Multi-Fidelity Local Surrogate Model for Computationally Efficient Microwave Component Design Optimization.

Area of Science:

• Engineering
• Computational Electromagnetics
• Optimization Techniques

Background:

• High-fidelity models in optimization are computationally expensive.
• Existing surrogate-model methods may lack efficiency or accuracy.
• Reducing computational cost while maintaining accuracy is crucial for complex designs.

Purpose of the Study:

• To propose a robust and computationally efficient multi-fidelity local surrogate-model optimization method.
• To minimize the number of high-fidelity model evaluations.
• To enhance optimization speed and improve optimal solution accuracy.

Main Methods:

• Utilizes multi-fidelity coarse models and polynomial interpolation for local surrogate modeling.
• Employs iterative local region modeling and optimization with a judgment factor for region size updates.
• Refines the final local surrogate model using space mapping techniques for high accuracy.

Main Results:

• Demonstrated effectiveness in designing a bandpass filter and a compact ultra-wide-band (UWB) multiple-in multiple-out (MIMO) antenna.
• Optimized designs met the specifications of the high-fidelity model.
• Achieved better convergence and substantial reduction in computational cost compared to existing methods.

Conclusions:

• The proposed multi-fidelity method offers a significant improvement over traditional optimization techniques.
• It provides a computationally efficient and accurate approach for complex engineering designs.
• The method effectively balances computational cost with the accuracy of the final optimized design.