以概率积分法为例对泰勒级数展开法的不足进行了深入研究.首先给出了概率积分法的预计参数,分析了根据参数定义和实测资料直接求定法的不足;论述了曲线拟合的基本思想和泰勒级数展开法的迭代步骤,指出泰勒级数展开法具有迭代易失真、收敛速度慢以及计算量大等不足;针对这些不足提出借助Broyden算法的基本思想建立迭代模型,并给出了新模型的计算步骤;最后对改进前、后的算法进行了对比分析,验证了改进后的算法的优越性.
A further study was made on the deficiency of Taylor series expansion method taking Probability Integral Method as an example. Firstly the predicting parameters of Probability Integral Method were introduced, at the same time, analyzing the shortcoming of directly calculating method according to definition of parameter and measured data; the basic idea of curve fitting and the iterative steps of Taylor series expansion method were dissertated, and some shortages of Taylor series expansion method were pointed out, such as iterative distortion and convergence slow and so on ; and therefore improving the algorithm a new iteration model was established based on the Broyden algorithm and its calculation steps were given ; Finally, this paper made a comparative analysis between original and improved, the result shows that the latter is better than before