| 摘要: |
| 针对目前市场上种子分选装置多以传统机械式、半自动式为主，智能化程度不高、分拣准确率较低的问题，基于机器视觉技术设计一种智能种子分选机，主要由传输系统、供料系统、图像采集系统、种子筛选系统和控制系统五大部分构成。以‘郑丹958’玉米种子为研究对象，以种子气流下斜抛的水平距离和传送带传输速度为寻优条件进行试验，确定最优气泵压力值和控制器脉冲频率。对960个种子样本（优质、劣质种子分别为824和136 个）通过目标检测模型进行质量识别，判别种子质量，对种子状态进行标注框选和坐标记录。使用PLC（SIMATIC S7-200 CN，CPU224XP）分选模块控制直动式电磁阀组，对种子进行分选试验。结果表明：1）最优组合气泵压力值为0.3 MPa，控制器脉冲频率为3 175 Hz。2）优质、劣质种子识别率分别为93.69%和91.91%，种子分选率为89.6%。该分选机能够有效满足种子分选要求。 |
| 关键词: 机器视觉 分选机 种子 线扫描相机 PLC |
| DOI：10.11841/j.issn.1007-4333.2024.08.17 |
| 投稿时间：2023-11-08 |
| 基金项目:国家大宗蔬菜产业技术体系（CARS-23）；河北省重点研发计划项目（20327403D）；河北省省属高校基本科研业务费项目（KY2022020） |
|
| Design of an intelligent seed sorting machine based on machine vision |
|
FENG Yubo, ZHAO Xiaoshun*, FAN Xiaofei, TIAN Ruitao, LIU Jingyan, SUN Lei
|
| (College of Mechanical and Electrical Engineering， Hebei Agricultural University， Baoding 071000， China) |
| Abstract: |
| Current seed sorting machines on the market are mostly traditional mechanical or semi-automatic type， which have low automation level and low sorting accuracy. In view of the problem， an intelligent seed sorting machine was designed based on machine vision technology. It is composed of transmission system， seed-feeding system， image acquisition system， seed-screening system and control system. Taking the ‘Zheng Dan 958’ corn seeds as the research object， the horizontal distance of the seed flow and the transmission speed of the conveyor belt is tested as the preferential conditions to determine the pressure value of the optimal air pump and the controller pulse frequency. A total of 960 seed samples （The number of high-quality and inferior seeds was 824 and 136， respectively.） are identified by the target detection model. The quality of the seeds was evaluated， and the seed state is marked and coordinate recorded. The PLC （SIMATIC S7-200 CN， CPU224XP） sorting module was used to control the direct-action solenoid valve set to perform sorting tests on the seeds. The results show that： 1） The pressure value of the optimal combination air pump is 0.3 MPa， and the controller pulse frequency is 3 175 Hz. 2） The recognition rate of high quality and inferior seeds was 93.69% and 91.91%， respectively， and the seed sorting rate was 89.6%. In conclusion， the machine can effectively meet the seed sorting requirements. |
| Key words: machine vision seed sorting machine seed line scan camera PLC |
