计算机科学
卷积神经网络
人工智能
特征(语言学)
超参数
模式识别(心理学)
断层(地质)
噪音(视频)
深度学习
理论(学习稳定性)
卷积(计算机科学)
小波变换
离散小波变换
小波
特征提取
人工神经网络
故障检测与隔离
算法
迭代法
群体行为
模拟电子学
可靠性
粒子群优化
混叠
财产(哲学)
区间(图论)
作者
G. Puvaneswari,A. Ahilan,R. Sundar,C. Anuradha
标识
DOI:10.1016/j.asej.2026.104400
摘要
Diagnosing analog circuit faults is essential to maintaining the stability and dependability of contemporary electronic systems. This paper proposes a novel hybrid deep learning framework, termed SbEO-ITCNN, which integrates Satin Bowerbird Optimization (SBO) and Egret Swarm Optimization (ESO) for efficient hyperparameter tuning of an Iterative Temporal Convolutional Neural Network (ITCNN). Initially, Wrapped Discrete Wavelet Transform (W-DWT) is employed to extract multi-resolution time–frequency features, effectively reducing noise and feature redundancy. The extracted features are then classified using ITCNN, which enhances temporal feature learning and improves classification performance. To further optimize the model, the proposed SbEO algorithm balances exploration and exploitation for robust parameter tuning. The proposed method is validated on multiple analog circuits, including Sallen–Key band-pass, Leapfrog, Butterworth low-pass, and Biquad high-pass filters. Experimental results demonstrate superior performance, achieving classification accuracies of 99.62 %, 99.21 %, 99.41 %, and 99.36 %, respectively, along with reduced computational time compared to existing techniques.
科研通智能强力驱动
Strongly Powered by AbleSci AI