养分专家系统推荐施肥对棉花产量及肥料效率的影响

[目的]新疆棉花种植中存在肥料用量较多和不合理施用带来的肥料利用率低问题,建立基于产量反应和效率的棉花推荐施肥方法棉花养分专家系统(Nutrient Expert for Cotton,NE),研究其在棉花上的应用效果.[方法]2020年,田间试验在库尔勒市和尉犁县,共设置7个处理,分别为棉花养分专家系统推荐施肥(NE...     

Indicators of soil quality and crop productivity assessment at a long-term experiment site in the lower Indo-Gangetic plains

Deterioration of soil quality under resource-intensive modern agriculture in the face of global climate change poses a huge risk to food security. Because of the complex nature, estimators of soil quality often rely upon a limited set of soil attributes, along with statistical data reduction techniques, for developing quality indices, whilst overlooking biological aspects and regional climatic variability. This study screened the most suitable soil quality indexing approaches for a rice-oilseed-based cropping system in the lower Indo-Gangetic plains (IGP). For this, surface soil samples (0–15 cm) were collected from an ongoing long-term fertilizer experiment with a rice-mustard-sesame cropping system in the IGP. The following treatments were assessed for their effect on soil quality: T₁-control, T₂-NPK (recommended NPK doses), T₃-NPKG (NPK + in situ green manuring), T₄-NPKGB (NPK + in situ green manuring + biofertilizer) and T₅-NPKF (NPK + farm yard manure FYM). We found that total organic carbon (TOC), β-glucosidase, CaCl₂ extractable S, alkaline KMnO₄ oxidizable N, activity of urease, amidase enzyme and mean weight diameter (MWD) were sensitive key indicators of soil quality. The NPKF treatment maintained the highest soil quality status (0.80–0.91), both under productivity and environmental protection goals, owing to the availability of decomposable carbon. Regression analysis showed a better agreement of equivalent rice yield with expert opinion (EO; R² = 0.89) than principal component analysis (PCA; R² = 0.76). Finally, we found that the expert opinion approach with the nonlinear scoring function was the best tool for soil quality assessment of the region.     

饲料加工工艺设计专家系统的设计

采用面向对象的设计方法，开发了基于Ｗｉｎｄｏｗｓ的配合饲料加工工艺流程设计专家系统（ＦＰＥＳ）的实现技术，该系统根据人类专家的设计知识，从饲料厂工程设计从大纲入手，能自动完成设备选型，设计计算，参数选择，流程图绘制等工作。     

基于图像自动标注与改进YOLO v5的番茄病害识别系统

针对作物病害识别系统功能单一，缺乏系统升级机制，人工升级系统成本较大的问题，以番茄病害为例，提出了基于OpenCV的番茄叶片图像自动标注算法和改进YOLO v5的番茄病害识别模型；结合数据集自动划分、模型自动训练与评估、手机APP自动创建与更新理念，设计了一种可以自动升级的番茄病害识别系统；引入专家审查校正机制，提高了系统识别结果的可靠性。实验结果表明，该系统实现了对番茄的健康叶片与9类病害叶片进行识别，可以在实际应用中通过手机APP识别番茄病害的同时自动扩充番茄病害图像数据集，并根据数据扩充量自动启动系统的升级优化流程，由此不断提升该系统的番茄病害识别性能。该系统为番茄生产提供了一个便捷、可靠的番茄病害识别工具。     

中国人工智能国际合作顶尖专家识别与评价

[目的 /意义]旨在通过对中国人工智能国际合作顶尖专家进行识别与评价,以期为加强中国人工智能国际合作提供参考。[方法 /过程]以人工智能领域H指数大于40的专家确定为顶尖专家进行研究,依托Aminer科技大数据平台,对中国人工智能国际合作的顶尖专家进行识别,并利用Topsis法对专家的综合学术水平进行评价。[结果 /结论]针对与中国有国际合作的顶尖专家识别结果,发现:(1)中、美两国的专家合作数远高于其余国家之和;(2)专家以男性为主导,女性比例极低;(3)专家所属机构分布在全球22个国家(地区),机构分布与各地区的科技、经济实力情况大体一致;(4)部分专家拥有学校研究人员及企业科学家双重身份;(5)中、美专家在论文学术领域的国际合作占据主导地位。国际迁移合作以中国学者到国外学习深造或工作为主,同时对国外专家引进的迁移合作,仍以美国专家作为优先对象。同时,Topsis评价结果显示中国人工智能的国际合作虽然已经全面铺开,但是与部分顶尖专家仍然未建立有效的直接交流合作。     

基于SpringBoot框架的群养母猪饲喂站管理系统设计与试验

母猪的群体养殖是一种能保证动物福利，提高母猪生产性能的重要养殖管理模式。为实现母猪群养过程的数字化及智能化，基于自主研发的群养母猪饲喂站硬件平台，采用B/S模式设计一个群养母猪饲喂站管理系统。系统后端使用基于Java语言的SpringBoot框架作为系统组件管理和运行的主要框架；设计后端数据库中的数据结构，优化后台系统中持久层数据的操作访问；使用Netty实现饲喂站及客户端与后台间的实时通信；基于Spring事务管理，封装可复用的业务逻辑，设计母猪信息管理、饲喂计划专家系统、饲喂站实时监控等系统功能模块。系统前端采用渐进式Vue框架实现用户与各个功能模块的人机交互。试验表明：本系统能实现对群养母猪饲喂站的控制和状态监测，并根据母猪特性和专家知识库动态生成每天的饲喂计划；在基础实验条件下，系统的饲喂站并发量可超过120台，理论最大母猪容纳量超过5 400 头，能够满足大规模母猪饲喂的需求。     

安徽省单季中稻生产智能决策支持系统的研制

本系统的研制是基于面向对象的构件技术和雄风专家系统开发工具的集成环境来完成的.以安徽省单季中稻生产为对象,以实现水稻生产的高产、优质、高效为目标,综合了水稻栽培、植保、土肥、农业生态等各个领域专家的知识经验,采用从定性到定量综合集成的方法建立.系统具有逻辑推理判断与决策能力,并有较高的精确性和可靠性.     

Laconex: A rule-based implementation to advise on plot-level land conservation in Vilathikulam,Southeastern India

Knowledge about land conservation exists along a spatial continuum, represented at the broadest level (with least depth) by international-level experts who have international-level expertise, and at the narrowest level (with most depth) by the individual farmer with the most site-specific knowledge at the plot level. This continuum covers many different aspects of knowledge about land degradation and its amelioration, including technical and cultural knowledge. This paper reports an effort (one in a series of steps) to gather, formalize, codify, and operationalize knowledge about land conservation from two different levels of the continuum. Land conservation knowledge was elicited from farmers and extension agents in Vilathikulam taluk (an area of contiguous villages) in Tamil Nadu, southeastern India, using semi-structured interviews. The knowledge thus gained was combined with knowledge from the literature on soils and land conservation for the area. This information was organized into factual knowledge and process knowledge, and implemented in the form of general recommendations, constraints, and specific recommendations. The process of arriving at these categories was organized into several steps using a spatial anchor and soil type as keys. These were operationalized in a rule-based computer program (LAnd CONservation EXpert -LACONEX), which offers land conservation advice at the plot level for Vilathikulam taluk , using an expert system development environment.