SmartDO

Using SmartDO's Fusion Technology of AI & Multiphysics Automatic Design Optimization Technology to Build Accurate Digital Twin

With 5G communication technology advancing into high-frequency Ka bands, improving signal transmission efficiency and coverage has become a key challenge. This article demonstrates how to use SmartDO's fusion technology of AI & multiphysics automatic design optimization technology to establish an accurate Digital Twin based on existing reflector CAE model. It calibrates the 5G frequency-selective reflector CAE simulation model against physical test data automatically.

Figure 1 shows the 3D electromagnetic field distribution of the 5G frequency-selective reflector CAE model.

Figure 1 3D Electromagnetic Field Distribution of the Reflector CAE Model

Figure 2 shows the hardware test setup and the physical appearance of the 5G frequency-selective reflector.

Figure 2 Hardware Test Setup and the Physical Appearance of the 5G Reflector

Background of 5G Frequency-Selective Reflector Design

The core goal of the 5G frequency-selective reflector is to provide high reflectivity at the main frequency (28±0.3 GHz), while suppressing reflection outside the main band to avoid interference.
Key design considerations include:

• Adjustment of geometric pattern (Metal pattern) parameters: P, L, W, UT, LT
• Material characteristics (Dk, Df) across multiple frequency points
• Incident and reflection angle settings (θ = 45° used in this study)

Using SmartDO's Fusion Technology of AI & Multiphysics Automatic Design Optimization Technology to Build Accurate Digital Twin


Digital Twin technology enables complete virtualization of physical reflectors, creating a digital model for synchronized simulation and measurement comparison. The process includes:
• Using CAD and CAE for high frequency electromagnetic software to build a complete geometric model of the reflector
• Importing measured data to establish physical-digital data mapping
• Establishing the mapping between model parameters (geometry and material constants) and reflection spectrum
• Integrating SmartDO for optimization tuning and fitting

Figure 3 shows the SmartDO control panel, presenting the process flow to help users clearly understand the execution relationships and procedures.



Figure 3 SmartDO Control panel, Presenting the Process Flow of Setting and Optimization



SmartDO Digital Twin Parameter Optimization Architecture


SmartDO’s fusion technology of AI & multiphysics automatic design optimization technology accelerates multi-variable parameter space search and matching, especially when multiple frequency points must simultaneously satisfy multiple constraint conditions. This study adopts the following method:
• Use SmartDO as the optimization core to build an objective function that minimizes the error between the measured and simulated curves
• Use Dk/Df values at 4 different frequencies as design variables - totaling 8 design variables (DVs)
• Set 18 constraint conditions across 9 frequency points
• Apply SmartDO's automatic optimization + AI algorithm to explore Dk and Df tuning across frequencies for the Digital Twin model

Figure 4 shows the comparison between measured and simulated results before and after SmartDO optimization. According to the results, the frequency-selective reflector design achieves optimal Return Loss (-0.42 dB) near the main frequency of 28.0 GHz, and also shows good reflectivity at 25.13 GHz and 30.16 GHz. After fitting the model with SmartDO's fusion technology of AI & multiphysics automatic design optimization technology, the error between measured and simulated data is significantly reduced.


Figure 4 The Comparison Between Measured and Simulated Results Before and After SmartDO Optimization



Advantages of SmartDO Digital Twin Auto-Tuning Optimization

Implementing SmartDO’s fusion technology of AI & multiphysics automatic design optimization technology to calibrate Digital Twin offers the following advantages over traditional trial-and-error methods:

• One-page GUI configuration
• One-click automatic optimization design
• Proved in optimization problems with more than 200 design variables
• Multiphysics automatic design optimization

For more information about SmartDO, please contact us.