成熟水稻-收获机械信息感知方法及自适应巡航收获技术

It is bottleneck problems of rice combine harvesters, such as lacking collection of operating status information, intelligent decision-making of operating performance and precision operation system, which lead to unstable operating performance, operating efficiency that varies from person to person, poor adaptability of harvesting and high intensity of maneuvering. Aiming to solve these problems, research had to be done on exploring the attribute information (e.g. mature rice’s density, plant height and hierarchic ears of rice) as well as real-time perception method of landform information, and establishing the on-line accurate calculation model for feed rate. Studied the real-time monitoring method of grain breakage rate and impurity rate, real-time control model of cutting height and cutting amplitude and Beidou GNSS/Inertial Navigation INS integrated centimeter-level positioning method to obtain online information on the status of harvesting machine. Revealing the matching rules of rice attributes and combine harvester operating parameters and constructing a dynamic correlation control theory between rice-machinery systems and a multi-parameter correlation control model forms an adaptive cruising harvesting intelligent decision support system under complex harvesting environment to improve the operation performance, efficiency, adaptability and maneuvering comfort of the machine; integrated development of a self-adaptive cruise harvesting intelligent control system of the rice combine harvester, test and optimize. The research results of the project will significantly enhance the intelligent level of rice combine harvester and the core competitiveness of our country products, and have important scientific significance and engineering application value for ensuring food security in China

针对水稻联合收获机缺乏作业状态信息收集、作业性能智能决策和精准作业系统导致作业性能不稳定、作业效率因人而异、收获适应性差、操纵强度大等瓶颈问题，探索待收获区域成熟水稻密度、株高、穗层等属性信息和地表形貌信息实时感知方法，建立喂入量在线精准计算模型；研究籽粒破碎率与含杂率实时监测方法、割茬高度和割幅实时控制模型以及基于北斗GNSS/惯导INS组合导航的连续厘米级定位方法，获取机器收获时工作状态在线信息；揭示水稻属性与联合收获机工作参数的匹配规律，构建水稻-机器系统之间的动态关联调控理论和多参数关联调控模型，形成复杂收获环境下自适应巡航收获智能决策支持系统，提高机器的作业性能、效率、适应性和操纵舒适性；集成开发水稻联合收获机自适应巡航收获智能控制系统，并进行试验与优化。项目研究成果将显著提升我国水稻联合收获机的智能化水平和产品的核心竞争力，对于保障我国粮食安全具有重要的科学意义和工程应用价值。

传统水稻联合收获机缺乏作业状态信息收集、作业性能智能决策和精准作业能力，导致整机作业性能不稳定、作业效率因人而异、收获适应性差、操纵强度大。为解决以上瓶颈问题，本研究以整机作业性能和作业效率的最佳平衡为目标，重点探索了适收期成熟水稻属性和机器状态信息在线感知方法和水稻属性和机器系统动态关联调控策略，开发了成熟水稻自适应巡航收获系统。主要研究内容和结论如下：.（1）基于深度高程图像结合灰度直方图扫描计算得到水稻株高信息，平均绝对误差为2.92cm；基于语义分割和图像特征，以骨架线节点和穗头像素面积为特征因子建立水稻密度预估模型（R2为0.835）；采用最大类间方差法，输出作物收获边界的横向偏移平均绝对误差为2像素，航向夹角平均绝对误差为1.65°；设计了双自由度随地仿形传感器，构建了割茬高度灰色预测调控系统，平均偏差控制在21mm以内；研制了水稻籽粒含杂破碎率一体化检测装置并开发了实时检测算法，实际机收时含杂率检测相对误差为2.6%~9.1%，破碎率相对误差为6.0%~12.8%；搭建了GNSS/INS松组合导航系统，基于卡尔曼滤波算法实现GNSS和INS的导航信息的有效融合;.（2）以水稻株高、密度、割幅和留茬高度为输入，前进速度为输出，建立了水稻联合收获机喂入量多参数模糊控制模型，实现了喂入量稳定调控；以籽粒清选夹带损失率、含杂率和破碎率为输出，筛片开度、风机进风口面积和脱粒间隙为输入，采用粒子群优化与小波神经网络融合对脱粒清选系统状态空间模型进行辨识，基于模型预测控制方法建立脱粒清选多参数协同控制策略，实现了损失率、含杂率和破碎率的快速稳定控制，系统超调量≤0.06%、响应时间≤3s；.（3）研制了机器作业状态实时监测、基于物联网的机器作业状态数据无线传输以及基于虚拟现实技术的机器作业状态远程管控模块，集成开发了水稻联合收获机自适应巡航收获车载电子信息系统，并通过多轮田间试验不同工况下自适应巡航收获控制系统的实时性、鲁棒性和收获适应性。