计算方法实验报告92.docxVIP

计算方法上机实习报告实验名称：曲线拟合问题的数值实验班级姓名：新能源科学与工程（9121086701），李旨敬（912108670122）摘要本次实验运用最小二乘法思想利用计算机程序进行实验，并对结果进行讨论。问题对实验数据（xi，yi），i=1，2，…，N，要求：利用最小二乘法原理，编制确定拟合多项式Pn（x）=a0+a1x+a2x2+…+anxn，n《N，的数值软件，要求在软件中能任意令拟合多项式Pn（x）中第p项系数为0，并自动调整相应正规方程组。（2）已知实验数据如下表xyxy51.0029300.9979101.0023350.9978151.0000400.9981200.9990450.9987250.9983500.9996按照上表，拟合多项式Pn（x）=a0+a1x+a2x2+…+anxn并分别对n=2，3，4，5，6进行数值计算，最后根据各自偏差的平方和大小分析拟合效果的好坏。数学模型根据最小二乘法思想，拟合多项式各点值与实际值的差的平方和越小，拟合效果越好，此题中，令权系数（反映该点数据的重要程度）ω=1，根据这个思想得到的正规方程组为然后在解此方程组,可以借用高斯消元法进行求解,求解后,并进行偏差平方和的计算。可利用如下的公式。偏差平方和=利用上正规方程组可使此项最小。程序结构流程图符号意义标志符数学符号类型形式作用nN整型1方程组的最大次数aA双精度N系数矩阵bB双精度N方程组的右端常数项xxX双精度NxN求和项wcδ双精度N偏差平方程序代码Private Sub Command1_Click()List1.ClearList2.ClearList3.ClearList4.ClearDim x(1 To 10) As Single, y(1 To 10) As SingleDim n As Integer, i As Integer, j As Integer, i_tem1 As String, i_tem2 As String, i_tem3 As String, k As IntegerDim a() As Double, b() As Double, xx() As Double, m() As DoubleFor i = 1 To 10x(i) = 5 * iNexty(1) = 1.0029: y(2) = 1.0023: y(3) = 1: y(4) = 0.999: y(5) = 0.9983: y(6) = 0.9979: y(7) = 0.9978: y(8) = 0.9981: y(9) = 0.9987: y(10) = 0.9996n = Combo1.List(Combo1.ListIndex)ReDima(1 To n + 1), b(1 To n + 1), xx(1 To n + 1, 1 To n + 1), m(1 To n + 1, 1 To n + 1)For i = 1 To n + 1 i_tem1 = For k = 1 To 10 b(i) = b(i) + (x(k) ^ (i - 1)) * y(k)‘求方程右边常数矩阵 Next i_tem2 = Format(b(i), 0.0000) For j = 1 To n + 1 For k = 1 To 10 xx(i, j) = xx(i, j) + x(k) ^ (i + j - 2)‘求正规方程组的求和项组成的矩阵 Next i_tem1 = i_tem1 Space(4) Format(xx(i, j), 0.0000) Next List1.AddItem i_tem1 List1.AddItem List2.AddItem i_tem2 List2.AddItem Next‘此循环是为了求解正规方程组的求和项组成的矩阵和右方常数矩阵For k = 1 To nFor i = k + 1 To n + 1m(i, k) = xx(i, k) / xx(k, k) For j = k To n + 1xx(i, j) = xx(i, j) - m(i, k) * xx(k, j) Nextb(i) = b(i) - b(k) * m(i, k)NextNext‘利用高斯消元法求解a(n + 1) = b(n + 1) / xx(n + 1, n + 1)Dim sum As DoubleFor i = n To 1 Step -1sum = 0 For k = i + 1 To n + 1 Step 1sum = sum + xx(i, k) * a(k) Nexta(i) = (b(i) - sum) / xx(i, i)Next‘此循环求解拟合多