摘要
针对传统的贴片机对于阻容等材料的位置和PCB的器件布局强限制的问题。提出了一种对贴片机的整体结构进行改装的新做法。即CRNN来进行字符的识别,使用卷积对字符的特征进行提取,再用RNN或者LSTM类的神经网络对序列进行预测,得出可能的字符。将所获取的阻容的位置和数值信息通过串口通信传到控制的四维模组平台的单片机上。单片机进行定位PCB上的元件位置并自动放料。有助于提高科研样品试制或研发小批量的工作效率。
Aiming at the problem that the traditional placement machine has strong restrictions on the position of the material such as the RC and the device layout of the PCB. A new approach to the modification of the overall structure of the placement machine is proposed. That is, CRNN is used to identify characters, and the features of the characters are extracted using convolution, and the sequence is predicted by the neural network of RNN or LSTM to obtain possible characters. The position and numerical information of the obtained resistance capacity are transmitted to the single-chip microcomputer of the controlled four-dimensional module platform through serial communication. The MCU performs positioning of the components on the PCB and automatically discharges the material. It can help improve the productivity of scientific research samples or research and development of small batches.
作者
任天成
金博闻
俞月伦
黄继业
Ren Tiancheng;Jin Bowen;Yu Yuelun;Huang Jiye(Hangzhou Dianzi University,Electronic Information,Hangzhou Zhejiang,310018)
出处
《电子测试》
2019年第15期135-138,共4页
Electronic Test
基金
2018年国家级大学生创新创业训练计划(201810336012)
关键词
阻容分类
CRNN
自动定位
PCB焊接
4维模组
Resistor volume classification
CRNN
automatic positioning
PCB soldering
4D module
