便携式柑橘虫害实时检测系统的研制与试验

为实现柑橘虫害的快速、准确识别,帮助果农及时掌握果园内虫害的危害程度和分布情况,该研究结合嵌入式图像处理技术设计了一套基于深度卷积神经网络的柑橘虫害实时检测系统。优选MoblieNet作为虫害图像特征提取网络,区域候选网络生成害虫的初步位置候选框,快速区域卷积神经网络(Faster Region Convolutional Neural Networks,Faster R-CNN)实现候选框的分类和定位。检测系统根据目标图像中虫害数量计算危害程度,按照正常、轻度、中度、重度4个等级判定柑橘虫害的严重程度,形成虫害识别与级别定量化测评软件。最后引入北斗模块获取采样点位置信息,进一步处理成可视化的虫害热力图。结果表明,该方法可实现对柑橘红蜘蛛和蚜虫的快速准确检测,识别准确率分别达到91.0%和89.0%,单帧图像平均处理速度低至286ms。该系统实现了柑橘虫害的精准识别与定位,可为农药喷洒作业提供精准信息服务。     

基于自监督学习的温室移动机器人位姿跟踪

为实现温室环境下机器人行进过程中的位置及姿态跟踪,该研究提出一种基于时序一致性约束的自监督位姿变换估计模型。模型用软遮罩,处理视频帧间静止造成的位姿预测值收缩现象,进一步用归一化遮罩,解决非刚体场景和目标遮挡问题。设计了一种星型扩张卷积,并基于该卷积,为模型构建自编码器。在采集自种植作物为番茄的日光温室视频数据上开展训练和测试试验。结果表明,与不采用遮罩处理的模型相比,采用软遮罩的模型,位置和姿态估计相对误差分别减少5.06个百分点和11.05个百分点,采用归一化遮罩的模型,这2项误差则分别减少4.15个百分点和3.86个百分点,2种遮罩均可显著提高模型精度;星型扩张卷积对降低模型误差是有效的,在网络参数不变的前提下,该卷积使姿态估计相对误差减少7.54个百分点;时序一致性约束使姿态估计均方根误差下降36.48%,每百帧累积姿态角误差降低54.75%,该约束可用于提高模型精度及稳定性;该研究的位置及姿态估计相对误差分别为8.29%和5.71%,与Monodepth2相比,减少了8.61%和6.83%。该研究可为温室移动机器人导航系统设计提供参考。     

黄土高原水蚀风蚀交错带小流域土壤水分季节变化特征与主控因素

土壤水分是黄土高原水蚀风蚀交错带生态环境恢复的关键因子,具有明显的时空异质性。以水蚀风蚀交错带的代表性流域—老爷满渠小流域为对象,采用网格法(50 m×50 m)共布设了73个样点,原位观测0—5 m土壤含水率,共测定23次(2013年6月至2019年10月),通过获取每个样点的环境因子,结合经典统计、地统计、随机森林等方法,分析了小流域尺度不同土层深度(每层1 m,共5层)土壤含水率的季节变化特征与影响因素。结果表明:不同土层土壤含水率的空间分布特征和季节性变化规律不同;对于0—1 m土层,土壤含水率在夏季和冬季之间存在显著性差异(p0.05),而对于1 m以下土层,春季平均土壤含水率高于其他季节,但不显著;无论在何种季节,不同土地利用方式、壤土与砂土间的土壤含水率在3 m以上土层均存在显著差异(p0.05),而阴、阳坡的土壤含水率在所有土层均存在显著差异(p0.05);在不同季节,土壤含水率与容重和砂粒呈负相关,与其他环境因子(有机碳含量、黏粒、粉粒、有机碳密度、土地利用、坡向、土壤质地和pH)呈正相关;除有机碳密度和黏粒较为稳定外,土壤含水率与环境因子的相关性均随土层深度增加呈减少趋势;环境因子对土壤含水率空间变异的整体相对贡献表现为土壤性质地形土地利用。研究结果可为研究区深层土壤水资源管理、土壤水文观测与模拟、植被优化布局等提供参考。     

