Direct search for the optimum of a simulated process continuing noise
The objective of this work was to study the optimization of a function whose measurement contains error or noise. A search method developed for noise free systems was applied to a noisy function using minimum variance estimators to improve the accuracy of the functional values. The results of the new search method are compared to those of a modified stochastic approximation method which has good convergence on noisy systems. At moderate noise levels (σ ≤ 0.5), the new method appears to converge faster than the best method reported for the modified stochastic approximation although at higher noise levels (σ > 0.5), this advantage seems to be lost.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.