A Knowledg-based Approach To The Satisfactory Solution In Optikization And Its Applications

Based on the satisfactory principle in de cision making proposed by Simon, this paper presents a now knowleage-based approach to the satisfactory solution in optimization problems. Both operations research algorithms and knowledge engineering technology are employed and supplemented with each other in this approach. In order to develop this new approach an essen- tial theorem, i.e., dividing optimization principle, and a new concept, i.e., turning region, have been proposed. From the methodology point of view the former is similar to the Bellman's optimization principle in dynamic programming, and the latter is the generalization of extreme point. The major steps of the knowledge-based approach to the satisfactory solution for linearly constrained optimization problems are as follows: Give an initial dividing hyperplane and move it in one direction step by step to find the turning region, which then should be compressed till the criterion of the satisfactory solution satisfies. Finally ,a numerical example in industrial systems illustrates that the proposed approach can effectively provide the satisfactory solution with less computer time.