Comparison of Optimization Techniques for Application to Food Product and Process Development

摘要: The super-simplex optimization was modified by incorporating a quadratic regression subroutine. new algorithm found to be more efficient in converging at the optimum than following algorithms: fractional factorial designs, sequential single-factor search, pattern-search method, Morgan-Deming simplex and super-simplex. efficiency of dependent on the‘fitness equation model equation. Response surface analysis accurate only when boundary covered optimum. When constraint imposed restrict search within boundary, unlike other algorithms including weighted centroid circumvented problem stalling boundary. Since it is not unusual for restricted boundaries, could useful optimizing food processing analysis. technique applied -food research, even higher expected backward stepwise multiple instead subroutine with equations set formulae. To obtain fit better true response surface, all information, graphics, should utilized.