Modeling of the Digital Class-D Amplifier Based on Deep Double Feedback Elman Neural Network

• Published in: Advanced Data Mining and Applications Vol. 13088; pp. 314 - 325
• Main Authors: Yu, Zeqi, Jiang, Bingbing, Liu, Haokai
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2022; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Digital class-D amplifier; Double feedback Elman neural network; Nonlinear modeling; Restricted Boltzmann machine
• ISBN: 9783030954079; 3030954072
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-95408-6_24

This paper presents a double feedback Elman neural network (DFENN) based on restricted Boltzmann machine (RBM), which we call RBM-DFENN, for modeling of the digital class-D Amplifier. RBM is the basic module of constructing deep learning networks, and the precision of the deep learning network is higher than that of the general network. The Elman neural network (ENN) was first proposed for speech signals, which has memory effect compared with the back propagation neural network (BPNN). Compared with the traditional ENN, the DFENN has stronger memory effect. Therefore, the RBM-DFENN proposed in this paper models the digital class-D amplifier. Experimental results show that the proposed model can accurately describe the memory effect of the system, and the modeling accuracy is higher.