of the meteorological observation data acquisition is weak,the redundancy of data is high,and the number of single values in the observation data interval is large,the precision of equivalence partitioning is low,an attribute reduction algorithm for Meteorological Observation data Interval-value based on Genetic Algorithm (MOIvGA) was proposed.Firstly,by improving the similarity degree of interval value,the proposed algorithm could be suitable for both single value equivalence relation judgment and interval value similarity analysis.Secondly,the convergence of the algorithm was improved by the improved adaptive genetic algorithm.Finally,the simulation experiments show that the number of the iterations of the proposed algorithm is reduced by 22,compared with the method which operated AGAv (Adaptive Genetic Attribute reduction) algorithm to solve the optimal value.In the time interval of 1 hour precipitation classification,the average classification accuracy of the MOIvGA ()t-Reduction in Interval-valued decision table based on Dependence) algorithm is 6.3％ higher than that of RIvD algorithm;the accuracy of no rain forecasting is increased by 7.13％;at the same time,the classification accuracy can be significantly impoved by the attribute subset received by operating the MOIvGA algorithm.Therefore,the MOIvGA algorithm can increase the convergence rate and the classification accuracy in the analysis of interval value meteorological observation data.

The quantitative analysis and simulation study on the relationship between prehistoric settlement sites and natural environment has become a hot spot in the study of the relationship between human and land in the prehistoric settlement sites. This article used five indicators, landform, soil, exposure, slope and eleva-tion, to construct the fuzzy evaluation system about prehistoric human settlement exponent in Zheng-zhou-Luoyang Area, which is based on factor detection method based on geographic detector to gain the weight of every factor, according to environmental driving factors grading standard and scale value, we used trapezoid and triangle membership function to calculate the value of every factor, then we used comprehensive evaluation of the weighted average type and isometric method to gain the result of prehistoric human settle-ment exponent's comprehensive level in Zhengzhou-Luoyang Area. Comprehensive level 1 for the western re-gion of the site, the level of 2 are large number of distribution, is a Hybrid-Around the Center type, grade 3 is the northwest loess type, grade 4 is the Central Hills-Flood Alluvial Plain, grade 5 is the eastern Platform-Ter-race-Plain type. Overlay them with the actual site of later period of the Yangshao distribution and distribution map of the site of the Dragon Mountain, statistics of the sites in each level, it is found that the vast majority of concentration in the middle of the class and above, the model is feasible, at the same time, the accuracy of the site was found is higher. To carry out specific analysis on comprehensive level 5 class, according to the SPSS data detection function, the nature is divided into Ⅰ, Ⅱ, Ⅲ class, count respectively the membership degree of the five factors in the statistics of various types are calculated, and the degree of conformity and contribution of the comprehensive grade value are compared before and after the weight. It is found that the site of type I high livable degree is the highest degree of conformity with the characteristic of the slope factor of 0 degrees, and the height of the 100-200 m is the second;Type Ⅱ high degree of habitable site for elevation of 100-200 m conform to the highest degree, to high livable degree in geomorphology and soils on the coincidence degree was significantly higher than that of type I, class Ⅲ high degree livable sites representing class Ⅱ soil and to-pography of the highest degree of compliance. In the same livable degree, two geographical positions, in other conditions are the same, the position to meet the characteristics of Ⅱ type environment than type I more new sites mining potential, and meet the needs of type Ⅲ environmental features position than type Ⅱ more poten-tial. Most livable degree high or higher site distributions in the Zhengzhou-Luoyang area in northern and cen-tral regions of Yiluo river basin, is near the river interchange and along the main stream distribution.This arti-cle also respectively based on the actual site of the later period of the Yangshao and Longshan period distribu-tion data of the level to verify the accuracy of the evaluation system. The results show that in the later period of the Yangshao, site environment index above the intermediate site ratio was 87.1%;In the period of Long-shan, the site of the intermediate above proportion was 85.5%, indicated that the model of evaluation result is accurate and reliable.

校园信息门户是高校的名片,为了顺应移动互联网微时代的发展趋势,全面实现数字化校园,共享高校资源与应用服务,构建高校微门户是主要发展趋势.针对数字化校园数据量大、个性化需求程度高等特点,结合分布式云计算处理方案,构建基于云计算的高校微门户,对海量服务数据进行深度挖掘分析,并利用个性化推荐算法实现信息分类与推送.实践表明,该系统能优化高校资源配置,对各类服务对象进行精准服务.

