水田土壤承载信息连续感知与拖拉机运动姿态预测估计

According the problem that degrees of mechanical automation and productivity need a further improvement in the process of rice production, a composite technology which involves continuous perception of paddy fields bearing information and prediction of motion postures of paddy tractor is proposed by the program applicant. The purpose of it is tomeet requirements for agricultural machinery, and improves the precise, intelligent, effective, as well as high-quality working. The applicant projected need to solve the following key problems:. 1) A tractor-implement rigid model will be built to describe tractor motions in paddy fields;. 2) The innovative invention ahead of front wheels sensing rod mounted is used to gather displacement, velocity, and force information from paddy soils. The interaction mechanism between paddy field soils and tractors as well as it between sensing-rod and paddy soil will be clarified. Eventually, a expressing signage-relationship model forward among sensing rod, bear capacity of paddy soils, tractor wheels will be constructed;. 3) The forward dual-track spectra data will be estimated using the present terrain information inverted by real-time motion attitudes of tractor;. 4) Based on the above analysis, the Kalman prediction algorithm will be developed to predict the continuous motion postures in a certain period of time, usually a few seconds.. The core technologies and related devices mentioned above will serve as a theoretical underpinning for mechanical controls in precise farming, early warning for tractor stuck, construction of tractor model, and predictive control of motion postures under paddy field circumstances.

针对水稻生产水田作业机械自动化程度和作业效率有待进一步提高问题，以实现水田作业机械智能精准与高质高效作业为目标，项目申请者提出通过水田土壤承载信息连续感知技术和水田拖拉机运动姿态预测估计技术，以提高水田作业机械作业质量和作业效率。申请者拟建立“水田拖拉机－悬挂机具”整车刚体模型以描述拖拉机水田运动，设计拖拉机左右前轮前方触杆连续传感装置，阐明“触杆－水田土壤”和“水田土壤－拖拉机轮”互作机理，建立“触杆－水田土壤承载－拖拉机轮”前方沉陷模型，利用拖拉机实时运动姿态反演的拖拉机四轮当前地形为初始地形，预测估计拖拉机前方连续的“双轨路谱”，设计Kalman预测估计算法预测估计拖拉机一定时间内的连续运动姿态。为水田农业机械精准作业控制和水田拖拉机陷车等提供预测控制信息，为水田拖拉机模型建立和运动姿态预测控制提供理论基础。

针对水稻生产水田作业机械自动化程度和作业效率有待进一步提高问题，项目以水田犁底层信息感知和不平度量化表征，“拖拉机—机具”整车建模，拖拉机作业姿态预测为切入点开展了共性科学问题和关键技术研究，以期为水田机械智能精准与高质高效作业提供普适性借鉴依据。针对水田犁底层信息化和量化研究，首先，设计了水田犁底信息感知触感装置连续测定了犁底层高程和农机前进阻力及其表达方法；其次，构建了水田土壤承载计算的JKR数值分析模型，通过实验测定和标定试验获得JKR模型参数：含水率45.66%，密度1809 kg/m3，坍落度233.5 mm，扩展度350.8 mm，泊松比0.42，剪切模量0.4 MPa，土壤—土壤恢复系数0.01，土壤—土壤静摩擦系数0.1；最后，以水田拖拉机行驶时z轴加速度作为犁底不平度表征，采用Welch法对采集得到的加速度序列进行谱估计，并依据ISO8608:2016和GB/T 7031-2005标准规定的多倍频程数据平滑方法对谱估计结果进行平滑，得到了水泥地不平度为A级，砂石路为B级，水田犁底介于B~C之间。为了构建水田“拖拉机—机具”整车模型，首先，分析了以插秧机动力配套为代表的水田农机重心分布区间（后轮承载系数0.65~0.7）；在此基础上，基于MapleSim建立了平地机调平机构关键部件多体动力学模型，采用高速相机及其图像分析软件TEMA 验证机具质心位置和姿态角，结果表明实测平地铲质心坐标与仿真运算最大误差约为10 cm，姿态运动与多体动力学模型分析结果基本一致；最后，通过解析法分别构建了水田整车的行驶动力学模型和行驶运动学模型，以模型参数可获取性为出发点，确定并辨识了整车扩展运动学模型参数，得到了水田车辆运动侧滑估计。研究围绕时间序列理论，离线构建了采集自多地的水田拖拉机作业姿态的时序模型，通过对各模型结构共性的分析，确定以AR（15）描述农机俯仰角运动变化，设计了相应的递推最小二乘法估计模型参数，实现了对农机未来1 s内运动姿态的在线连续预测估计。综上所述，项目严格按照研究计划执行，完成了预期研究目标，取得了一定基础性和具有借鉴意义的研究成果。