A Comparison of Regression and Neural Network Based for Multiple Response Optimization in a Real Case Study of Gasoline Production Process

Most of existing researches for multi response optimization are based on regression analysis. however, the artificial neural network can be applied for the problem. in this paper, two approaches are proposed by consideration of both methods. in the first approach, regression model of the controllable factors and s/n (signal to noise) ratio of each response has been achieved, and then a fuzzy programming has been applied to find the optimal factors' levels. in the second approach, a tuned artificial neural network (ann) is used to relate controllable factors and overall exponential desirability function then genetic algorithm (ga) is used to find factors’ optimum values. mentioned approaches have been discussed in a real case study of oil refining industry. experimental results for the suggested levels confirm efficiency of the both proposed methods; however, the neural network based approach seems to be more suitable for our case study.