摘要
通过对内点多目标线性规划中采用有效鞍点来产生搜寻方向一文的研究 ,提出了新的仿射尺度内点多目标算法。该方法首先求出每个目标函数在最优规划中所作的贡献率 ,然后解一个单目标线性规划问题 ,其解作为有效鞍点的初始值。在迭代过程中 ,投影单个价值向量 (单个目标函数的系数 )以求出投影方向 ,然后求出这些投影方向及现行迭代点到鞍点的方向对最优解的较真实的贡献率 ,从而获得下一次迭代的方向。在进行下次迭代前 ,求出新的边界点与鞍点对最优解的较真实贡献率 ,以确定新的边界点是否取代鞍点。新算法组合使用了层次分析法 (AHP)中的标权转换法和内点法 ,将 q个有效鞍点减少为一个。实例计算表明 。
According to the efficient anchoring point for generating search directions in interior multiobjective linear programming(MOLP), the paper presented the new affine-scaling interior MOLP algorithm. First the contribution rate of each objective function to optimal solution was found as the initial value, then a single - objective linear programming problem was solved and got an efficient anchor point with which to start the iterative process. During the iterative process, in light of the projected individual cost vectors (the coefficient of objective function) , we can get the projected gradients and real contribution rates of them to optimal solution for the anchor point. Using these contribution rates, we generate a combined step direction along which the next step is taken. Before proceeding the next iteration, use the scale - weight transformation method in AHP to confirm whether anchor point is always efficient. Scale - weight transformation method in AHP and affine- scaling interior points algorithm reduced the number of efficient anchor points to one. The computing examples proved that this method is feasible.
出处
《西南石油学院学报》
CSCD
2000年第4期80-83,共4页
Journal of Southwest Petroleum Institute
关键词
多目标线性规划(MOLP)
有效鞍点
层次分析法(AHP)
标权转换法
仿射尺度内点法
multiobjective linear programming
efficient anchor points
analytic hierarchy process
scale-weight transformation method
affine-scaling interior point algorithm
