基于协同分布式随机模型预测控制的水稻联合收获机作业性能协调控制方法研究

In view of the difficulties in monitoring of grain impurity ratio and grain broken ratio on-line, the inability to detect abnormal monitoring values of operation status in time which will lead to driver's misjudgment and the malfunction of control system, and the poor harvesting adaptability and low efficiency in different working condition, this project gives a potential key to improve harvesting performance: firstly, the heat transfer models for different grain components in the detection space will be established based on the active infrared thermal imaging technology, and a high-resolution grain impurity and broken ratio on-line detection device will be developed. Then, a sensor fault diagnosis and isolation system will be constructed using an advanced data processing algorithm to provide reliable input signals for the control system. At last, a cooperative control method for adjusting feeding, threshing &separation and cleaning working process will be developed to suit for uncertain harvesting environments, such as different rice varieties, different planting modes and different harvesting time. The established control system can be expected to realize the coordinated control of harvesting performance, such as forward speed, grain sieve loss ratio, grain impurity ratio, grain separation loss ratio, and grain broken ratio. Harvesting efficiency and adaptability of rice combine harvesters can be significantly improved under the complex harvesting conditions by using the developed control method. Corresponding theories, methods and technology also can be used for harvesting wheat, rape, soybean and other crops with good performance. The results of the project also can provide theoretical and technical support for ensuring food security in China.

针对水稻联合收获机田间作业过程中籽粒含杂率、破碎率在线动态检测困难；无法及时发现作业状态监测值异常而使驾驶员误判、控制系统误动作；联合收获机在不同作业环境下收获适应性差、作业效率低等问题，基于主动红外热成像技术，通过建立粮箱内谷物各成分在检测空间内的传热模型，研制高分辨率籽粒含杂率、破碎率在线检测装置；应用先进数据处理算法，构建传感器故障诊断与隔离系统，为控制系统提供可靠的输入信号；探索水稻品种多样、种植模式不一和收获时间差异等不确定收获环境下，联合收获机行走喂入、脱粒分离、清选工作过程的协同控制方法，实现前进速度、籽粒清选损失率和粮箱籽粒含杂率、籽粒夹带损失率和粮箱籽粒破碎等作业性能的协调控制，显著提高复杂收获环境下水稻联合收获机的作业效率与收获适应性。相应的理论、方法与技术还可用于小麦、油菜、大豆等其他作物的高效、自适应机械化收获，可为保障我国粮食安全提供理论和技术支撑。

针对水稻联合收获机田间作业过程中作业性能指标在线动态检测困难，联合收获机在不同作业环境下收获适应性差、作业效率低等问题，本项目通过分析粮箱内谷物各成分形态特征，根据检测实时性要求，研究检测装置软硬件配置方案，比较了不同图像处理算法（单阈值、自动阈值分割方法）对图像区域不同特征的分割效果，着重比较了不同自动阈值分割方法对图像分割的影响，计算了谷物各成分在检测区域内的面积占比，建立了籽粒含杂率和破碎率计算模型，确定了检测装置的合理安装位置和采样周期，实现籽粒含杂率及破碎率的实时监测，试验结果表明，含杂率和破碎率监测的相对误差分别为≤9.42%, ≤8.10%。研发了联合收获机多传感器信息耦合及故障诊断技术，通过设计基于高速摄影和高速信号采集的籽粒碰撞试验装置，优化了籽粒损失监测传感器信号调制电路，籽粒清选损失监测误差≤4.48%。以STM32处理器为核心，构建了由数据采集、数据输出、数据接收、数据显示、文件系统等模块组成的联合收获机的作业信息采集系统，实现了主要作业参数的实时动态采集。分析了各传感器在不同收获条件下正常监测数据及传感器故障时监测数据特征，根据故障传感器监测数据的变化规律，对比各先进数据处理算法对不同类型传感器的故障诊断与预测效果，开发可视化传感器故障诊断人机交互界面，提示驾驶员故障传感器位置。建立了联合收获机作业性能优化控制模型，分析了收获过程中不同前进速度和工作参数组合下行走喂入、清选过程内各动态变量时间序列，根据动态变量间关联性，分析过程模型参数与待收获田块水稻特性（水稻品种、含水率等）和收获时间等因素的关系，分别构建了行走喂入和清选工作过程内工作参数控制模型，并进行了田间收获试验验证，试验结果表明整机总损失率≤1.81%，籽粒含杂率≤1.44%，籽粒破碎率≤0.98%，优于国家标准相关规定。相关技术可以有效解决传统联合收获机在收获过程中作业性能不稳定、适应性差的问题，应用前景广泛。