Fishery Modernization ›› 2025, Vol. 52 ›› Issue (6): 97-105. doi: 10.26958/j.cnki.1007-9580.2025.06.012
Previous Articles Next Articles
GUO Sanci1，SU Zhenheng1，LIU Benxue2（1 Henan Mechanical and Electrical Vocational College, Zhengzhou 451192, Henan,China；#br# 2 Zhengzhou University school of mechanical and power engineering, Zhengzhou 450001, Henan,China)
郭三刺1，苏振恒1，刘本学2(1 河南机电职业学院机电工程学院，河南 郑州 451192；
2 郑州大学机械与动力工程学院，河南 郑州 450001)
Abstract: Feeding rail cars are indispensable in the feeding link of modern fishery breeding, in order to solve the problems of low accuracy of the motion displacement model of the fish pond feeding railcar and the complex operation of the human-computer interaction system, a railcar drive control system based on automation and machine learning technology was designed. With PLC as the core controller, a DTU is integrated to realize remote control. The whole process of automatic feeding operation was written in LAD language. The motion displacement of the small railcar is recorded by using the proximity switch and RFID locator, and the motion displacement model is fitted by the Lasso Regression Algorithm. Based on HMI language, a human-computer interaction system was developed to arrange information and data such as motion displacement models, operation buttons, and indicator arranged in a 'pin' - shaped layout for intuitive information display. The results showed that the mean square error (MSE) of the fitted motion displacement model was 20.7449, the mean absolute error (MAE) was 3.0849, and the coefficient of determination (R2) was 0.9892. The proportion of users who were satisfied with the evaluation indexes of the human-computer interaction system was more than 90%. The average Frecher distance coefficient of the gear speed change curve of the three full-process positioning feeding operations was 9.289. The drive control system operates stably and reliably, and can better meet the actual production needs.
Key words:
PLC,
fish feeding rail car,
drive control systems,
fréche distance,
lasso regression algorithm
摘要： 投饲轨道车在智能化渔业养殖投饲环节中不可或缺，为解决鱼池投饲轨道车运动位移模型精度低和人机交互系统操作复杂等问题，基于自动化和机器学习技术设计了一种轨道车驱动控制系统。以PLC为核心控制器搭载DTU实现远程控制；采用LAD语言编写了全程自动化投饲作业程序；利用接近开关和RFID读写器感知记录小型轨道车运动位移，运用Lasso回归算法拟合运动位移模型；基于HMI语言开发了人机交互系统，将运动位移模型、操作按钮和指示灯等信息与数据以“品”字形布局。结果显示，拟合运动位移模型的均方误差（MSE）为20.7449、平均绝对误差（MAE）为3.0849和决定系数（R2）为0.9892；用户对人机交互系统评价指标满意以上程度占比均大于90%；3次全程定位投饲作业齿轮转速变化曲线的平均弗雷歇距离系数为9.289。研究表明，鱼池投饲轨道车驱动控制系统运行稳定可靠，可以较好地满足实际生产需求。
