缺苗断垄对机播夏玉米产量及产量构成因素的影响

为探讨缺苗断垄对机械化播种夏玉米的影响,在大田状态下设置2个品种不同缺苗断垄程度进行试验,并对缺苗级别、断垄级别、籽粒产量及其产量构成因素进行统计分析。结果表明,缺苗断垄引起产量降低的主要原因是株数和穗数减少,随着缺苗断垄级别的增加,株数穗数逐渐减少,产量逐渐降低。产量降低幅度小于缺苗断垄级别的增加幅度,在缺苗断垄率为10.00%～40.00%的情况下,产量降低幅度为3.60%～28.51%;在缺苗断垄率相同的情况下,鲁单981比郑单958的减产幅度大些。     

一个DNA计算的图表示问题

DAN计算是新兴的一种智能计算方法,2维平面中建立合适的图表示方法对研究DNA序列有重要作用。对他人证明的一种图表示方法的退化度进行了纠正,并给出了反例说明其真实退化度。     

作物模型资源构造平台(CMRCP)的构建研究

为有效促进作物系统数字化研究的深入发展,将集成制造的工业化思想引入作物模型领域,提出了作物模型资源(CMR)及作物模型资源构造平台(CMRCP)的概念.在提炼作物生长通用概念模型的基础上,将作物模型资源分解为通用模型结构、模型算法表示和模型资源映射三部分;并通过模型概念与/或树、模型算法与/或图与模型实现形态相结合,提出了综合性模型资源表示方法.另外,借鉴自适应软件开发方法,基于组件框架、组件组装、模型解析等技术,提出了平台构造原理,表明模型构建任务应该由作物元模型、建模学习机制、模型构造引擎和建模支撑工具共同协作完成.结果表明,构建具有学习能力、自适应和组件化特征的作物模型资源构造平台,可促进作物建模知识资源和软件开发资源的相互转化,并可积累易于共享与复用的作物模型资源库,降低作物模型构建与开发难度,提高农业模型软件开发效率和质量,为作物智能建模提供构建平台.     

基于K-SVD和正交匹配追踪稀疏表示的稻飞虱图像分类方法

针对当前稻飞虱图像分类研究中存在图像识别速度慢、分类精度低的不足,该文提出一种基于K-SVD和正交匹配追踪(orthogonal matching pursuit, OMP)稀疏表示的稻飞虱图像分类方法。首先,根据稻飞虱的趋光性特点,使用团队自主研发的野外昆虫图像采集装置自动获取稻田害虫图像;然后,利用K-SVD算法对稻飞虱图像特征的过完备字典进行更新构造,结合OMP算法对原始输入图像的特征信号进行稀疏表示;最后,通过求解输入图像的重构误差对昆虫图像进行分类。在相同的试验条件下,与传统的图像分类算法(SVM、BP神经网络)进行比较。实验结果表明,该文提出的基于K-SVD和OMP算法的稻飞虱图像稀疏表示分类方法可对稻飞虱与非稻飞虱进行快速准确的分类,分类速度达到6.0帧/s,平均分类精度达到93.7%。与SVM和BP神经网络相比,分类速度分别提高了5和5.5帧/s;分类精度分别提高了15.7和28.2个百分点,为稻飞虱的防治预警工作提供了信息与技术支持。     

Using patterns of reef fish assemblages to refine a Habitat Classification System for marine parks in NSW,Australia

• 1. The Solitary Islands Marine Park (SIMP) in New South Wales (NSW), Australia, is located in a tropical‐temperate biotone, with variable influence of the tropical East Australian Current (EAC) from inshore to offshore. Marine communities on the extensive, subtidal, rocky reefs are poorly described. As a result, the current Habitat Classification System (HCS) used as part of the process to determine the marine park zoning arrangements, which places all shallow reef (<25 m) in the same category, is unlikely to represent real biotic patterns.
• 2. To evaluate the influence of five key factors (distance from shore, reef type, dominant benthos, latitude, and depth range) fish assemblages were widely surveyed, using 30‐minute timed counts, at 68 sites across the extent of shallow reef within the park. Relationships between assemblage patterns and levels of the different factors were subsequently examined using multivariate analyses.
• 3. Patterns of reef fish assemblages were most strongly correlated with distance from shore. Three distinct assemblages occurred on inshore (<1.5 km), mid‐shelf (1.5–6 km), and offshore (>6 km) reefs. Differences in assemblage structure by reef type were also apparent on inshore and offshore reefs, but not on mid‐shelf reefs. Correlations with the other factors were weak. The cross‐shelf pattern was persistent over the scale of years.
• 4. The results of the study provide strong support for incorporating distance‐from‐shore categories into a refined HCS. This will improve its ability to represent biological diversity as reflected by patterns of reef fish assemblages. Further research is required to determine the wider application of the HCS to other marine parks in NSW and to determine if it also, effectively, represents other components of biodiversity. Copyright © 2009 John Wiley & Sons, Ltd.

