2025, Vol. 6, Issue 1, Part A
AI-Driven BP neural network approaches for enhancing analog circuit efficiency using simulation-based training and genetic optimization
Author(s): Matías González, Valentina Rojas and Joaquín Pérez
Abstract: The optimization of analog circuits is a critical challenge in modern electronics, demanding innovative approaches to improve performance and efficiency. This study explores the integration of Genetic Algorithm (GA) optimization with Backpropagation Neural Networks (BP-NNs) to enhance analog circuit design. The primary objective is to develop a hybrid model that leverages the predictive power of BP-NNs and the adaptive optimization capabilities of GA to address limitations such as slow convergence, overfitting, and inefficiencies in traditional methods. The research utilized SPICE-based simulation data as training and testing datasets, focusing on key circuit parameters like signal-to-noise ratio (SNR), power consumption, and bandwidth. A comparative analysis was conducted between the hybrid BP-NN model and the standard BP-NN to evaluate performance improvements. The hybrid model demonstrated significant enhancements across all metrics. It achieved a prediction accuracy of 98.7% compared to 91.2% for the standard BP-NN, with a reduction in mean squared error (MSE) from 0.045 to 0.0085. Training efficiency was improved, with the hybrid model requiring 35 epochs for convergence compared to 50 epochs for the standard model. Furthermore, circuit performance metrics showed substantial improvements: SNR increased by 18.4%, power consumption reduced by 12.7%, and bandwidth improved by 15.3%. Statistical analysis confirmed the significance of these results, with a p-value < 0.05 in paired t-tests. The study concludes that integrating GA with BP-NNs offers a robust, efficient, and generalizable framework for analog circuit design. Practical recommendations include incorporating this hybrid approach into design workflows, enhancing simulation environments for AI integration, and promoting interdisciplinary collaborations. The findings pave the way for advancing electronic design methodologies, bridging the gap between AI research and real-world applications.
DOI: 10.22271/27084531.2025.v6.i1a.78
Pages: 01-06 | Views: 64 | Downloads: 25
Download Full Article: Click Here
How to cite this article:
Matías González, Valentina Rojas, Joaquín Pérez. AI-Driven BP neural network approaches for enhancing analog circuit efficiency using simulation-based training and genetic optimization. Int J Res Circuits Devices Syst 2025;6(1):01-06. DOI: 10.22271/27084531.2025.v6.i1a.78
