基于数字孪生技术的气吸滚筒式排种器设计与试验

Design and Experiment of Pneumatic Cylinder Seed-metering Device Based on Digital Twin

摘要:

针对气力式排种器设计依赖度高、周期长的问题，本文提出数字孪生技术应用于排种器设计优化过程。以气吸滚筒式排种器为例，构建了包括需求模型、概念方案模型、结构模型、参数优化模型的设计数字孪生体。首先，基于排种器设计历史数据和实例数据进行需求分析，利用系统建模语言(SysML)建立需求模型;其次，通过形态学矩阵生成多组概念方案，采用层次分析法评价迭代，确定最优方案，完成概念方案模型设计;再次，开展排种器结构设计，采用CFD-DEM耦合法进行结构仿真，以仿真结果为依据评价并迭代优化结构设计，建立结构模型;最后，利用粒子群优化-反向传播（PSO-BP）神经网络预测模型与多目标粒子群优化算法（MOPSO），以低漏播率和重播率为目标，迭代优化排种器吸孔直径、滚筒转速、负压，得到参数优化模型。搭建虚实交互试验台，采集气吸滚筒式排种器物理实体转速、排种结果，计算合格率、漏播率、重播率，对比排种器虚拟数字孪生体与物理实体数据。结果表明，物理实体排种结果与设计数字孪生体预测结果合格率、漏播率、重播率绝对误差分别为2.67%、0.36%、3.03%，验证了气吸滚筒式排种器设计数字孪生体的准确性，为基于数字孪生技术的排种器数字化设计及设计过程优化提供思路。

Abstract:

Aiming at the problems of high dependence and long cycle in the design of pneumatic seed metering device, the application of digital twin technology in the design optimization process of seed metering device was proposed. Taking the air-suction drum seed metering device as an example, the design digital twins, including demand model, conceptual scheme model, structural model and parameter optimization model were constructed. Firstly, the requirement analysis was carried out based on the historical data and instance data of the seed metering device design, and the requirement model was established by using the system modeling language (SysML). Secondly, through the morphological matrix, multiple sets of conceptual schemes were generated, and the analytic hierarchy process was used to evaluate the iteration, determine the optimal scheme, and complete the conceptual scheme model design. Thirdly, the structural design of the seed metering device was carried out, and the CFD-DEM coupling method was used to simulate the structure. Based on the simulation results, the structural design was evaluated and optimized iteratively, and the structural model was established. Finally, the particle swarm optimization-back propagation (PSO-BP) neural network prediction model and multi-objective particle swarm optimization algorithm (MOPSO) were used to iteratively optimize the suction hole diameter, drum speed and negative pressure of the seed metering device with the goal of low leakage rate and replay rate, and the parameter optimization model was obtained. A virtual-actual interaction test bench was built to collect the rotation speed and seeding results of the physical entity of the air-suction drum seed metering device, calculate the qualified rate, missed seeding rate and replay rate, and compare the virtual digital twin and physical entity data of the seed metering device. The results showed that the absolute errors of the qualified rate, leakage rate and replay rate between the physical entity seeding results and the design digital twin prediction results were 2.67%, 0.36% and 3.03%, respectively. The accuracy of the design digital twin of the air-suction drum seed metering device was verified, which provided ideas for the digital design research and design process optimization of the seed metering device based on digital twin technology.