摘要
针对直拉硅晶体生长引晶流程中生长界面温度无法自动测量和校准的问题,本文提出一种基于光圈图像特征与最小二乘支持向量机相结合的温度模式分类检测方法。以数字相机获取的籽晶熔接处的光圈图像作为输入数据,利用图像处理算法提取光圈特征,并以人工校准产生的分类数据和持续生长的后验数据为训练样本,对最小二乘支持向量机分类模型进行训练。实际生长测试证明,可通过多个分类器的组合使用,将生长界面温度在红外测温仪的基础上校准到满足自动引晶所需要的温度。
In order to solve the problem that the growth interface temperature cannot be automatically measured and calibrated in the process of crystal growth of Czochralski crystal,a temperature pattern classification method based on aperture image feature and least squares support vector machine is proposed.The seed splice aperture image obtained by digital camera is taken as the input data,the aperture feature is extracted by image processing algorithm,and the classification data generated by manual calibration and the posterior data of continuous growth are used as training samples,the least squares support vector machine classification model for training.The actual growth test shows that the temperature of the growth interface can be calibrated on the basis of infrared thermometer to meet the temperature needed for automatic crystallization.
作者
赵跃
王欣
ZHAO Yue;WANG Xin(Crystal Growth Equipement and System Integration Engineering Research Center,Faculty of Automation and Information Engineering,Xi’an University of Technology,Xi’an 710048,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2018年第4期573-578,共6页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金重点项目(61533014)
作者简介
赵跃(1972-),男,西安理工大学自动化与信息工程学院副教授,博士生导师,主要研究方向为信号处理和复杂系统的建模与控制,zhaoyue@xaut.edu.cn;;王欣(1994-),西安理工大学自动化与信息工程学院硕士研究生,主要研究方向为模式识别、机器学习、图像处理,595419317@qq.com。