棉秆热解分级冷凝产物改良盐碱土效果

为揭示分级冷凝热解液对盐碱土的改良作用,该研究以三级冷凝方式作为棉秆热解液精制方法,分析分级冷凝对产物组成影响;在此基础上,利用盆栽试验,以不同热解温度下获得的棉秆分级冷凝第3级产物稀释不同倍数后施入盐碱土,考察分级冷凝第3级产物对于小白菜生长、土壤pH值、全盐量、蔗糖酶、脲酶、碱性磷酸酶、过氧化氢酶、有机质,碱解氮、速效磷和速效钾的影响。结果表明：1）分级冷凝可以达到棉秆热解液从源头分离的目的,实现酚类与酸类物质的粗分离,分级冷凝第3级产物的酚类与酸类物质浓度分别为单级产物的0.50～0.55与1.21～1.35倍。2）灌施分级冷凝第3级产物可以显著促进小白菜的萌发与生长（P<0.05）,热解温度与稀释倍数均会影响促进作用。中温热解以及较大的稀释倍数对种子萌发更有利,500℃热解获得分级冷凝第3级产物稀释600倍灌施时,小白菜的长势最好,发芽率、株高与株质量分别提高32.12%、95.11%与120.03%,丙二醛含量降低48.39%。3）灌施分级冷凝第3级产物会显著影响土壤酶活性,除稀释50倍时降低土壤脲酶与碱性磷酸酶活性外,其他稀释倍数下土壤中酶活性显著增加（P<0....     

PCB removal,soil enzyme activities,and microbial community structures during the phytoremediation by alfalfa in field soils

Purpose  

An in situ phytoremediation trial was developed in order to investigate the function of alfalfa during a 2-year bioremediation of an agricultural soil contaminated with polychlorinated biphenyls (PCBs). The study was conducted with the aim to better understand the application potential of PCB phytoremediation at field scale.     

Characterization of phytochemicals and antioxidant activities of a purple tomato (Solanum lycopersicum L.)

A newly developed nongenetically modified purple tomato V118 was investigated for its phytochemical compositions and antioxidant activities. A highly efficient and sensitive UPLC method was developed for both the phenolics and carotenoids, which showed that in addition to the phytochemicals commonly known for tomatoes, V118 had a unique composition of anthocyanins. The total carotenoid content of V118 was 234.78 μg/g dry weight (DW), and the total phenolic content was 659.11 mg GAE/100 g DW. The antioxidant activities of the lipophilic extract as measured by the PCL and ORAC-L assays were 30.11 μmol TE/g DW and 11.97 μmol TE/g DW, respectively, while the hydrophilic extracts as determined by the ORAC-H and FRAP assays were 323.23 μmol TE/g DW and 54.95 μmol AAE/g DW, respectively. The LC-MS study showed three major anthocyanins, which were mainly acylglycosides of petunidin and malvidin. This study showed that purple tomatoes such as V118 possess additional phytochemicals like anthocyanins, which can potentially have added health benefits.     

Development of a Sequential Decision-Making Model for Controlling Multiple Air Pollutants Under Stochastic Uncertainty

Most of previous programming methods for air-quality management merely considered single pollutant from point sources. However, air pollution control is characterized by multiple pollutants from various sources. Meanwhile, uncertain information in the decision-making process cannot be neglected in the real-world cases. Thus, an inexact multistage stochastic programming model with joint chance constraints based on the air quality index (air-quality management model with joint chance constraints (AQM-JCC)) is developed for controlling multiple pollutants deriving from point and mobile sources and applied to a regional air-quality management system. In the model, integrated air quality associated with the joint probability existing in terms of environmental constraints is evaluated; uncertainties expressed as probability distributions and interval values are addressed; risks of violating the overall air-quality target under joint chance constraints are examined; and dynamics of system uncertainties and decision processes under a complete set of scenarios within a multistage context are reflected. The results indicate that useful solutions for air quality management practices in sequential stochastic decision environments have been generated, which can help decision makers to identify cost-effective control strategies for overall air quality improvement under uncertainties.     

