车体称重调簧试验台视觉定位控制系统研究

Research on visual positioning control system of vehicle weighing spring test bench

基于中南大学研制的车体称重调簧试验台硬件设计,摒弃了硬件改动较大的基于位移传感器的闭环控制方案,而是利用现有资源及机器视觉技术代替人眼完成定位控制操作。围绕前端图像采集、图像识别及特征提取、后端半闭环反馈控制3部分搭建基于图像信息的手眼视觉伺服系统模型,提出基于Open CV机器视觉库的参数自适应改进Hough变换算法识别目标特征轮廓,应用TMS320C6748系列嵌入式DSP图像处理平台,提供一套多路视频信号分时定位控制的通用解决方案。根据实测结果,本视觉定位控制系统可以准确识别试验所需3种不同车型的二系簧定位销形状,并且系统的定位控制精度在±2 mm以内,满足系统定位精度要求的±5 mm,相比手动方式缩短定位补偿时间75%,提高试验台自动化程度和试验效率。

The article eliminated the closed-loop control scheme which employs the displacement sensor and applied machine vision technology in position control based on spring test platform of CSU.It focused on front-end image acquisition,image recognition and feature extraction,back-end semi-closed-loop feedback control.A model of eye-in-hand visual servo system was built based on image and the adaptive improved Hough transform algorithm for identifying target feature was put forward based on open CV.TMS320C6748 series embedded DSP image platform was applied to provide general solution of multi-channel video signal timing control.According to results,the visual position control system can identify the shape of secondary pin of five locomotives precisely and the control accuracy is within±2 mm,which meets the system positioning accuracy requirements of±5 mm.Compared to manual way,it shortens 75%of position compensation time and improves the efficiency and automation of test platform.