摘要
目的:针对变压器用油面温控器校准时间长,示值与开关动作参数要分开校准,人工读数准确性不高等问题,开发了基于深度学习的变压器用油面温控器全自动校准系统。方法:通过程序控制恒温槽产生校准需要的温度场以及固定升降温速率曲线。利用标准表结合二等铂电阻采集标准温度。利用多路转换开关实时采集油面温控器开关信号,实现自动记录开关动作状态。采用基于深度学习的指针式仪表自动识别读数方法,实现对于油面温控器仪表表盘的自动读数识别。该系统可根据相关检定规程,对油面温控器测温头的示值误差、两表偏差以及温度开关的接点动作误差同时进行校准。结果:实验测试其指针表盘识别准确率为94.6%,达到人工读数的水平。结论:该系统具有较高的准确性以及可靠性,可以提高油面温控器的校准效率。
Aims:Aiming at the time-consuming problem,and the low accuracy of manual readings for transformer oil surface thermostats,an automatic calibration system for transformer oil surface thermostats based on deep learning was developed.Methods:The temperature field required for calibration and the fixed temperature rise-and-fall rate curves were generated by the program control thermostat.The standard temperature was collected by the standard thermometer combined with the second-class platinum resistance.The switch signal of the oil surface thermostat was collected real-timely by the multiplexer switch to the automatic recording of the switch actionstate.The deep learning-based automatic identification reading method of the pointer instrument was adopted to realize the automatic reading identification of the instrument panel of the oil surface thermostat.The system could simultaneously calibrate the indication error of the temperature measuring head of the oil surface thermostat.The deviation of the two meters and the contact action error of the temperature switched according to the relevant verification procedures.Results:The accuracy of the pointer dial recognition was 94.6%,reaching the level of manual reading.Conclusions:The system has high accuracy and reliability,which can improve the calibration efficiency of the oil surface thermostat.
作者
黄嘉旻
孙斌
孙超平
梅怡嘉
赵玉晓
HUANG Jiamin;SUN Bin;SUN Chaoping;MEI Yijia;ZHAO Yuxiao(College of Metrology and Measurement Engineering,China Jiliang University,Hangzhou 310018,China;Huzhou Institute of Quality Supervision and Inspection,Huzhou 313099,China)
出处
《中国计量大学学报》
2023年第4期501-510,共10页
Journal of China University of Metrology
基金
国家自然科学基金项目(No.51475440)。
关键词
油面温控器校准
深度学习
两表偏差
温度开关
oil surface thermostat calibration
deep learning
two-meter deviation
temperature switch
作者简介
通信作者:孙斌(1972-),男,教授,博士,主要研究方向为热工参数检测与自动校准。E-mail:bsun555@cjlu.edu.cn。