Authors

Abstract

To address the challenges of unstable planting spacing and susceptibility to operational fluctuations during the operation of electric-driven potato planters, an electric-driven spacing control system based on a BP neural network–PID controller was designed. The system uses actual output data from the rotational speed sensor, analog voltage signals from the control terminal, and control error information as inputs. By leveraging the online learning and adaptive tuning capabilities of the BP neural network, PID parameters are dynamically generated and optimized for real-time operating conditions, thereby achieving precise speed control of the seeding actuator. By integrating the structural design of the electric-driven potato planter with the spacing control mechanism, a mathematical model of the brushless DC motor and transmission system was established. Based on this model, a BP neural-network-based PID control strategy was developed. A MATLAB/Simulink simulation platform was constructed for comparative validation. Compared with a traditional PID controller tuned by empirical trial-and-error, the proposed method demonstrated superior control performance. The traditional PID exhibited approximately 10%–15% overshoot with oscillations during step response, whereas the neural network PID maintained a comparable rise time with negligible overshoot and a smoother response. Finally, both control algorithms were deployed on prototype machines for field trials to validate their effectiveness and engineering applicability under real-world conditions. Field test results indicated that under neural network control, the maximum row spacing error was 1.1 cm, with an average absolute relative error of 2.7%, meeting the row spacing accuracy requirements for electric-driven potato planters. These findings provide a theoretical basis and practical reference for the design of row spacing control systems in potato planters.

Abstract in Chinese

针对电驱马铃薯播种机作业过程中播种株距难以稳定、易受工况波动影响的问题，本文设计了一种基于BP神经网络PID的电驱株距控制系统。该系统以转速传感器反馈的实际输出、控制端模拟量电压以及控制误差等信息为输入，利用BP神经网络在线学习与自适应整定，实时给出更匹配工况的PID参数，从而实现对排种执行机构的精细调速控制。结合电驱马铃薯播种机的结构方案与株距控制机理，建立直流无刷电机及传动环节的数学模型，并在此基础上构建BP神经网络PID控制策略。进一步搭建Matlab/Simulink仿真平台开展对比验证：与采用经验试凑法整定的传统PID相比，所提方法控制性能更优；传统PID在阶跃响应中出现约10%到15%的超调并伴随一定振荡，而神经网络PID在基本保持上升速度的同时几乎无超调、响应更为平稳。最后，将两种控制算法分别部署于样机并开展田间试验，以进一步验证其在实际作业条件下的有效性与工程适用性。田间试验结果表明，神经网络控制下最大株距误差为1.1cm，平均绝对相对误差为2.7%，满足电驱马铃薯播种株距精度要求，研究成果可为马铃薯播种机株距控制系统提供理论依据与实践参考。