逐次最小二乘在高精度曲面建模方法(HASM)中的应用

Application of Sequential Least Square in HASM Technique

  • 摘要: 从高精度曲面建模方法(HASM)曲面方程出发,结合数据平差理论,建立了基于独立单元的计算模型,并且采用逐次最小二乘法对计算方程进行分组求解。对HASM-AD和几种经典方法的精度进行了交叉验证,结果表明,该算法模拟误差的均方根差远小于其他空间插值方法。

     

    Abstract: High accuracy surface modeling(HASM) technique is designed to improve accuracy in surface fitting,based on the surface theory.However,the constraints imperfection in the model led to deviation and the integral iterative algorithm adopted in HASM technique resulted in high computation complexity and huge memory usage usually exceeding the PC capacity,so that it became difficult to put HASM technique into application.So an innovative model is put forward based on the sequential least square in computation on the basis of independent units.The technique applies sequential division to computation,which can divide equations into groups to process in sequence.Cross-validation has indicated that HASM-AD technique excelled other spatial interpolation techniques in accuracy.Simultaneously,the studies has demonstrated that the HASM-AD algorithm has prominently reduced the computation complexity,relieved memory usage and effectively broken up the restriction from the extent of data sets on computation.

     

/

返回文章
返回