潮位相关分析中的最小二乘法研究

Study on least square method for correlation analysis of tidal elevation

在潮位相关分析中,若2个相关变量均存在误差且几率相仿时,应用传统最小二乘法可能获得不甚合理的线性相关结果,所拟合的直线不具备可逆性。文中从直线拟合的数学定义出发,探索潮位相关分析中的最小二乘法问题。初步应用研究表明,文中提出的方法可以有效地解决传统最小二乘法在线性相关分析中出现的不合理现象,使分析结果与实际符合良好,回归方程具有良好的可逆性。

The traditional least square has evident error when it is used to fit line whose two variables both have error and the error is almost similar, sometimes may bring illogical results and the fitted line without the reversibility. Based on the mathematic definition, this paper studies the problem that the least square method applied to correlation analysis of tidal elevation. The elementary applied research makes clear that the present method could solve the illogical phenomenon in the correlation analysis of tidal elevation with the traditional least square method, the present solution well accords with the actual instance and has well reversibility. It could be applied to the scientific studies on the observation data.