Approach for Load State Detection in MIN

Some thoroughly research on the basic method for load state detection in MIN is introduced. Load sources of SCP and SSP in MIN are analyzed in detail after investigating the relationship between inter resource and load state in the system. And load detection parameters and model are presented. Moreover, a novel strategy to detect system load state based on fusion of multi detection data is presented. By some test, the theory and method are proved most feasible. Our research show that load state of MIN can be recognized by detecting utility of internal resource utilizing of single node in MIN, and multi redundant detecting data fusion is effective approach to improve the reliability and trust degree of the result.     

坝上生态农业工程生态效益监测与评价研究

利用遥感技术、地理信息系统和全球定位系统对河北坝上实施生态农业工程建设后的生态效益进行了监测与评价。定量描述了坝上生态环境的动态变化趋势，并对坝上生态农业工程的气候效益、土壤改良效益、农业水文生态效益进行了分析与评价。研究结果表明，通过工程治理，使农业生态系统的功能增强，结构合理，取得了显著的生态效益。     

五轴数控加工叶片无干涉刀位轨迹的计算

根据微分几何特性，提出五坐标数控加工叶片时对叶片的工作面和背面的刀位轨迹计算分开予以讨论，对工作面而言，利用平头立铣刀加工时不存在干涉，而对背面而言，可能存在干涉。根据干涉的种类，分别对全局干涉和局部刀具干涉提出了快速判定依据，并对干涉点进行了修正。指出利用Iso-scallop法可获得最优的轨迹和蔗，并给出了实例。     

基于随机森林回归算法的小麦叶片SPAD值遥感估算

使用机器学习中的随机森林(RF)回归算法构建小麦叶片SPAD值遥感反演模型。以2010—2013年江苏地区试验点稻茬小麦3个生育期(拔节、孕穗、开花)的叶片为材料,结合我国自主研发的环境减灾卫星HJ-1对研究区域进行同步监测,分析了各生育期叶片SPAD值与8种植被指数间的相关性;以0.01水平下显著相关的植被指数作为输入参数,使用RF回归算法构建了每个生育期的小麦SPAD反演算法模型,即RF-SPAD模型,以支持向量回归(SVR)和反向传播(BP)神经网络算法构建的SVR-SPAD模型和BP-SPAD模型作为比较模型,以R2和均方根误差(RMSE)为指标,分析了每个生育期3个模型的学习能力和回归预测能力,结果表明:RF-SPAD模型在3个生育期都表现出最强的学习能力,R2和RMSE在拔节期分别为0.89和1.54,孕穗期分别为0.85和1.49,开花期分别为0.80和1.71;RF-SPAD模型在3个生育期的回归预测能力都高于BP-SPAD模型,高于或接近于SVR-SPAD模型,R2和RMSE在拔节期分别为0.55和2.11,孕穗期分别为0.72和2.20,开花期分别为0.60和3.16。     

播种检测技术的研究进展与趋势

播种作业过程中,一旦出现播种异常现象应能及时报警,并检测出故障的性质和位置,以便机手排除故障,提高播种机的播种质量。要实现这一目的,就必须对播种性能施全程监控。为此,介绍了播种检测技术发展概况,指出了提高精密播种机田间播种质量的关键问题,分析了播种检测技术的发展趋势。     

注水肉无损检测技术现状与发展趋势分析

针对目前市场上注水肉现象加重问题,有必要研究注水肉的检测技术。首先对注水肉的传统检测方法进行简要论述,分析总结了传统方法的弊端;然后对现阶段肉品含水率的4种主要无损检测方法:生物电阻抗法、核磁共振方法、超声波法和近红外光谱法的研究现状及特点进行了总结分析,着重叙述了近红外光谱技术在肉品含水率及品质检测中的研究情况及技术特点,提出了未来技术发展方向。     

