search

youxch / Inverse-design-of-patch-antennas

This repository hosts a simple demonstration of a deep learning approach for the inverse design of patch antennas. The goal is to explore energy-efficient designs and to significantly reduce simulation cost compared to conventional methods.
MIT License
64 stars 9 forks source link

A simulation file validation issue. #5

Open youxch opened 7 months ago

youxch commented 7 months ago

hair-an A set of data was randomly selected and subjected to full wave simulation. But the expected results of the model were not obtained, and I don't know where the problem lies. Here's my object built by CST2022. I appreciate it if you could take your time to review the simulation file and see if you can identify the problem. The data I choose is from file"Output_Max.txt",line 2. 7.000 71.000 40.000 -6.500 8.000 9.000 2.400 3.415 1015.000 9.002 By calculation, the actual value should be l= 71.0 w= 40.0 wu1= 2.0 wu2= 32.0 wu4= 6.0 lu1= 3.87 wu3= 2.4 lu2= 5.81 wf= -6.5 h= 7 fL=2400MHz, fH=3415MHz Gain=9.002dBi You can get my file by clicking this URL https://github.com/hair-an/CST-repetition-file-of-Inverse-design-of-patch-antennas

youxch commented 7 months ago

Thank you for your feedback. I have conducted a simulation validation using the parameters you provided, and the actual values indeed match your calculations. Below are the model structure, reflection coefficient, and realized gain at 2.45 GHz, which show very little difference from the predicted values. Please review them. 373e2d3b0425ebb6b9884021ad0bd84 f8af56ea25ce182419761d8a2a50248 a3e8b6cdd42d95d7989da018f5ccb57

hair-an commented 7 months ago

Thank you for your reply, It was my carelessness that caused errors in the simulation. After correctly placing the feeding position, although there were still some differences, a result similar to yours appeared, perhaps it may due to the size of the ground? Anyway it proves that the DLmodel is correct and your work is Inspired.