Alloys-by-Design Strategies Using Stochastic Multi-Objective Optimization

摘要: Abstract : The objective was to develop and demonstrate a technique for multi-objective optimization of the chemical composition steel alloys with use an existing experimental database. involves organization execution iterative experiment, which results in set Pareto optimum compositions. algorithms response surface building known as IOSO used where surfaces are built accordance information. In experiments information on alloy properties neighborhood is accumulated, makes it possible increase accuracy obtained. After each iteration this technique, new compositions formed assumed be optimal, experiment evaluation should carried out. For work, artificial neural networks were that utilized radial-basis functions modified order build surfaces. modifications consisted selection ANN parameters at stage their training based two criteria: minimal curvature surface, provision best predictive given subset test points. procedure demonstrated work successful efficient