.png){kind=icon}
.png){kind=icon}
Return
Multi-objective design method for power amplifiers based on convolutional neural networks
.png){kind=icon}
PRE
AI
DOI：10.1016/j.mejo.2026.107093.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Traditional circuit design methods are heavily reliant on the experience of designers and find it difficult to meet the modern requirements for precision and efficiency.Inspired by image and semantic recognition, a multi-objective design strategy based on Convolutional Neural Networks (CNN) and Particle Swarm Optimization (PSO) for Power Amplifier (PA) design is proposed in this paper. This approach establishes a mapping relationship between design parameters and performance metrics through a neural network model. The fitness function is minimized to maximize prediction accuracy. Subsequently, the optimal solution is efficiently located by the PSO algorithm.Experimental results show that replacing traditional simulation software with CNNs not only significantly improves training efficiency and accuracy but also enables designers without in-depth circuit knowledge to rapidly and accurately obtain superior circuit parameters. Compared with traditional manual calculations, this approach achieves better results within only 95 s, greatly improving design efficiency and demonstrating strong potential for large-scale circuit design.
Keywords:
Convolutional neural networks
Power amplifiers
Multi-objective design
Particle swarm optimization
Journal
IF:
2.3
Papers:
296
Citations:
4.7K