Space Mapping Optimization using Neural Networks for Efficient Parameter Estimation

Abstract: This project focuses on optimizing input parameters of a partial derivative function of a fine model using Neural network-based Space Mapping Optimization (SMO). The fine model is known for its high accuracy but is computationally expensive. On the other hand, the coarse model is represented by a neural network, which is much faster but less accurate. The SMO approach is applied to bridge the gap between these two models and estimate the optimal input parameters for the fine model. Additionally, this project involves a comprehensive review of previously available Neuro Modeling Space Mapping techniques, which are also used in this project to enhance the optimization process. By utilizing SMO with a neural network-based coarse model, we aim to demonstrate the effectiveness of this method in optimizing complex functions efficiently. The proposed approach of using Neural Network based Space Mapping offers a promising solution to this optimization problem.