多元非线性多项式智能拟合法及其应用研究

提出基于矩阵法结合遗传算法的多元非线性多项式智能拟合法。该法先将多元非线性多项式转化成多元线性多项式，建立最小二乘法标准矩阵，用遗传算法对多项式的项数、各项类型、次数等进行搜索组合，得到最优的拟合函数式，从而实现智能拟合的目的。用该法拟合了金祖源乱堆填料层压降通用关联图，结果令人满意。智能拟合法与传统的非智能拟合法相比具有可以自动寻求最优多项式、拟合精度高的优点。用Visual Basic6．0编程实现智能拟合法，算法稳定，收敛速度快。

Research on intelligent fitting method for multivariate nonlinear polynomials and its application

The intelligent fitting method for multivariate nonlinear polynomials was presented,which was based on combining matrixmethod with genetic algorithm.The multivariate nonlinear polynomials was firstly transformed to multivariate linear polynomials,andthe least squares standard matrix was established.Then the number of polynomials term,the type and the order of every term weresearched and combined to obtain the optimum fitting function by genetic algorithm,and the purpose of intelligent fitting was realized.The Jinzuyuan's general correlation graph of pressure drop of huddle packed layer was fitted by using the above method,and the resultswere satisfactory.The intelligent fitting method has more advantages than the traditional unintelligent method,such as automatically se-lecting optimum polynomials and higher precision.The intelligent fitting method programmed by Visual Basic 6.0 has good stability andfast convergent speed.