Staircase Re-sampling Technique as an Independent Strategy for Noisy Optimization Problems

Abstract

Engineering problems generally based on collecting and processing of the data taken from real-life problems. Even the data represents the real-word data quantitative and qualitative properties, the data contains noise -measurement noise or disturbance-. To get an accurate solution for this noisy optimization problem some additional techniques are formed inside the optimization algorithm to handle the noise. For this reason, in this research staircase-like dynamic re-sampling method is proposed. This technique is based on increase the amount of re-calculation of the objective functions as the generations goes on. The impact of this technique on the benchmark problems are empirically demonstrated by using four different optimization algorithms and proposed dynamic re-sampling method is compared with the static re-sampling method. The results show that the impact of the proposed algorithm increases as the problem becomes more complex and harder to solve with respect to the computational resources.