Prediction of SET on SRAM Based on WOA-BP Neural Network

Wanting Zhou*, Man Li, Lei Li, Kuo Hui Yeh

*此作品的通信作者

研究成果: Article同行評審

7 引文 斯高帕斯(Scopus)

摘要

The incidence of high-energy particles into the semiconductor device would induce single event transients (SETs), which is a main threaten to MOS device. And the incidence distances and Linear Energy Transfers (LETs) have important effects on the SET current. A machine learning method based on Whale Optimization Algorithm-Back Propagation neural network (WOA-BPNN) model considering injection distances and LETs has been proposed to predict SET current in this paper. And this method could effectively reduce the simulation time from hours to seconds compared to device model. The current data that predicted by this method has been compared with the Technology Computer Aided Design (TCAD) simulation result which obtained in the background of the 40 nm process technology, the regression coefficient between the predicted value based on the proposed method and the TCAD simulation result was 99.76%, and the maximum integral relative error was 0.287% while the minimum integral relative error is 0.04%. Besides, the proposed method is also compared with PSO-BPNN (Particle Swarm Optimization, PSO) and GA-BPNN (Genetic Algorithm, GA), and the results demonstrated that the WOA-BPNN has prediction accuracy and timing saving advantages over the other two methods.

原文English
頁(從 - 到)267-273
頁數7
期刊Journal of Internet Technology
24
發行號2
DOIs
出版狀態Published - 3月 2023

指紋

深入研究「Prediction of SET on SRAM Based on WOA-BP Neural Network」主題。共同形成了獨特的指紋。

引用此