H5N1亚型AIVHA抗原表位的高效表达及其抗原活性分析

血凝素（hemagglutinin,HA）蛋白是禽流感病毒（avian influenza virus,AIV）的一个重要表面抗原性蛋白,在疾病诊断和防治上有重要意义。本研究为了探讨一种更为简便有效的HA重组蛋白表达途径,利用生物信息学软件,对H5N1亚型AIVHA基因编码的氨基酸序列进行分析,在分析其在大肠杆菌中的密码子偏好性、稀有密码子分布情况及有关蛋白的抗原性等重要特性后,构建了HA抗原表位重组表达质粒pET-32a（＋）-HA。经测试,该重组质粒在1mmol/LIPTG诱导剂作用下诱导过夜,能在大肠杆菌Rosetta-gami B（DE3）中高效表达,并得到48.1kD大小的目的重组表达蛋白。重组蛋白用6×His-tagged protein纯化试剂盒纯化后,与福氏佐剂等量混合制备成抗原,以200μg/鸡的剂量皮下注射2月龄SPF鸡3次,采血分离血清。Western-Blot试验结果表明,该重组表达蛋白能分别与所制备的高免鸡血清及H5N1亚型AIV阳性血清发生特异性反应,在硝酸纤维素膜上出现特异性杂交带。说明本试验研究的HA抗原重组表达蛋白具有良好的免疫原性和反应原性,保留了HA蛋白的抗原活性,提示该重组蛋白在H5亚型AIV的防治技术研究中具有重要的实际应用价值。     

基于语义本体的柑橘肥水管理决策支持系统

利用信息技术实现柑橘精准生产管理是果树信息化管理的重点和难点。该文针对山地果园肥水信息化精准管理问题,研发了基于语义本体的柑橘肥水管理决策支持系统。柑橘肥水语义本体是决策支持系统的核心,它以资源描述框架三元组整合涉农信息。系统实现了施肥、生理病害防治和排灌监测3个功能。系统验证结果表明:1)施肥查询模块能够根据树龄、施肥时期、土壤质地、地形和产量5个因素计算氮、磷、钾施肥量,经216组数据验证其输出正确率100%;2)经100组涵盖12种营养素缺乏和过剩生理病害数据验证,病症查询模块能够根据输入病症正确判断生理病害;3)排灌检测模块能根据不同土壤质地对排灌实时监测,经小型气象站获取的土壤含水率数据验证,系统正确预警率达100%。经361组性能测试数据结果表明系统平均响应时间在0.23 s(浏览器端应用)和0.58 s(手机端应用)内,具备较优性能。该研究为农业领域知识建模和异构多源数据整合等问题提供了可行方案。     

生物质灰渣与化肥混合对氨挥发的影响

采用室内恒温培养,碱液吸收法测定灰渣-化肥中氮的挥发量,研究了锯木灰、谷壳灰、玉米灰、水稻灰分别与不同化肥配比下,在一定时间内的氨挥发规律。结果表明:四种灰渣在N1、N2、N3水平下,随着时间的增加氨的挥发量和挥发率呈先增大后减小的趋势,各处理的峰值多出现在10~15 d的培养阶段;谷壳灰、玉米灰、水稻灰添加磷酸二氢钾能明显抑制混合物中氨的挥发,而锯木灰添加氯化钾、磷酸二氢钾均能抑制氨的挥发;在四种灰渣中,锯木灰处理的氨挥发率明显高于其他三种灰渣,特别是在N3水平下比同期的其他灰渣处理高5倍。Elovich方程和抛物线扩散方程均能较好拟合氨的挥发量随时间的变化,其相关系数均达到极显著水平。