Geometry prediction of the antenna design using machine learning method

Geometry prediction

1. GEOMETRY PREDICTION OF THE ANTENNA DESIGN USING MACHINE LEARNING METHOD Agusriandi Eko Setijadi, S.T., M.T., Ph.D. Dr. Ir. Achmad Affandi, DEA [email protected] INDONESIA, 11 December 2023

2. Outline 01 Background 03 Methodology 04 Result and Discussion 02 Literature Review 05 Conclusion

3. Problems of accuracy and effectiveness Design & Optimization issues Background

4. Literature Review Principle of Predicting The principle of predicting is a method used in antenna design to predict the performance of an antenna system before it is built. This is done by using mathematical models and simulations to predict the antenna's radiation pattern, gain, and other important parameters. Applications in Antenna Design The principle of predicting is used extensively in antenna design for wireless communication systems. By using this method, engineers can design antennas that meet the specific requirements of a particular system, such as frequency range, bandwidth, and gain . This helps to ensure that the antenna system performs optimally and provides reliable communication.

5. THERE IS A NEED FOR NEW METHODS THAT CAN PRODUCE PREDICTIONS FOR THE DESIGN AND OPTIMIZATION OF ANTENNA BY UTILIZING ML

6. Methodology Experiment Dataset Preparation Training Model Testing and Validation Result and Analysis Conclusion

7. Experiment substrate = FR-4 dielectric constant = 4.5 thickness = 1.5 mm Frequency = 13 GHz Width = 6.253 mm Length = 4.788 mm Feed lines = 3.5 mm

8. Dataset Preparation The success of a machine learning model in antenna design depends largely on the quality of the dataset used to train it. Here are some steps to prepare the dataset: Collect data on antenna designs and their corresponding performance metrics, such as gain and radiation pattern. Preprocess the data by normalizing or standardizing the input features and output metrics to ensure they are comparable and have similar ranges. Split the dataset into training, validation, and test sets. The training set is used to train the model, the validation set is used to tune the hyperparameters and prevent overfitting, and the test set is used to evaluate the final performance of the model.

9. Data Collection Collect data on existing antenna designs and their performance characteristics. Data Preprocessing Clean and format the collected data to ensure it is suitable for use in the machine learning model. Feature Selection Identify the most relevant features from the preprocessed data to use in the machine learning model. Training Model

10. Testing and Validation Testing Testing involves measuring the performance of the designed antenna in a controlled environment. This can be done using specialized equipment such as anechoic chambers or network analyzers. The data collected from testing is used to validate the antenna design and make necessary adjustments. Validation Validation involves testing the designed antenna in real-world scenarios to ensure that it meets the required specifications and performs optimally. This can be done using simulation software or by deploying the antenna in a real-world environment. The data collected from validation is used to fine-tune the antenna design and ensure that it performs as expected.

11. Results and Analysis Simulation Results Our simulations show that the antenna designs generated using machine learning algorithms outperform traditional designs by up to 20%. Optimization Results Our machine learning algorithm was able to optimize antenna designs for both size and performance, resulting in smaller and more efficient antennas. Performance Analysis Our analysis of antenna performance data shows that the machine learning-generated designs have a more stable and consistent performance across different operating conditions.

12. Experiment and Simulation Result and Discussion

13. s11 = s paramater sL = substrate length L = Length patch fw = feed width w = width patch The number of data rows collected 243 rows from several times the Sweep parameters Generate Dataset

14. Training Model

15. Comparison of results from experiments and simulations of ML

16. Conclusion

17. Thank You