     

Perceptions and representations of biodiversity in Lausanne,Switzerland: Acknowledging the importance of residential environments

While the ecological mechanisms involved in biodiversity loss within urban environments are widely studied, the literature often overlooks its social-ecological drivers. One of these drivers is the relationship between perceptions and representations of ecosystems and species within specific residential contexts. This study therefore aims to address the following questions: how do residents perceive biodiversity in relation to their social representation of nature within specific residential environments? Are there differences in the social representation and perception of greenspace types and species based on residents' demographic attributes (gender and age) and characteristics of their place of residence (district and housing types)? Through a quantitative survey and the analysis of 148 responses from Lausanne residents using textometry, cartography, and inferential statistics, this research demonstrates two main findings. Firstly, it confirms that the perception of species remains limited, which aligns with previous scholarly work. Secondly, it reveals that the social representations of nature and biodiversity can either correlate with or be disconnected from the perceptual process at the ecosystem level, depending on the residents' living environments. The impact of residential environments on biodiversity perception and representations is discussed, along with the implications of iconic biodiversity traps for raising public awareness about biodiversity preservation in cities.     

基于切换字典的林区小气候监测数据压缩感知方法

为降低林区小气候监测站的数据传输功耗,提出了一种切换字典的数据压缩感知方法,在对样本数据进行特征表征与分类的基础上,合理切换使用离散傅里叶变换基（Discrete Fourier transform,DFT）固定字典或K-SVD（K-singular value decomposition）学习字典,对样本数据进行稀疏表达。采用高斯函数对样本数据进行拟合,以拟合决定系数R2和拟合均方根误差（RMSE）为切换因子,定义了字典的切换策略。选用空气温度、空气湿度、土壤温度、土壤湿度作为测试对象,实验验证切换策略的可行性。实验表明,在林区小气候监测站中,当稀疏度和压缩率均相同时,结合DFT和K-SVD两种字典的优势,基于切换字典的数据压缩感知算法比单一字典具有更小的重构误差。经功耗测试实验,当稀疏度K=16时,采用切换字典的数据压缩感知算法,使监测站的平均每日电能消耗降低了16.35%,保证了林区小气候监测站的低功耗运行和数据可靠传输。     

基于BERT的水稻表型知识图谱实体关系抽取研究

针对水稻表型知识图谱中的实体关系抽取问题,根据植物本体论提出了一种对水稻的基因、环境、表型等表型组学实体进行关系分类的方法。首先,获取水稻表型组学数据,并进行标注和分类;随后,提取关系数据集中的词向量、位置向量及句子向量,基于双向转换编码表示模型（BERT）构建水稻表型组学关系抽取模型;最后,将BERT模型与卷积神经网络模型、分段卷积网络模型进行结果比较。结果表明,在3种关系抽取模型中,BERT模型表现更佳,精度达95.11%、F1值为95.85%。     

人造板分裁的优化设计

本文利用无尺寸表示矩阵、尺寸向量矩阵建立了人造板分裁的数学模型,在考虑人造板分裁的工艺要求下,建立了人造板分裁的优化目标函数和约束函数,其目标函数以最小的下脚料面积为目标函数,从而获得最低的生产成本。该优化方法简单可行,在家具生产中有着普遍的实用价值。     

杂交早粳笹优8142的栽培特性分析及高产配套技术调查

通过劳模示范田栽培调查，分析了偏高秆、偏大穗粳爪交组合优８１４２的栽培特性，并总结以大垅双行稀植为中心的高产配套技术。