油桃外部缺陷的高光谱成像检测

采用高光谱(420~1 000 nm)成像技术对"中油9号"油桃的4种外部缺陷(裂纹果、锈病果、异形果和暗伤果)进行检测判别。对400个样本(4种外部缺陷样本和完好样本)运用偏最小二乘回归(PLSR)从全波段中分别提取了10条特征波长,分别为497、534、657、677、696、709、745、823、868、943 nm。缺陷样本的高光谱图像经过主成分分析后,对876 nm下的单波段图像通过掩膜、Sobel算子处理,并对主成分图像经过区域生长算法实现缺陷样本的缺陷区域分割。对光谱数据进行主成分分析得到前10个主成分值,并对图像数据采用灰度共生矩阵(GLCM)提取得到6项图像纹理指标(均值、对比度、相关性、能量、同质性、熵值)。将主成分值和纹理值融合建立极限学习机(ELM)模型对油桃外部缺陷进行检测判别。结果表明,该模型对缺陷样本的判别正确率为91.67%,完好样本的正确率为100%。     

融合光谱与图像信息的河套蜜瓜糖度在线检测试验系统

针对蜜瓜糖度在线检测的需求,设计了融合漫透射光谱与图像信息的河套蜜瓜糖度在线检测试验系统,该系统包括硬件平台和软件系统两部分。硬件平台主要包括蜜瓜输送装置、光谱采集装置、图像采集装置和控制系统4部分。软件系统基于Microsoft Visual C++6.0语言,结合Omni Driver软件、Fly Capture2及Open CV软件开发。系统可实现蜜瓜光谱与图像信息的自动采集、显示及保存,可实现对漫透射光谱预处理,获取糖度检测所需光谱数据,对图像预处理,提取蜜瓜外观特征图像信息(R、G、B颜色值和蜜瓜体积)。在此基础上,系统可通过融合漫透射光谱与图像信息的蜜瓜糖度检测模型,计算蜜瓜糖度。系统同时可实现检测个数统计,外观特征信息及糖度的实时显示、保存等功能。测试试验表明,该试验系统检测1个样品用时1.2 s,糖度检测均方根误差为1.22,可满足河套蜜瓜糖度在线检测试验需求,为进一步开展河套蜜瓜糖度在线检测研究奠定了基础。     

Monitoring forage production for farmers’ decision making

Objective management of grazing livestock production systems needs monitoring of forage production at the managerial unit level. Our objectives were to develop a system that routinely estimates forage above-ground net primary production (ANPP) at the spatial and temporal resolution required by farmers in the Pampas of Argentina, and to facilitate adoption of the system by end users as a managerial support tool. Our approach was based on the radiation use efficiency (RUE) logic, which proposes that ANPP is determined by the amount of photosynthetically active radiation absorbed by the canopy (APAR), and the efficiency with which that energy is transformed in above-ground dry matter (radiation use efficiency, RUE). APAR is the product of incoming photosynthetically active radiation (PAR) and the fraction absorbed by the canopy (fPAR). We estimated fPAR as a non-linear function of MODIS normalized difference vegetation index (NDVI). RUE was empirically estimated for the two principal forage resources of the region, yielding the following relations: ANPP = 0.6 × APAR + 12, (R² = 0.86; p < 0.001; n = 18) for the upland sown pastures, and ANPP = 0.27 × APAR + 26, (R² = 0.74; p < 0.001; n = 18) for the lowland naturalized pastures, with ANPP in g/m²/60 days and APAR in MJ/m²/60 days. The models were able to predict independent ANPP values with acceptable accuracy. Computational procedures were automated and run in a Relational Data Base Manager System that stored and managed all the information. The system is currently monitoring 212,794 ha in 83 farms and provides monthly ANPP values for the previous month and a history of the last 6 years. The data so generated show ANPP differences between the two major forage resources, considerable variability of a given month’s ANPP among years and paddocks, and contrasting among-farm differences in the efficiency of conversion of ANPP and forage supplements into beef production. The system was well accepted by end users who utilize it mainly for making near real time decisions according to last month ANPP, and explaining results of previous production cycles by incorporating ANPP as an explicative variable. However, there were differences among farmers in the degree of utilization, apparently related to the advisor’s attitude toward this new technology. Our results indicate that (1) forage production of large extensions can be monthly monitored at the paddock level by a small laboratory with capabilities in geographic information systems, and (2) advisors and farmers apply this information to their managerial decisions.