氧气A吸收带偏最小二乘基线拟合方法
发布时间：2018-07-27 14:22
【摘要】：为了研究被测目标在氧气A带的自发射光谱与透射率的关系,根据透射率的计算方法,提出了采用偏最小二乘回归法对氧气A带的光谱进行基线拟合,搭建了实验系统,验证了该方法拟合基线的准确度。首先,以黑体辐射理论为依据,给出氧气A带平均透射率计算方法,以实际被测目标光谱为研究对象,以带外数据为依据,利用偏最小二乘法拟合被测目标在氧气A带的基线。为提高拟合精度,剔除了测量奇异点,并利用基线拟合不确定度来评价偏最小二乘回归法拟合基线的准确度。为了验证该方法的准确性,以卤素灯为光源,在0~130m范围内,获得不同距离处的光谱曲线,以及相同距离不同分辨率下的光谱曲线,将各种曲线分别进行基线拟合,分析各自的标准偏差。结果表明,同一距离处不同分辨率下的平均标准偏差为0.23%,随着分辨率的降低,基线拟合不确定度变小,信噪比增大;基线拟合不确定度还与测试设备的分辨率有关,分辨率越高,带外信息基线拟合不确定度越大,反之,带外基线拟合不确定度越小。
[Abstract]:In order to study the relationship between the self-emission spectrum and transmittance of the target in the oxygen A band, according to the calculation method of transmittance, the partial least square regression method is proposed to fit the baseline of the oxygen A band spectrum, and an experimental system is set up. The accuracy of the method is verified. Firstly, based on the blackbody radiation theory, the calculation method of average transmittance of oxygen A band is given. Taking the actual measured target spectrum as the research object and the out-of-band data as the basis, the partial least square method is used to fit the baseline of the measured target in the oxygen A band. In order to improve the fitting accuracy, the singular points were eliminated, and the accuracy of partial least square regression was evaluated by using the uncertainty of baseline fitting. In order to verify the accuracy of the method, the spectral curves at different distances and the spectral curves at the same distance and different resolution were obtained in the range of 0 ~ 130 m using halogen lamp as the light source. Analyze their respective standard deviations. The results show that the average standard deviation at different resolution at the same distance is 0.23. With the decrease of resolution, the uncertainty of baseline fitting becomes smaller and SNR increases, and the uncertainty of baseline fitting is related to the resolution of test equipment. The higher the resolution, the greater the uncertainty of baseline fitting with out-of-band information, and the less the uncertainty of out-of-band baseline fitting.
【作者单位】： 中北大学山西省光电信息与仪器工程技术研究中心;中国科学院长春光学精密机械与物理研究所应用光学国家重点实验室;
【基金】：国际科技合作项目(2013DFR10150) 国家自然科学基金仪器专项基金(61127015);国家自然科学基金青年科学基金项目(61505180)
【分类号】：O433
本文编号：2148102
[Abstract]:In order to study the relationship between the self-emission spectrum and transmittance of the target in the oxygen A band, according to the calculation method of transmittance, the partial least square regression method is proposed to fit the baseline of the oxygen A band spectrum, and an experimental system is set up. The accuracy of the method is verified. Firstly, based on the blackbody radiation theory, the calculation method of average transmittance of oxygen A band is given. Taking the actual measured target spectrum as the research object and the out-of-band data as the basis, the partial least square method is used to fit the baseline of the measured target in the oxygen A band. In order to improve the fitting accuracy, the singular points were eliminated, and the accuracy of partial least square regression was evaluated by using the uncertainty of baseline fitting. In order to verify the accuracy of the method, the spectral curves at different distances and the spectral curves at the same distance and different resolution were obtained in the range of 0 ~ 130 m using halogen lamp as the light source. Analyze their respective standard deviations. The results show that the average standard deviation at different resolution at the same distance is 0.23. With the decrease of resolution, the uncertainty of baseline fitting becomes smaller and SNR increases, and the uncertainty of baseline fitting is related to the resolution of test equipment. The higher the resolution, the greater the uncertainty of baseline fitting with out-of-band information, and the less the uncertainty of out-of-band baseline fitting.
【作者单位】： 中北大学山西省光电信息与仪器工程技术研究中心;中国科学院长春光学精密机械与物理研究所应用光学国家重点实验室;
【基金】：国际科技合作项目(2013DFR10150) 国家自然科学基金仪器专项基金(61127015);国家自然科学基金青年科学基金项目(61505180)
【分类号】：O433
【相似文献】
相关期刊论文 前1条
1 宫元彬;李思纯;杨德森;时胜国;;利用偏最小二乘回归的噪声源分离量化[J];声学技术;2013年S1期
，本文编号：2148102
本文链接：https://www.wllwen.com/kejilunwen/wulilw/2148102.html
