신경회로망을 이용한 최적화 모형의 통제도구 = UNIK-OPT/NN : Neural Network based Adaptive Optimal Controller on the Optimization Models

다국어 초록 (Multilingual Abstract)

When we apply optimization model to most planning problems, additional information tends to be accumulated since the initial model building and its partial implementation. Particularly, if such information means restrictions to the values on a set of the designated decision variables, it may cause the infeasibility or severe degradation of optimal solution in the optimization model. The only way to overcome this is the coefficients should be adjusted so that the desired values on the infeasible decision variables can be obtained. Let us call such an effort an adaptive optimal control on the optimization model. Since there is no known analytical method to perform such adaptive optimal control on the optimization model, we proposed a neural network approach for the adaptive optimal control and validated with the scheduling problems in a refinery plant. To help the adaptive optimal control procedure, we develop a tool UNIK-OPT/NN which integrates the neural network model with the semantically represented linear, programming model that are generated by the knowledge-assisted optimization modeler UNIK-OPT. UNIK-OPT/NN is also applied to the refinery case and it is shown that the user can quickly and easily develop an adaptive optimal control model by using UNIK-OPT/NN.