3.灵敏度辩析.ppt

第七节 灵敏度分析 * * 一、定义 其单纯形表如一下所示 1 0.5 -2 0 0 4 x5 0 0 0 x5 0 -0.125 0.5 1 0 2 x2 3 0.25 x4 0 0 0 1 4 x1 2 x3 x2 x1 b xB CB 0 3 2 cj -14 0 0 -1.5 -0.125 0 0 1 0 0 4 16 x4 0 1 0 x5 0 0 0 4 0 12 x5 0 0 x4 0 1 2 1 8 x3 0 x3 x2 x1 b xB CB 0 3 2 cj σ 2 3 0 0 0 初始单纯形表 σ 0 1 0 x6 0 0 0 2 2 3 1 30 x6 0 0 1 x5 0 1 2 5 8 3 150 x7 0 1 x4 7 0 x7 0 6 4 8 120 x5 0 x3 x2 x1 b xB CB 1 5 4 cj σ 4 5 1 7 0 0 0 例 2: 初始单纯形表 -13/15 8/15 -1/15 x6 0 0 -1/15 1 2/3 4/3 0 8 x4 7 -4/15 2/15 x5 0 1 0 5/3 13/3 0 92 x7 0 0 x4 7 0 x7 0 2/3 1/3 1 14 x1 4 x3 x2 x1 b xB cB 1 5 4 cj σ 0 -17/3 -19/3 0 -1/15 -52/15 0 1 -1 2 1 0 2 x2 3 -3 3 x4 0 -1 -1 x5 0 -3 0 0 -8 σ -1 0 1 1 x1 2 x3 x2 x1 b xB cB 1 3 2 cj 1 -1 2 1 0 2 x2 3 -1 -1 x5 0 -3 3 x4 0 -3 0 0 σ -1 0 1 1 x1 2 x3 x2 x1 b xB cB 1 3 2 cj 例4:下面是一张LP问题的最优单纯 形表,观察其基变量、非基变量目标 函数系数的改变对检验数的影响 即， 1 -1 2 1 0 2 x2 3 -1 -1 x5 0 -3 3 x4 0 -3 0 0 σ -1 0 1 1 x1 2 x3 x2 x1 b xB cB 1 3 2 cj 要求xj=0 当σn+1?0,σn+2?0时最优解不变;否则,运用对偶单纯形法迭代求最优解 化非标准表为标准表，若ｂｎ＋１??０，则最优解仍为最优解 若ｂｎ＋１?０，则用对偶单纯形法迭代找出最优解 1 -1 2 1 0 2 x2 3 -1 -1 x5 0 0 -3 3 x4 0 0 0 -3 0 0 -8 σ -1 0 1 1 x1 2 x3 x2 x1 b xB cB 1 3 2 cj 0 1 -1 2 1 0 2 x2 3 -1 -1 -1 x5 1 3 -2 0 0 -1 x6 0 -3 3 x4 0 0 x6 -3 0 0 -8 σ -1 0 1 1 x1 2 x3 x2 x1 b xB cB x6 1 0 2 0 1 0 1 x2 3 0 1 0 x5 0 -1 -3 2 0 0 1 x5 0 -6 0 x4 0 -1 -1 x6 0 -1 0 0 -7 σ 1 0 1 2 x1 2 x3 x2 x1 b xB cB 1 3 2 cj 1 -1 1 1 0 2 x2 3 -1 -1 x5 0 -3 -3 0 0 -8 σ 3 x4 0 -1 0 1 1 x1 2 x3 x2 x1 b xB cB 1 3 2 xj 解: （1） λc3*?-σ3=3 λc4*?-σ4=3 λc5*?-σ5=1 所以-3? λ?1/2 σT=(0,0,-3,-3,-1) λ?3 λ?-3 λ?1/2 C*=(0,0,1,-1,2) 0 1 -1 1 1 0 2 x2 3 -1 0 -1 x5 0 1 0 0 1 1 4 x6 0 -3 3 x4 0 0 0 x6 0 -3 0 0 -8 σ -1 0 1 1 x1 2 x3 x2 x1 b xB cB 1 3