摘要
高空气象探测技术的发展与探测仪器的进步密不可分。气压传感器作为探空仪搭载的核心器件,在高海拔极端环境下受到制作工艺及温度波动等因素的影响,会出现测量数据精度受损的现象。针对上述问题,通过分析气压传感器的标准气压值和误差之间的关系,推导了误差波动描述算式,把气压传感器标校任务转化为非线性回归方程系数的拟合任务,并设计一种用于误差系数拟合的RBF模型。将拟合后的系数结合误差波动算式来计算最终的标校气压值,以达到提高气压传感器标校精度的目的。实验结果表明:RBF神经网络拟合系数获取的标校值与传统的BP神经网络相比,可使测量误差的偏差度≤0.1%,有效提高了高空气象探测中的气压传感器的测量精度。
The development of high-altitude meteorological detection technology is inseparable from the progress of detection instruments.As the core component of the radiosonde,the air pressure sensor is very likely to be affected by the factors such as manufacturing process and temperature fluctuations in extreme environments at high-altitude so that the accuracy of measurement data can be damaged.In view of the above problems,by analyzing the relations between the standard pressure value and the error of the pressure sensor,this paper deduces the error fluctuation description formula,and transforms the calibration task of the pressure sensor into the fitting task of nonlinear regression equation coefficients,and designs a RBF model for the fitting of error coefficients.The final calibration air pressure value is calculated by the fitted coefficient combined with the error fluctuation formula so as to improve the calibration accuracy of the air pressure sensor.The experimental results show that compared with the traditional BP neural network,the calibration value obtained by the fitting coefficient of the RBF network can reduce the deviation degree of the measurement error to 0.1% or less,which effectively improves the measurement accuracy of air pressure sensor in high-altitude meteorological detection.
作者
刘春玲
王涛
刘鹏宇
Liu Chunling;Wang Tao;Liu Pengyu(Henan Meteorological Service Center,Zhengzhou 450003,China;Faculty of Information Technology of Beijing University of Technology,Beijing 100124,China;Beijing Laboratory of Advanced Information Networks,Beijing 100124,China)
出处
《气象与环境科学》
2023年第3期106-111,共6页
Meteorological and Environmental Sciences
基金
国家重点研发计划(2018YFF01010100)
河南省气象局重点实验室科研基金项目(KM202213)。
关键词
气压传感器
标校
非线性拟合
误差预测
air pressure sensor
calibration
nonlinear fitting
error prediction