人工智能算法在潮位数据插补中的应用  

Application of artificial intelligence algorithms in tide level data interpolation

作　　者：钱春 QIAN Chun(Shanghai Hydrological Administration,Shanghai 200232,China)

机构地区：[1]上海市水文总站,上海200232

出　　处：《水利水电快报》2025年第3期21-24,共4页Express Water Resources & Hydropower Information

摘　　要：为了提高传统潮位插补方法的计算精度,以黄浦江的4个站点为研究对象,采用了LightGBM(Light Gradient Boosting Machine)人工智能算法构建了潮位序列相关关系模型,并进行了误差分析。结果表明:LightGBM算法能够有效建立输入特征与输出响应之间的复杂非线性关系,适用于潮汐波动传播过程中的内在相关性分析。随着输入特征值维度的增加,模型预测精度在初期快速提高,随后逐渐趋于稳定,且预测误差随着源数据站点与目标数据站点距离的减小而减小,计算得出了误差均方根为0.0169 m的目标站潮位预测插补值序列,验证了该方法在潮位资料插补计算中的有效性。研究成果为潮位插补提供了一种新的方法,有助于提高潮位观测数据的连续性。To improve the accuracy of traditional tidal level interpolation methods,taking four stations on the Huangpu River as the research objects,the LightGBM(Light Gradient Boosting Machine)artificial intelligence algorithm was adopted to construct a correlation model of the tidal level sequence,and an error analysis was carried out.The results showed that the LightGBM algorithm could effectively establish the complex nonlinear relationship between input features and output responses,which was applicable to the analysis of the inherent correlation in the process of tidal fluctuation propagation.As the dimension of input feature values increased,the prediction accuracy of the model improved rapidly in the initial stage and then gradually tended to be stable.Moreover,the prediction error decreased as the distance between the source data station and the target data station decreased.It was calculated that the root mean square error of interpolated sequence of tidal level predictions for the target station was 0.0169 m,verifying the effectiveness of this method in the interpolation calculation of tidal level data.The research results can provide a new method for tidal level interpolation and are of great significance for improving the continuity of tidal level observation data.

关 键 词：水文测验 LightGBM 人工智能算法 潮位插补 黄浦江 

分 类 号：P333.9[天文地球—水文科学]