基于处方图的水稻侧深变量施肥控制系统设计与试验

Design and Experiment of Rice Side-deep Variable Rate Fertilization Control System Based on Prescription Diagrams

摘要:

为解决水稻侧深施肥过程中由于缺乏科学处方与智能决策而造成的肥料资源浪费、无法实现空间差异化肥力补给的问题，本文设计了一种基于处方图的水稻侧深变量施肥控制系统。结合排肥理论分析得出变量施肥作业中影响排肥量的可控因素(排肥轴转速、外槽轮工作长度、机组前进速度)，探究施肥控制策略，构建并训练基于神经网络的智能决策模型；利用土壤养分平衡法结合克里金空间插值法在ArcGIS中生成施肥处方图；最后设计水稻侧深变量施肥控制系统并集成到水稻插秧机上，开展性能测试试验。双变量控制模型肥量控制精度试验结果表明，最大排肥误差为3.27%，最小误差为0.06%，总平均误差为1.23%。目标施肥量播量准确性试验结果表明，尿素、磷酸二胺、硫酸钾、掺混复合肥平均播量误差分别为4.60%、4.41%、4.18%和3.66%，平均示量误差分别为5.04%、4.83%、3.81%和4.84%。基于电子处方图的田间试验中系统获取有效定位信息54 867个，定位错误点数6个，均发生在田块边界附近，分区定点变量施肥平均播量误差为4.23%。结果表明集成创制的水稻插秧侧深变量施肥一体机具有较高的播量精度和稳定性，控制系统具有较高的肥量控制精度，利用电子处方图指导施肥作业切实可靠，可实现插秧智能变量施肥一体化作业。

关键词:水稻;侧深变量施肥;处方图;双变量控制;智能决策

Abstract:

To address the issues of fertilizer resource wastage and the inability to achieve spatially differentiated nutrient supply in rice side-deep fertilization due to the lack of scientific prescriptions and intelligent decision-making, a prescription map-based variable-rate fertilization control system for rice side-deep fertilization was designed. Combining fertilizer application theory analysis, controllable factors influencing the fertilizer application rate in variable rate fertilization operations, including fertilizer shaft rotation speed, external groove wheel working length, and machine forward speed were identified to explore fertilization control strategies and construct and train an intelligent decision-making model based on neural networks. Using the soil nutrient balance method combined with the Kriging spatial interpolation method in ArcGIS, a fertilization prescription map was generated. Subsequently, the hardware and software for the rice side-deep variable rate fertilization control system were designed and integrated into a rice transplanter for performance testing. The precision test of the bivariate control model for fertilizer amount showed a maximum application error of 3.27%, a minimum error of 0.06%, and an overall average error of 1.23%. Results from the target fertilization amount accuracy test indicated that the average application rate errors for urea, diammonium phosphate, potassium sulfate, and blended compound fertilizer were 4.60%, 4.41%, 4.18%, and 3.66%, respectively, with average dosage errors of 5.04%, 4.83%, 3.81%, and 4.84%. In field tests guided by electronic prescription maps, the system obtained 54 867 valid positioning points with six positioning errors, all occurring near field boundaries, the average broadcasting amount error for partitioned and fixed-point variable fertilization was 4.23%. The results demonstrated that the integrated rice transplanter with side-deep variable rate fertilization capabilities had high application rate accuracy and stability, and the control system exhibited high precision in fertilizer amount control. Using electronic prescription maps to guide fertilization operations was practical and reliable, enabling integrated intelligent variable-rate fertilization during rice transplanting.

Key words:rice;side-deep variable rate fertilization;prescription diagram;bivariate control;intelligent decision-making