针对三维水下传感器网络存在的节点部署稀疏、水下节点昂贵、网络部署成本高、三维环境复杂等问题,提出了一种基于网格划分和虚拟力的网络部署策略.该策略研究了三维空间多面体填充问题,将水平面划分为一定大小的网格,对水面上的节点运行虚拟力算法,使节点均匀扩散开,落在同一网格的节点通过控制浮标与节点间的缆绳长度控制节点在垂直方向的移动,形成三维水下传感器网络.仿真实验结果表明,该策略能够以更小的节点数目达到更高的三维空间网络覆盖效率,从而有效地减少网络的部署成本.

针对全球范围内气候环境的变化,研究古时代聚落的演变发展及其与环境变化的关系,可深入了解人地关系,为制定聚居决策提供参考.以GIS为研究方法,对郑洛地区新石器时代裴李岗时期至龙山时期的995处聚落遗址进行统计,发现其中单一型聚落遗址546处,叠置型聚落遗址449处.单一型聚落遗址的数量变化为增长、下降、迅速增长,聚落遗址重心也随着环境的变化而出现移动,这与该地区气候历经的4个不同的变化阶段有着密切关系.在所统计的叠置型聚落遗址中,有11处包含了4个时期的聚落遗址.在裴李岗时期至龙山时期,聚落遗址的叠置系数由小到大,随着自然环境的相对稳定,聚落的继承性得到提升.

针对无线气象传感网内由于节点数量大、感知数据冗余度高而导致节点通信耗能过高的问题,提出了数据联合稀疏预处理模型,利用监测区域气象要素预报值和各簇头要素值计算出一个全网公共分量并对网内数据进行预处理.将分布式压缩感知应用于簇型传感网中,对各节点感知数据进行压缩观测,在汇聚节点进行数据重构,从根本上降低节点通信量,均衡负载;同时设计了一个基于公共分量异常数据稀疏方法.仿真实验中,相对于单独使用压缩感知,数据联合稀疏预处理模型能够有效利用数据时空相关性提高数据稀疏度,压缩性能提高了25％,重构性能提高46％;同时,异常数据处理方案能够以96％的高概率恢复异常数据.因此,该数据预处理模型能够提高数据重构效率,有效降低网内数据通信量,延长网络寿命.

随着互联网技术的普及与发展,高校移动社交网络平台逐渐因为其移动性好、携带便捷、及时性等特点,成为各大高校构建移动社交网络校园发展的主流.该文提出了一个基于情景感知的高校移动社交网络平台的构想,其利用情景感知的推荐算法,实现信息的收集与分析,为用户提供更加全面的服务.

针对传感器节点在三维监测区域中随机分布覆盖效率低下,并且不能达到关键区域重覆盖的问题,本文使用空间填充多面体,分别从确定性覆盖和随机覆盖两个方面,提出理想状态下覆盖冗余率最低和空间密度值最低的节点分布策略.首先将监测区域分为多个以传感器节点的传感半径为外接球直径的多面体,然后将传感器节点放置在多面体的顶点或是外接球重叠区域中,最后理论分析出同构节点分布的最佳位置.实验仿真表明,在相同覆盖重数的情况下,截角八面体的覆盖冗余率和空间密度值最低.

八旗制度是清朝的一项独特制度,模糊了民族意识,使政权忽视了对本民族文化的保护.由于蒙、汉等民族在人数上的优势,加之八旗子弟长期杂居的生活状态使满族的文化,尤其是语言和文字最终湮没.

针对密度分布不均的雷电定位资料,提出了一种基于OPTICS聚类算法的雷电临近预警模型。该模型运用OPTICS算法对雷暴天气连续时段的雷电定位资料进行聚类分析,有效剔除了影响雷暴云分布的稀疏点。在聚类分析结果基础上,利用"膨胀-侵蚀"算法还原雷暴云真实分布,根据雷暴云的移动趋势进行雷电落区预报。此外,针对传统预测算法运行时间长的缺陷,运用邻接表改进了OPTICS算法,且优化了可达队列更新策略。实验结果表明,基于改进的OPTICS算法所构建的雷电临近预报模型降低了算法运行时间,同时提高了雷电预报模型适应能力及预测的准确率。