苗带信息引导的水稻根区液态氮肥雾化深施机理研究

In current rice production, nitrogen applications are basically done manually. The application methods are not appropriate, and the efficiency of nitrogen uptake is low. These are the serious problems hindering the rice fertilization mechanization. This project, for the first time, proposes the technique of Atomization-Deep Fertilizing of liquid nitrogen in the root zone of rice, based on the supply and demand relationship between rice crop and nutrients, and the requirements of both machinery and agronomy. In addition, through exploring the theory and design innovation, intelligent machinery guided with the information of seedling-rows will be developed. In dealing with the difficulties of fertilizer deep placement, the project will use the approach that integrates methods of optics, mechanics, electronic, and hydraulics. The deep convolutional neural network and the image processing technology will be used to accurately perceive the information of seedling-rows at different seedling growth stages. The information will guide fertilization machines in operation to avoid damaging the crop and accurately apply fertilizer to the root zone of the crop. Modeling approach will be also taken through coupling the computational Fluid Dynamics (CFD) and Discrete Element Method (DEM) to simulate the dynamic interactions of machines with paddy soil and fertilizer materials. The results will reveal the unknown distribution of fertilizers in paddy soil, and lead to optimized design of fertilization machines. The proposed project will solve the theoretical and technical problems of precision fertilization into the root zone of rice crop, improve the uptake efficiency of nitrogenous fertilizer, and reduce agricultural non-point source pollution. It will also provide references for developing innovative mechanical fertilization technology for other crops.

在水稻生产过程中，氮肥的追施作业基本是由人工撒施完成的，追肥方式落后、肥料利用率低，是水稻施肥机械化中的最薄弱环节。本项目农机与农艺融合，根据水稻与养分的供需关系，首次提出水稻根区液态氮肥雾化深施的追肥方式，并进行机械装置创新设计理论和方法的探索，研制出具有苗带信息引导功能的智能追肥机械。在研究方法上，针对水稻追肥过程中机械深施的技术难点，光、机、电、液多领域结合开展研究。采用深度卷积神经网络和图像处理技术精准感知不同生育期苗带信息，引导施肥机械避开作物准确地将肥料追施于水稻根区；综合运用计算流体力学和离散元方法定量解析水田土壤-机械-肥料系统的动态作用过程，优化施肥器结构，揭示肥料在水田土壤中的散布规律。通过本项目的实施，不仅可以解决水稻根区精准机械追肥的理论和技术难题，提高水稻氮肥的利用率、减少农业面源污染，也可为其他作物精准机械施肥作业技术的创新提供参考和借鉴。

本项目针对水稻根区精准机械追肥开展了水稻叶色诊断技术研究、苗带图像信息识别方法与机具随行控制技术研究、液态氮肥射流注入深施机理研究、液态肥雾化开沟深施机理研究、射流喷施装置与开沟雾化施肥装置以及施肥机整机研制。首先针对水稻氮含量检测问题，提出了基于图像处理技术的水稻叶色分级检测算法，在田间光照环境下检测准确率达到了92%-95%。根据插秧期的水稻秧苗特征，提出了基于深度神经网络模型的苗带线提取算法，航向角参数提取偏差的误差为0.60°；根据返青期与分蘖期水稻秧苗特征，提出了基于图像处理方法的水稻苗带中心线检测算法，检测精度大于93%，2种苗带线提取方法均满足水稻精准施肥定位的精度要求。针对施肥作业过程中存在的偏差问题，基于自抗扰控制方法构建液压位置伺服系统，使用液压动力进行往复调节施肥装置保证其与苗带的安全距离。针对水稻精准施肥装置，基于有限元FLUENT和离散元EDEM构建了水田土壤模型，开展了开沟器-水田土壤互作机理研究，优化设计了1种新型仿生开沟器。基于计算流体力学软件Fluent构建了雾化喷嘴的的仿真模型，优化设计了1种新型扇形雾化喷嘴。利用计算流体力学软件Fluent建立了射流式施肥器的仿真模型，设计了1种新型施肥器。研发了2ZFY-6A型水稻插秧同步侧深施液肥机、射流式水稻追肥机2种装备，其中2ZFY-6A型水稻插秧同步侧深施液肥机施肥量误差为1.91%，各行排肥量一致性变异系数为1.06%，总排肥量一致性变异系数为0.16%，施肥深度合格率为93.93%，株肥侧向距离合格率为92.42%；射流式水稻追肥机各行施肥量变异系数小于1%，总排肥量稳变异系数为0.39%，施肥断条率为0%，施肥深度合格率为85.6%；两种机型各项指标均满足测试指标要求。本项目完成后，形成了比较完整的水稻根区精准机械施肥的理论方法，部分解决了水稻精准施肥的关键科学技术问题。