作者：靳萍, 李红志, 王磊

作者单位：国家海洋技术中心，天津 300112

关键词：感应传输CTD;随机森林;CEEMD;FFT谱分析;数据降噪

摘要：

鉴于感应传输温盐深测量仪（conductivity-temperature-depth profiler, CTD）海上试验测量误差的问题，考虑到温盐深各参数非线性非平稳变化的特点，提出基于时频分析的CTD数据降噪方法，即通过随机森林回归的鲁棒性和CEEMD（complete ensemble empirical mode decomposition）的自适应性，提取CTD高频谐波振荡，由FFT谱特点对高频成分细化，将噪声部分识别与去除。仿真结果表明，降噪后信噪比提升约9 dB，相关性99%以上；同时将该方法应用于威海锚泊平台感应传输CTD实测数据中，通过盐度检验的方法对降噪结果分析，试验结果表明：进口和国产两型CTD盐度的测量准确性提升1.3倍，降噪后的残差通过高斯白噪声的显著性检验，说明该方法有效、可靠。

Research of data denoising method for induction transmission CTD based on time-frequency analysis
JIN Ping, LI Hongzhi, WANG Lei
National Ocean Technology Center, Tianjin 300112 China
Abstract: In view of the measurement error of the CTD(conductivity-temperature-depth profiler) in the sea test, considering the non-linear and non-stationary characteristics of the parameters of temperature, salinity and depth, a noise reduction method of CTD data based on time-fequency analysis was proposed, that was, through the robustness of random forest regression and the self adaptability of CEEMD(complete ensemble empirical mode decomposition), the high frequency harmonic oscillations of CTD were extracted, and according to the FFT spectrum characteristics, the high frequency components were further refined to identify and remove the noise part. The simulation results showed that the signal-to-noise ratio was improved by about 9 dB and the correlation was more than 99% after noise reduction. At the same time, the method was applied to the measured data of inductive transmission CTD of Weihai mooring platform, and the noise reduction results were analyzed by means of salinity test. The test results showed that the salinity measurement accuracy between the imported and the domestic CTD was increased by 1.3 times, and the residual error after noise reduction passed the significance test of white Gaussian noise, which have proved the effectiveness and reliability of the method.
Keywords: inductive transmission CTD;random forest;CEEMD;FFT spectrum analysis;data denoising
2021, 47(5):24-32,57  收稿日期: 2021-01-11;收到修改稿日期: 2021-02-23
基金项目: 国家重点研发计划（2017YFC1403403，2016YFC1400502，2016YFC1400401）；广东省重点研发计划（2020B1111020001）
作者简介: 靳萍(1994-),女,山西大同市人,硕士研究生,专业方向为海洋观测技术
参考文献
[1] BOND N A, CRONIN M F, SABINE C, et al. Upper ocean response to Typhoon Choi-Wan as measured by the Kuroshio Extension Observatory mooring[J]. Journal of Geophysical Research, 2011, 116(C2): 1-8
[2] 黄谟涛, 翟国君, 欧阳永忠, 等. 海洋测量误差处理技术研究进展与展望[C]//海洋测绘综合性学术研讨会. 中国测绘学会, 2003: 12-19.
[3] 刘雪堂, 林红军. CTD剖面仪现场定标和误差分析[J]. 海洋技术, 1990, 9(1): 18-24
[4] SCOR Working Group 51. The acquisition, calibration, and analysis of CTD data[J]. UNESCO Technical Papers in Marine Science, 1988, 54(8): 92
[5] 许建平. 温盐深剖面仪资料的校正和处理技术[J]. 海洋技术, 1987, 6(4): 39-48
[6] 马继瑞. CTD资料处理的主要步骤和方法(续)[J]. 南海研究与开发, 1989(2): 70-79
[7] 梁广建, 徐刚, 龚建忠. 海洋调查水文资料异常值控制探讨[J]. 海洋测绘, 2002, 22(6): 53-55
[8] 刘首华, 陈满春, 董明媚, 等. 一种实用海洋浮标数据异常值质控方法[J]. 海洋通报, 2016, 35(3): 264-270
[9] UCHIDA H, SHIMADA K, KAWANO T. A method for data processing to obtain high-quality XCTD data[J]. Journal of Atmospheric and Oceanic Technology, 2011, 28(6): 816-826
[10] 郑一, 岳军, 纪乃华, 等. 基于PAADCP的海流流速的误差分析及校正方法[J]. 海洋科学, 2013, 37(5): 81-86
[11] 徐波, 黎会鹏, 周凤星, 等. 基于自适应CEEMD的非平稳信号分析方法[J]. 振动. 测试与诊断, 2020, 40(1): 54-61+203
[12] 陈志恒, 罗文斌, 常俊杰, 等. 基于EMD的神经网络空耦超声储油罐液位检测[J]. 中国测试, 2021, 47(1): 9-14
[13] 邓高峰, 赵震宇, 王珺, 等. 基于改进自编码器和随机森林的窃电检测方法[J]. 中国测试, 2020, 46(7): 83-89
[14] 胡清华, 汪运. 考虑数据噪声的鲁棒回归建模方法综述[J]. 西北大学学报(自然科学版), 2019, 49(4): 496-507
[15] 魏永星, 于金花, 李琦, 等. 实测海洋环境噪声数据谱级特性研究[J]. 海洋技术学报, 2016, 35(3): 36-39