广义简约梯度法是简约梯度法的扩展,用于处理含有非线性约束的有约束优化问题,核心思想是用等式约束来减少优化变量的个数。
理论方法
首先考虑优化问题:
min f ( X ) s . t . h j ( X ) ≤ 0 , j = 1 , 2 , . . . , m l k ( X ) = 0 , k = 1 , 2 , . . . , l x i l ≤ x i ≤ x i u , i = 1 , 2 , . . . , n \begin{aligned} \min \quad f(X)\\ \mathrm{ s.t. } \quad h_j(X)\leq0,&j=1,2,...,m\\ l_k(X)=0,&k=1,2,...,l\\ x_i^l\leq x_i\leq x_i^u,&i=1,2,...,n \end{aligned} minf(X)s.t.hj(X)≤0,lk(X)=0,xil≤xi≤xiu,j=1,2,...,mk=1,2,...,li=1,2,...,n
现对不等式约束增加松弛变量。可以描述为:
min f ( X ) s . t . h j ( X ) + x n + j = 0 , j = 1 , 2 , . . . , m h k ( X ) = 0 , k = 1 , 2 , . . . , l x i l ≤ x i ≤ x i u , i = 1 , 2 , . . . , n x n + j ≥ 0 , j = 1 , 2 , . . . , m \begin{aligned} \min \quad f(X)\\ \mathrm{ s.t. } \quad h_j(X)+x_{n+j}=0,&j=1,2,...,m\\ h_k(X)=0,&k=1,2,...,l\\ x_i^l\leq x_i\leq x_i^u,&i=1,2,...,n\\ x_{n+j}\geq 0,&j=1,2,...,m \end{aligned} minf(X)s.t.hj(X)+xn+j=0,hk(X)=0,xil≤xi≤xiu,xn+j≥0,j=1,2,...,mk